| Oct 4, 2026 @ 11:57 PM |
Bandita Baisak, Ranjan Kumar Baisak |
ranjan@xhield.tech |
https://linkedin.com/in/ranjanbaisak |
https://xhield.tech |
9880398951 |
Founder |
Couple |
two |
Tech |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Oct 4, 2026 @ 9:37 PM |
SUSHEEL |
susheelshetty2@gmail.com |
https://www.linkedin.com/in/susheel-chilashetty-507a6622a?utm_source=share&utm_campaign=share_via&utm_content=profile&utm_medium=android_app |
https://web.rootskart.in/ |
09741670371 |
CEO ENGINEERING |
WE ARE WORKING FROM PAST 2 YEARS ON THE PROJECT |
2 |
WE ARE 10 MEMBERS IN THE TEAM MAJORLY ALL ARE INTO THE AI BASED DECISION SPACE TECHNOLOGY BASED AND HAVING PRODUCED MULTIPLE CATELOGUE OF THE PRODUCT OVER THE TIME AND THE SOLUTION FREQUENCY HAVE MADE THE BEST OF THE SOLUTION AND AS PROBLEM SOLVER WE HAVE WORKED ALOT IN THE SEGMENT ON THE SAME AND HAVE MADE THE BEST USABLE PRODUCTS TO THE END CONSUMER UPON THE APPROVALS |
ROOTSKART |
https://web.rootskart.in/ |
KALABURAGI |
ROOTSKART is building the AI Decision Intelligence Layer for the physical Earth, converting continuous satellite and geospatial observations into intelligent predictions, risk insights and actionable decisions across critical industries. |
Organizations today have access to unprecedented volumes of satellite, geospatial, climate and IoT data, but lack an intelligent AI layer capable of turning this information into timely decisions. Existing platforms largely provide imagery, maps, analytics and alerts, leaving humans to interpret what is happening, why it matters, what will happen next and what action should be taken. This creates a critical gap between **Earth observation and real-world decision-making**. Data remains fragmented across systems, analysis is often manual and reactive, risks are difficult to prioritize, and intelligence is rarely continuous or predictive. As infrastructure, cities, climate, forests, mining and maritime systems become increasingly complex, organizations need AI that can continuously understand changing conditions, reason across multiple data sources, predict risks and translate intelligence into actionable decisions at scale. |
ROOTSKART is building an AI Decision Intelligence Layer for Earth that transforms continuous satellite and geospatial observations into predictive, contextual and actionable intelligence. The platform fuses optical, SAR, hyperspectral, thermal, climate, GIS and other permitted data sources through a unified Earth Data Engine, then applies Geospatial AI, computer vision, temporal intelligence, predictive models and AI reasoning to continuously understand changes across the physical world. Instead of simply producing images, maps or alerts, ROOTSKART determines **what is changing, why it matters, what could happen next, the associated risk and what action should be prioritized**. Its modular intelligence architecture can serve infrastructure, urban mapping, climate technology, forests, mining, maritime and other physical-world sectors through sector-specific AI decision models. The platform delivers intelligence through APIs, dashboards and enterprise workflows, creating a scalable **Observe → Understand → Predict → Decide → Act** system that converts Earth observation into real-world decisions. |
ROOTSKART is differentiated by making **AI Decision Intelligence—not satellite data or mapping—the core product**. Its proprietary architecture fuses multi-source Earth observations with geospatial AI, temporal intelligence, domain models and AI reasoning to move beyond detection and analytics toward **prediction, risk prioritization and recommended action**. The platform is sensor-agnostic and designed to work across infrastructure, urban, climate, forest, mining and maritime domains without rebuilding the core intelligence stack for each sector. Its defensibility comes from continuously accumulated geospatial-temporal datasets, domain-specific AI models, location-based intelligence, decision workflows and outcome feedback that improve the system over time. This creates a scalable Earth Data → AI Understanding → Prediction → Decision → Actionarchitecture rather than a conventional imagery, GIS or dashboard product. |
Revenue |
Revenue |
ROOTSKART targets organizations that make high-value decisions about physical assets, land, infrastructure and the environment, including: * **Government & public-sector agencies** — urban planning, infrastructure, disaster management, forests, mining, environment and maritime authorities. * **Infrastructure & construction companies** — monitoring projects, assets, land development and construction progress. * **Smart cities & urban technology companies** — urban mapping, growth monitoring, infrastructure planning and climate-risk intelligence. * **Mining & natural-resource companies** — mine monitoring, land-use change, environmental compliance and operational intelligence. * **Climate & environmental organizations** — climate-risk assessment, ecosystem monitoring, carbon and environmental intelligence. * **Forestry & conservation organizations** — forest-change detection, degradation monitoring and ecosystem intelligence. * **Maritime & logistics companies** — vessel, port, coastal and offshore intelligence. * **Insurance & financial institutions** — geospatial risk assessment, asset monitoring and event-based damage intelligence. * **GIS, engineering and technology companies** — integrating ROOTSKART’s AI Decision Intelligence through APIs and enterprise platforms. **Primary customers:** Government, large enterprises and organizations managing geographically distributed physical assets where faster, predictive decisions have significant operational or financial impact. |
$80 Billion |
$75 billion |
$62 billion |
ROOTSKART follows a diversified **B2B, B2G and AI-as-a-Service model**, monetizing its AI Decision Intelligence Layer rather than raw satellite data. Revenue comes through **annual SaaS subscriptions** for sector-specific intelligence platforms, **usage-based APIs and data services**, **enterprise licensing and customized AI deployments**, and **project-based intelligence solutions** for government and large organizations. Additional revenue can come from integrating ROOTSKART’s AI decision layer into existing GIS, enterprise and operational systems. As usage scales, the model combines recurring platform revenue with API consumption and high-value enterprise deployments, creating a scalable global revenue architecture. |
Our competitive landscape includes global Earth-intelligence and GeoAI companies such as **Planet, Maxar/Vantor, BlackSky, ICEYE, Orbital Insight, Esri and Palantir**, along with emerging AI-native geospatial companies. In India, **SatSure and Aganitha Space Technologies** are relevant adjacent players in satellite-data analytics and Earth intelligence. SatSure, in particular, is developing large Earth-observation AI models and an Earth-intelligence backbone. However, ROOTSKART is positioned specifically around the **AI Decision Intelligence Layer**—combining multi-source Earth observation with AI reasoning, prediction, risk prioritization and actionable decisions across multiple physical-world sectors, rather than focusing primarily on satellite imagery, GIS or a single vertical. |
ROOTSKART uses a B2B/B2G enterprise-led go-to-market strategy focused on converting high-value geospatial use cases into scalable AI Decision Intelligence deployments. We acquire customers through **paid pilots and proof-of-value projects** with government agencies, enterprises, infrastructure companies and sector partners, demonstrating measurable improvements in monitoring, risk detection, prediction and decision speed. Successful pilots are converted into annual SaaS subscriptions, enterprise deployments and API contracts. We build distribution through strategic partnerships with **GIS providers, satellite-data companies, system integrators, insurers, engineering firms and government ecosystems**, while industry demonstrations, targeted enterprise outreach, conferences and referrals generate additional demand. Our modular platform allows a successful deployment in one sector or geography to be replicated across customers and markets, creating a **land-and-expand model** with recurring subscription, API and enterprise revenue. |
ROOTSKART will follow a **land-and-expand B2B/B2G go-to-market strategy**, entering through high-value, data-intensive use cases where AI-driven decisions can deliver measurable operational impact. We will begin with focused deployments in **infrastructure, urban intelligence, climate, forestry, mining and maritime** through paid pilots and proof-of-value projects with enterprises and government organizations. Successful pilots will convert into annual AI SaaS, API and enterprise contracts. We will accelerate adoption through partnerships with satellite-data providers, GIS companies, system integrators and industry platforms, while using direct enterprise sales, strategic government programs, demonstrations and sector partnerships to build the customer pipeline. Once validated in one geography or sector, the same AI Decision Intelligence core will be replicated across industries and international markets, creating a scalable **Earth Intelligence platform rather than a project-based business**. |
Our long-term vision is to build **the global AI Decision Intelligence Layer for Earth**—an intelligent infrastructure that continuously understands the changing physical world through satellite, geospatial, climate, IoT and other data sources. ROOTSKART aims to evolve from sector-specific intelligence into a **unified Earth Intelligence platform** capable of detecting change, understanding context, predicting risks, simulating outcomes and recommending actions across infrastructure, cities, climate, forests, mining, maritime and other critical systems. We envision a world where organizations do not merely observe Earth through satellite imagery, but can **query, understand and make intelligent decisions about the physical world in near real time**. Our ultimate goal is to become a global AI infrastructure layer connecting **Earth observation to intelligent action at planetary scale**. |
yes incroporated |
NA |
Yes |
We are applying to Avinya to accelerate ROOTSKART’s transition from a proven geospatial intelligence platform into a **scalable AI Decision Intelligence Layer for Earth**. Avinya’s ecosystem can help us strengthen our deep-tech architecture, validate high-impact use cases with industry and government, access strategic mentors and partners, and accelerate enterprise adoption. We see the program as an opportunity to refine our technology, build stronger market pathways and expand from individual satellite-data applications into a **multi-sector Earth Intelligence platform** serving infrastructure, urban, climate, forest, mining and maritime use cases. With Avinya’s support, we aim to move faster from technology validation to scalable deployments and global market expansion. |
During the programme, we aim to strengthen and validate ROOTSKART’s **AI Decision Intelligence Layer** by integrating multi-source satellite and geospatial data with advanced AI for real-time change detection, prediction, risk assessment and decision support. We want to validate the platform through industry and government pilots across infrastructure, urban, climate, forest, mining and maritime applications, convert pilots into scalable commercial deployments, strengthen our enterprise-ready product and APIs, and build strategic partnerships for national and global expansion. Ultimately, we aim to demonstrate a **repeatable Space-to-Decision platform that can scale across sectors and geographies**. |
ANYTHING IS OK |
Yes |
Core engine |
ROOTSKART’s architecture is built around an AI Decision Intelligence Layer* that converts continuous Earth observations into actionable decisions. A multi-source data fabric ingests optical, SAR, hyperspectral, thermal, weather, climate, GIS and other permitted geospatial data. A geospatial AI pipeline performs data harmonization, cloud/noise handling, feature extraction, change detection and spatial-temporal analysis. Above this, computer vision, temporal models, predictive ML and domain-specific models generate context-aware intelligence. An AI reasoning layer combines these outputs with historical patterns, geographic context and domain knowledge to determine **what is changing, why it matters, what may happen next and what action should be prioritized**. Results are delivered through APIs, dashboards and enterprise workflows with confidence and explainability. The architecture follows **Observe → Understand → Predict → Decide → Act**, enabling the same core intelligence layer to serve infrastructure, urban, climate, forest, mining and maritime applications. |
Our key data advantage is the proprietary geospatial intelligence layer we build on top of satellite and public Earth-observation datasets. While satellite imagery itself may be commercially or publicly available, ROOTSKART creates proprietary **location-specific, temporal and domain-labelled datasets** through continuous processing, historical analysis, customer deployments and ground/operational validation. These datasets capture how assets, environments and risks change over time and are used to train and improve our AI models for detection, prediction and decision intelligence. Our growing library of geospatial features, temporal patterns, sector-specific labels, decision outcomes and feedback creates a compounding data advantage, enabling the platform to become more accurate and context-aware as deployments scale. |
ROOTSKART’s defensibility comes from building an **AI Decision Intelligence Layer**, rather than a single satellite-analysis model. Our architecture combines multi-source Earth observation, geospatial AI, temporal modelling, domain intelligence and AI reasoning to move from detection to **prediction, risk prioritization and recommended action**. Over time, each deployment generates proprietary geospatial-temporal training data, domain labels, historical patterns and real-world outcome feedback, creating a continuously improving intelligence layer. Our sensor-agnostic architecture also allows new satellite and data sources to be incorporated without rebuilding the system. The combination of **proprietary data, temporal intelligence, domain-specific models, decision workflows and feedback loops** creates a compounding technical advantage that becomes stronger as the platform scales across sectors and geographies. |
We evaluate ROOTSKART’s AI models through a combination of **technical accuracy, temporal reliability and real-world decision impact**. Key metrics include **precision, recall, F1-score, IoU and mAP** for detection and classification; **MAE, RMSE and R²** for prediction models; and **false-positive/false-negative rates, confidence calibration and spatial-temporal consistency** for reliability. We also measure **inference latency, coverage and data-quality robustness** across different sensors, locations and environmental conditions. At the decision layer, we track **decision accuracy, risk-prioritization accuracy, actionable-alert rate and reduction in analysis/response time**. Models are validated against ground observations, historical datasets and customer/field outcomes, with continuous monitoring, human-in-the-loop validation and model drift checks before production deployment. |
ROOTSKART follows a **privacy-by-design and security-by-design** approach. We minimize collection of personally identifiable information and primarily process satellite, geospatial, environmental and asset-level data. Customer data is logically isolated with role-based access controls, encryption in transit and at rest, secure API authentication, audit logging and controlled data access. We apply appropriate data-retention and deletion policies and ensure data is processed only for defined purposes. Our deployments are designed to align with applicable Indian data-protection, geospatial-data, contractual and sector-specific requirements, with additional controls for government and enterprise deployments. We also maintain human oversight for high-impact decisions, data-quality checks and model monitoring to improve reliability, traceability and responsible AI use. |
Yes. Our primary regulatory considerations relate to **geospatial data, satellite-data licensing and access, data protection, cross-border data transfer, and sector-specific regulations**. Requirements may vary depending on the country, resolution and source of Earth-observation data, customer type and application area. Certain government, critical-infrastructure or sensitive-location use cases may also require additional approvals or restrictions. ROOTSKART mitigates these risks by using authorized data sources, following applicable Indian and international geospatial and data-protection requirements, implementing customer-specific access controls, and designing the platform to support compliance requirements in each target market. We continuously monitor regulatory changes as we expand across sectors and geographies. |
At 10x scale, the primary challenges would be **data volume, compute requirements, model inference latency, data-source heterogeneity and quality control**. A 10x increase in locations and observations could increase processing loads and make model monitoring and validation more complex. We address this through a **cloud-native, modular and sensor-agnostic architecture** with distributed processing, automated data pipelines, scalable model-serving infrastructure, caching and asynchronous inference. We also use confidence thresholds, automated quality checks, model-drift monitoring and human-in-the-loop validation for critical outputs. This architecture allows compute and storage to scale independently while keeping the AI Decision Intelligence Layer reliable as the number of customers, locations and observations grows. |
Yes. ROOTSKART has a **fully in-house AI/ML and geospatial technology team** with years of experience in developing and deploying satellite and geospatial intelligence solutions. Our team has expertise in **Geospatial AI, computer vision, machine learning, deep learning, remote sensing, GIS, spatial-temporal modelling, predictive analytics and AI-based decision intelligence**. We develop the complete technology stack internally—from satellite-data ingestion, preprocessing and feature extraction to AI modelling, prediction, risk assessment and the **AI Decision Intelligence Layer**. This fully in-house capability enables us to rapidly develop, validate, deploy and continuously improve our proprietary AI systems without core dependency on external AI development teams. |
We currently use a combination of **open, licensed and proprietary datasets**. Our stack includes publicly available Earth-observation datasets such as **Sentinel-1/Sentinel-2**, along with licensed/commercial satellite datasets where higher resolution or specialized spectral information is required. We also integrate weather, climate, GIS, terrain and other permitted geospatial datasets. Our proprietary advantage is created through the **derived datasets** we generate—processed geospatial features, temporal change histories, domain-specific labels, location intelligence and validated ground/operational data from deployments. This combination allows ROOTSKART to build an AI Decision Intelligence Layer without being dependent on a single satellite provider or dataset. |
We will continuously improve model performance through a **closed-loop AI learning system** that combines new satellite observations, temporal data, ground validation and real-world decision outcomes. Models will be continuously evaluated using precision, recall, F1, IoU, prediction error, false-positive/negative rates, confidence and drift metrics. We will retrain models with newly labelled and validated data, improve sector-specific models, and use human-in-the-loop feedback for critical decisions. Automated data-quality checks, model monitoring, A/B testing and periodic validation across new geographies and conditions will ensure reliability. As deployments scale, the resulting **data → prediction → outcome → feedback** loop will continuously strengthen ROOTSKART’s AI Decision Intelligence Layer. |
**Both India and global markets.** ROOTSKART is being built as a **global-first Space AI company from India**, using India as our technology and innovation base while designing the AI Decision Intelligence Layer for international deployment from the outset. Our sensor-agnostic and sector-agnostic architecture enables us to serve customers across infrastructure, urban intelligence, climate, forestry, mining and maritime markets worldwide. We aim to establish India as our core development and innovation hub while expanding through global enterprise partnerships, satellite-data ecosystems and international deployments. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/7qhHIOQGN59p-ROOTSKART_AI_Pitch_Deck.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/7qhHIOQGN59p-ROOTSKART_AI_Pitch_Deck.pptx |
https://drive.google.com/file/d/1usE7qeJkPTTMUtA7x_wQOn5BSwWoLmco/view?usp=drivesdk |
Yes. ROOTSKART is a **mission-driven Space AI company** focused on making Earth observation more intelligent, accessible and actionable. Our mission is to build an **AI Decision Intelligence Layer for Earth** that helps organizations understand environmental and physical-world changes earlier, predict emerging risks and make faster, evidence-based decisions. By applying satellite and geospatial intelligence across infrastructure, urban development, climate, forests, mining and maritime systems, we aim to improve resource efficiency, strengthen climate resilience, support sustainable development and reduce the impact of preventable risks. Our long-term goal is to create globally scalable AI infrastructure that transforms **Earth observation into intelligent action**, enabling better decisions for people, businesses and the planet. |
NA |
KITS KARNATAKA AND GOVERNMENT OF KARNATAKA |
checked |
| Oct 2, 2026 @ 12:32 PM |
George Johnson and Radha Baran Mohanty |
george.johnson@datamatrixai.co.in |
http://www.linkedin.com/in/georgejohnsonin |
http://www.datamatrixai.co.in |
7829440444 and 8847822146 |
Co - Founders |
We were part of the cohort for GenAI and Prompt Engineering course by Indian Institute of Science(IISc) Bangalore during 2024 and we came together to form the company legally from April 2025 |
All |
Datamatrixai is founded by two veterans with technology and business acumen. George spent in leadership position as MD for consulting firm and Radha spent 28 years with Infosys heading Delivery for North America for Media Entertainment and Communications vertical. Radha is an IIT alumnus. The products we have developed so far are real industry problems. Our strengths are:
1. Applying AI to real enterprise problems across existing data, documents, systems, and human conversations.
2. Deep domain orientation: We have developed use cases around power distribution, BFSI, education, retail/e-commerce, and other enterprise environments, giving it a practical understanding of industry workflows rather than just technology capabilities.
3. Ability to combine multiple AI approaches
We have used Generative AI, predictive AI, rule-based algorithms, semantic search and SLM/LLM technologies depending on the business problem. That is more useful to enterprises than forcing every problem into a GenAI solution.
4. Experienced technology team:
The combination of experienced AI engineers, architects and enterprise technology professionals gives DatamatrixAI an ability to understand both business problems and enterprise implementation challenges.
5. Ability to sense market forces and dynamics: Developing technology is one thing, but unerstanding what the market needs and how to approach especially in B2B enterprises is a challenge which the founders having spent decades in the corporate sector and leadership position can understand what could work and what needs time.
6. Building an organisation: In the IT world, talent is the key and getting superior talent at affordable costs is another challenge and motivating them to deliver is a different challenge. We have the capabliity to build a progressive and can do culture within the organsation for a long term sustainabilty.
7. Harmony: The founders trust each other and has an harmonious relationship which is essential for effective decision making. |
Datamatrix.AI Pvt Ltd |
https://datamatrixai.co.in |
Bengaluru and Bhubaneshwar |
At DatamatrixAI, we're building next-generation AI infrastructure to help businesses unlock the full potential of AI/ML and GenAI. Our platform leverages cutting-edge technologies to automate complex workflows, enhance decision-making, and reduce operational overhead. We design and develop scalable, and production-ready GenAI solutions with a big eye on security and privacy. |
Voice Bots have become a necessity for enterprises of any scale. We want to build self-service enterprise voice agent without the developer's intervention. |
Customer writes a prompt for agent creation. AI platform converts this to Agent specifications containing Objectives, Target Customer, Business rules, voice type, and language to be used, Questions to be asked, escalation rules etc. For eg., The Prompt is "Build a Malayalam voice bot, that answers inbound calls. The Bot understands whether the call is a sales enquiry, or to register a service complaint or check a complaint's status, collects the required details, saves them to the CRM, and then transfers the call to a customer care agent. |
The platform enables generating a Voice Bot Factory which is as easy as Create, Deploy and Run in any language and with business rules and make changes as per wish. |
two |
one |
B2B Enterprises having call centers such as 1. Telecom, Banking, Insurance, Manufacturing, Automobile firms, NBFC's, Retail, E-Commerce, Travel and Tourism, Hospitality, Heathcare. 2. Small Enterprises like Clinics, Restaurants, Fitness Centers, Travel Agents, Educational Institutions, Doctors, Advocates, etc |
NA |
NA |
NA |
SaaS Model |
None as of now |
We use enterprise sales, account-based outreach, strategic industry relationships and partnerships to reach decision-makers in BFSI, utilities, telecom, retail and other high-volume customer-interaction sectors. |
Our GTM is to start with one measurable voice-agent use case, demonstrate ROI, convert to a recurring SaaS/usage contract, and expand across functions through self-service agent creation. We initially focus on India and GCC markets, supported by direct sales, strategic partners and industry networks. |
We are strong at generating intelligence from the data. We intend to offer backward integration process where the voice is processed for insights and intelligence for actionable decision making. |
Pvt Ltd |
0 |
one |
For mentorship in Technology, Business. Leverage the strong IIT network and its strong brand equity with various industry stakeholders |
Enhance the product capabilities and make it a go to product for AI based Voice Bots |
Yes |
one |
one |
|
|
|
|
|
|
|
|
|
|
Both |
|
|
|
Yes |
Did not understand this question hence unable to answer. |
checked |
| Oct 2, 2026 @ 1:28 AM |
Chitra Javali and Girish Revadigar |
chitra.javali@trividhilabs.com |
https://www.linkedin.com/in/chitra-javali-phd-785ba34b?utm_source=share_via&utm_content=profile&utm_medium=member_android |
http://NA |
+916366438134 |
Dr. Chitra Javali - Founder and Director, PhD in CSE (Cybersecurity) M.Tech (Network. & Int. Eng) B.E (E&C ) • PhD -UNSW Sydney, Australia. World Rank - 19 • Postdoctoral Fellowship – National University Singapore (NUS), World Rank - 10. • Recipient of Google PhD Fellowship and N2Women Young Researcher Fellowship, • Winner of multiple prestigious international awards for research. • Industry Exp.: 10+ Yrs, Academic Exp.: 4 Yrs. • International Patents (Primary Investigator): 8. Dr. Girish Revadigar, Co-Founder and Director PhD in CSE (Cybersecurity) M.Tech (Ind. Electronics) B.E (E&C ) • PhD - UNSW Sydney, Australia. World Rank – 19 • Postdoctoral Fellowship – Singapore University of Technology and Design (SUTD) • Top-200 most qualified young researchers worldwide 2016 (By Heidelberg Laureate Forum). • Winner of multiple International Awards for research • Industry experience: 15+ Years • International Patents ( Primary Investigator): 73 |
My co-founder and I have been working together since our Master's (M.Tech) where we initially worked together as a team for graduation projects. Later on we have been collaborating to write research papers while pursuing PhD. Together we have more than 10 research papers published in top tier conferences and journals. |
one |
We have pursued PhD in Cybersecurity domain and worked in International standardisation bodies in the domain of telecom and digital identity. We together have 80 international patents in cybersecurity domain and have more than 12+ years of industry experience. |
Trividhi Labs Private Limited |
https://trividhilabs.com/ |
Belagavi |
We are building a Verifiable Agent Trust Infrastructure — Identity, Authorization and Accountability for Autonomous AI Agents that enables an AI Agent to cryptographically prove who it is, who authorized it, the data it may access, actions it may perform, and provide verifiable evidence of those actions across organizational boundaries. |
As AI systems evolve from conversational assistants to autonomous agents capable of accessing sensitive data, invoking enterprise tools, communicating with other agents and executing transactions, conventional human-centric authentication mechanisms are insufficient. A fundamental challenge is establishing cryptographically verifiable answers to the questions: Who is the agent? Who owns or operates it? What is it authorized to do? On whose behalf is it acting? And can its actions be independently audited? |
The proposed framework develops a decentralized trust framework using W3C Decentralized Identifiers, Verifiable Credentials, Verifiable Presentations, privacy-preserving disclosure and policy-based authorization to establish identity, capabilities, delegated authority and accountability for autonomous AI agents. |
ABC |
two |
Pilots |
The initial buyer is a CISO, identity lead, AI platform owner or product security lead whose organization already plans agents that touch sensitive APIs. |
INR 6,000 Cr |
INR 480 Cr |
INR 12 Cr ARR at 5 year |
Revenue Model |
Microsoft Entra Agent ID provides agent identity and related enterprise security controls. Okta for AI Agents covers agent discovery, access and governance, and announced additional runtime and delegation capabilities. CyberArk and other machine identity or privileged access vendors are relevant adjacent competitors. The proposed focus is portable authority evidence across organizational boundaries, with scoped delegation and signed receipts tied to tool execution. |
The initial buyer is a CISO, identity lead, AI platform owner or product security lead whose organization already plans agents that touch sensitive APIs. Discovery interviews should identify one workflow with an accountable resource owner and a realistic integration path. The first commercial motion is an 8 to 12 week paid pilot with agreed success metrics. Conversion occurs when the customer approves annual coverage for the workflow; expansion adds agents, APIs and organizational partners. |
Healthcare and education design partners can establish credible cases, while EHR and ERP vendors, system integrators and IAM consultancies offer distribution paths. India is the first operating market, with export opportunities through partners and standards based integration. |
In year one, our Agent Trust Infrastructure project will test the authority chain in healthcare and education. In years two and three, we plan to package connectors and policy patterns for repeatable enterprise deployments. By years four and five, our project aims to support independent verification of agent authority across more sectors and national boundaries. |
Incorporated on 7 February,2026 |
We have not raised any funding yet. |
one |
Need to answer |
Need to answer |
We would prefer virtual participation. |
one |
two |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
- |
|
|
|
|
|
|
| Oct 1, 2026 @ 5:52 PM |
Sumana Krishnaiahsetty Batchu |
sumana.ROSM@gmail.com |
http://www.linkedin.com/in/sumanabk |
http://NA |
09620405550 |
Sumana.B.K - Director, Venkatesh |
We are alumni of the same institute |
2 |
We complement each other's strengths |
ROSM Global Services Pvt Ltd |
NA |
Bangalore |
We are an innovative Ed-Tech Platform based on patented technology |
We help reduce knowledge gaps in students that can adversely affect their academic progress |
VBooks provides a dual-format integrated learning content by contextually linking text and video lessons so that students can switch between the two formats as required without losing context, thereby saving time and effort. Platform provides an integrated search and navigation across text, video-lesson and notes pertaining to a lesson so as to provide a better learning experience |
We have 3 granted patents supporting the platform technology |
MVP |
Users |
All Educational Institutes. Initial target customers will be private schools for students from 5th grade to 12th grade |
15,00,00,000 students across India |
25,00,000 Karnataka students from 5th-10th grades |
10,00,000 Bangalore students from 5th-10th grades |
Subscription free per school student and also creator fees for teachers/institutes creating content on VBooks |
None, product is based on a patented technology |
By demonstrating the product to high-fit private schools' management |
Reach schools through events, phygital marketing and sales channel partners where possible |
Provide futuristic dual-format digital textbooks called VBooks that not only improves quality of education, but also the learning experience |
Private limited |
NA |
Yes |
To get access to coaching, capital and market |
Gain paying customers at the earliest |
Yes |
Yes |
Supporting feature |
VBOOKs™ uses a multi-tier, decoupled AI architecture designed for curriculum-grounded learning rather than generic video generation: • Scripting & Structuring: LLMs (Gemini / GPT via an abstraction AI Gateway) convert textbook sections into structured JSON scene graphs. Formulas, definitions, and headers are extracted verbatim to eliminate hallucination. • AI Verification (Fact-Check): A secondary AI verification pass cross-checks every scene against the source text. Non-compliant scenes are held for human review. • Voice Narration: Enterprise neural TTS produces measured, natural academic voiceovers across languages (Indian English, Hindi, Kannada). • Visual Synthesis: Rather than unpredictable diffusion video, we use a hybrid engine: code-drawn procedural visuals (React, SVG, Canvas) combined with AI-generated text-free conceptual backdrops, compiled into deterministic MP4s using headless Chromium (Remotion) and FFmpeg. • App Stack: TanStack Start / React 19, TypeScript, PostgreSQL with Row-Level Security (RLS), and S3-compatible edge object storage. |
Curriculum-grounded learning graphs: Our data pipeline processes verified school and board syllabi (CBSE, ICSE, state boards) into granular, time-aligned concept maps (textbook sentence ↔️ video timestamp). • Fine-grained engagement telemetry: We capture proprietary behavioural data on where students pause, rewind, annotate, switch between text and video, and struggle with specific concepts. • Teacher feedback loop: Teacher corrections, approvals, and custom notes continuously refine our pedagogical scene-generation prompts. |
Patented bidirectional contextual switching between video playback and textbook text saves time & effort. 10x–50x Unit Economics Advantage: Generative video models (Runway, Sora, Veo) cost $5–$30+ per lesson and hallucinate text/math. Our hybrid code-drawn engine delivers crisp mathematical diagrams and verified voiceover at ~₹80–₹160 ($1–$2) per lesson. Enterprise School Governance: Full multi-tenant hierarchy (Principals, Teachers, Students, Parents) with mandatory human-in-the-loop review queues before content is published. |
Factual Faithfulness & Verbatim Retention: Automated string and token distance checks ensuring formulas, laws, and definitions match the source text with 0% semantic drift. • Temporal Sync Accuracy: Scene audio duration matched to sub-100ms keyframe transitions, avoiding visual-narration drift. • Teacher Acceptance Rate: Percentage of AI-generated lessons approved without manual edits in the review queue (target >90%). • Generation Success Rate: Pipeline completion rate across scripting, voice synthesis, image generation, and video compilation. • Student Retention: Pre- and post-lesson quiz score delta and completion rates. |
* Zero Data Leakage for AI: All LLM and TTS calls use enterprise API gateways with strict zero-data-retention agreements; school and student data is never used to train public foundation models. • Multi-Tenant Data Isolation: Enforced at the database layer via PostgreSQL Row-Level Security (RLS) and cryptographically signed JWT sessions. • Compliance-Ready: Built to adhere to India’s DPDP Act (2023) and global student privacy principles (GDPR / FERPA). • Transport & Storage Security: End-to-end TLS encryption in transit and AES-256 encryption at rest for all database records and media assets. |
Curriculum & Academic Compliance: Risk of unverified AI outputs misrepresenting state or national educational standards. Mitigated by our strict source-bounded ingestion (the AI can only reference the provided textbook) and teacher approval workflows. • Data Protection Laws: Evolving norms around student digital data under India’s DPDP Act. Mitigated by strict role-based access, institution data isolation, and not collecting unnecessary student PII. • Copyright in Training Data: We don't train custom base models on scraped data; we only process institution-authorized curriculum content through enterprise APIs. |
Upstream API Rate Limits: Concurrent LLM and image generation calls during peak school hours could hit upstream gateway TPM/RPM limits. Solution: Decoupled asynchronous task queues with automatic exponential backoff and multi-provider failover. • Video Rendering Capacity: Headless Chromium/FFmpeg rendering workers could experience queuing delays. Solution: Autoscaling containerized render workers (Docker/ECS) independent of the core web API. • Media Egress: Increased streaming bandwidth for audio and video. Solution: Aggressive global edge CDN caching and bytecode-light procedural SVG animation where possible. |
Yes. Our core engineering team specializes in: • Applied LLM orchestration, structured prompt chaining, and automated hallucination verification. • Computer graphics and media engineering (headless video rendering with Remotion, FFmpeg, Canvas/SVG procedural graphics). • Full-stack systems architecture, high-concurrency database design, and EdTech UX workflows. • Foundation model integration across Google Gemini, OpenAI, and state-of-the-art regional TTS engines. |
Content-Agnostic / BYOC Model: VBOOKs™ operates on a "Bring Your Own Curriculum" model. The system ingests textbooks and learning materials uploaded and owned by partner institutions or public open-standard curricula (e.g., NCERT). • No Unlicensed Scraping: We do not scrape copyrighted textbooks or third-party web content. • Proprietary Interaction Dataset: All generated scene graphs, temporal alignment metadata, and anonymized engagement logs are 100% owned by ROSM Global Services Pvt. Ltd. |
Human-in-the-Loop Feedback Loop: Logging teacher revisions and edits to continuously calibrate our system prompts and few-shot examples. • Automated STEM Benchmark Suites: Running automated regressions against complex mathematical equations, chemical formulas, and scientific diagrams to prevent rendering anomalies. • Specialized Visual Template Library: Expanding our procedural SVG component library so the system can depict physics simulations, biological processes, and financial tables with high visual fidelity. • Regional Language Tuning: Continuous evaluation of native accent, pronunciation, and pacing in Indian regional languages (Hindi, Kannada, Tamil, etc.). |
Both, with an initial strong focus on India. • India First: Deep curriculum support for CBSE, ICSE, and State Boards, multilingual voiceover (Indian English, Hindi, Kannada), and lightweight bandwidth optimization for Indian mobile networks. • Global Scalability: The underlying learning paradigm (patented video-text synchronization) and platform architecture are universally applicable to K-12 & higher education |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/izrzLcH4RbCU-VBooks-Acsel.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/izrzLcH4RbCU-VBooks-Acsel.pdf |
https://rosm-global.com/explainer |
We are an impact-focused organization, trying to improve learning quality, experience and impact. With VBooks students experience better knowledge assimilation and retention leading to improved learning outcomes. |
We are a woman-led startup with woman owning a majority stake |
Krishnamoorthy |
checked |
| Oct 1, 2026 @ 1:50 PM |
INAYATULLAH MAGAMI |
info@amazingenterprises.co.in |
https://www.linkedin.com/in/inayatullah-magami-4b9715249?utm_source=share_via&utm_content=profile&utm_medium=member_android |
http://www.amazingenterprises.co.in |
9844444507 |
I am the founder and director of AMAZING ENTERPRISES (OPC) Private limited. My background is in business development, trading and entrepreneurship. I am responsible for overall business strategy, operations, customer relationships, and developi8ng new business opportunities, My focus is on building a sustainable business by identifying market needs and creating reliable products and services |
We first met through our professional work and gradually developed a strong working relationship. We have been working together for several years, collaborating on various professional activities business and project |
one |
Our biggest strength as a team is our ability to communicate openly, trust each other, and work toward the same goal, we bring different skills and perspectives, support each other when challenges arise, and focus on finding practical solution together |
AMAZING ENTERPRISES (OPC) PRIVATE LIMITED |
www.amazingenterprises.co.in |
HUBLI |
Amazing enterprises is a trading and export business focused on suplying qulity food and agricultural products to domestic and international market |
We are solving the challenge of finding reliable suppliers who can provide quality food and agricultural products with consistent specifications, competitive pricing and dependable delivery, We aim to make sourcing easier and more transparent for buyers by connecting them with trusted supply partners |
Our solution is to provide a reliable transparent, and efficient sourcing platform for food and agricultural products, We connect buyers with trusted suppliers, ensure products mee required quality and specifications, offer competitive pricing, and coordinate dependable supply and delivery |
Our solution is unique because we combine reliable supplier relationships, quality verifications, customized product specifications Competitive sourceing and end to end coordination n one process. Our growing supplier network, market knowledge, Customer relationships, and experience in international trade create a strong foundation that becomes more difficult to replicate as we scale |
one |
two, Testimonials |
Our target customers are B2B buyers, Importers, distributors, Wholesalers, retailers, Hotels restaurants and food service businesses |
NA |
NA |
NA |
NA |
NA |
NA |
NA |
Our long term vision is to build Amazing Enterprises into a trusted global food products sourcing and export company, connecting quality Indian agricultural and food products with customers across International Markets. We aim to build long term buyer relationships, expand our product range and export markets, and become known for quality, reliability, competitive pricing, and consistent supply |
Amazing Enterprises (OPC) Private limited is an incorporated one person company (OPC) registered as a private limited company in India |
NA |
two |
We are applying to Avinya to gain access to the right mentorship,industry expertise, business networks and growth opportunities needed to scale our business. |
During the programme, we want to strengthen our business mode, validate our market strategy, and identify new international customers. We aim to improve our sales, operations and export capabilities through expert mentorship and industry connections, |
Yes, we are open to hybrid or virtual participation, We are comfortable attending sessions online and participating in in person session whenever required |
two |
Not-applicable |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Sep 30, 2026 @ 4:27 PM |
Amit Dhanania & Sunil Kumar Dhanania |
info@taxgem.co.in |
http://linkedin.com/in/amit-dhanania-970377102/ |
https://taxgem.co.in/ |
9883024805 |
Amit Dhanania - co-Founder , Sunil Kr Dhanania- Founder , Director - Ruchi Dhanania , Director - Neetu Dhanania ,Strategic Partner - Akash Dhanania , Research & Developent - NIdhi Sarpal |
10 |
2 |
Deep financial and compliance-domain expertise + practical understanding of the problems faced by individuals and businesses + technology/product execution. |
FOLDY |
foldy.co.in |
kolkata |
Foldy is a centralized financial and compliance document intelligence platform that helps individuals and businesses securely organize, manage and monitor their financial and compliance records in one place. |
Individuals and businesses manage important financial and compliance documents across emails, government portals, accountants, cloud folders and physical files. This makes it difficult to find records, track validity and renewals, identify missing documents, and maintain an organized compliance history. Foldy addresses this fragmented and manual approach by bringing these records into one structured digital workspace. |
Foldy provides a centralized digital workspace where users can securely store, organize and access financial, tax, compliance and business documents. Documents can be structured by year and category, with features for document search, validity and renewal tracking, missing-document identification and compliance record management. The platform is being developed with an AI roadmap that will add intelligent document classification, information extraction, search and personalized financial and compliance insights. |
Foldy combines financial and compliance domain knowledge with a unified document-management platform. Rather than focusing on a single tax, accounting or document category, Foldy is designed to create a consolidated financial and compliance record for individuals and businesses. Its foundation in Taxgem Professionals Pvt. Ltd. provides direct exposure to real-world tax and compliance workflows, helping us design the product around practical user needs. Over time, the structured document and compliance data layer will also support intelligent automation and AI-driven insights. |
MVP |
Users |
Individuals and salaried professionals, freelancers and consultants, startups and MSMEs, companies, and tax, legal and financial professionals who need a structured way to manage financial, tax, compliance and business documents. |
13.73 crore registered , individual taxpayers 8.9 crore , ITRs filed in AY 2025–26 4.83 crore non-filers |
NA |
NA |
Foldy follows a subscription-based SaaS model with tiered plans for individuals, SMEs and enterprises. Individual plans include Free, Foldy Plus at ₹999/year and Foldy Pro at ₹2,499/year. SME plans include Biz Basic at ₹4,999/year and Biz Plus at ₹9,999/year, while enterprise plans include higher-value offerings ranging from ₹18,000/year to ₹3,00,000/year depending on features and user licences. Revenue will be generated through subscriptions and higher-value business and enterprise plans. |
ClearTax, Zoho Books, TallyPrime and other tax, accounting, compliance and financial-management platforms are relevant competitors or adjacent solutions. Foldy differentiates itself by bringing personal and business financial data, compliance, taxation and banking information together in one unified financial intelligence and document-management platform. |
Foldy will initially leverage Taxgem Professional Pvt. Ltd.'s existing financial and compliance ecosystem, followed by digital marketing, referrals, partnerships with CAs and financial professionals, and targeted outreach to individuals, SMEs and businesses. The product will subsequently scale through digital acquisition, professional partnerships and product-led growth. |
Foldy will launch in phases, initially targeting individual users and SMEs with recurring tax, financial and compliance-management needs. We will leverage Taxgem's existing ecosystem for early market access and validation, supported by partnerships with CAs and financial professionals, digital marketing and targeted customer acquisition. We will subsequently expand toward larger businesses and enterprises through advanced compliance, tax, audit-readiness and financial-intelligence capabilities. |
Our vision is to build Foldy into a full-stack financial intelligence platform and the unified financial and compliance layer for Indian consumers and businesses. We aim to bring taxation, compliance, banking, investments, credit, insurance and financial documents into one secure ecosystem, reducing fragmentation and enabling timely compliance, better financial visibility and intelligent decision-making. |
Private Limited Company – Taxgem Professional Pvt. Ltd. |
NA |
Yes |
We are applying to Avinya to accelerate the development and market readiness of Foldy, a FinTech and Compliance Tech platform being developed by Taxgem Professional Pvt. Ltd. We are particularly looking for mentorship in AI and product development, access to industry and technology experts, support in validating our product and business model, and opportunities to connect with potential partners, customers and investors. Avinya's accelerator environment can help us strengthen Foldy's technology, go-to-market strategy and scalability as we move from product development toward market launch. |
During the program, we want to take Foldy from its current product-development stage toward a market-ready MVP, strengthen its technology and AI capabilities, validate the product with target users, refine our go-to-market and revenue strategy, establish relevant industry partnerships, and prepare the business for scalable market launch and future fundraising. |
Yes, we are open to hybrid and virtual participation. |
Yes |
Supporting feature |
Foldy's AI architecture is currently under development. The planned architecture will use AI/ML models on top of Foldy's structured financial, tax and compliance data layer to provide capabilities such as tax-filing intelligence, document understanding, information extraction, intelligent search and compliance insights. The final model and technology stack will be determined during product development and validation based on accuracy, security, cost and reliability requirements. |
Foldy's initial advantage comes from Taxgem Professional Pvt. Ltd.'s domain expertise and experience in taxation, audit and regulatory compliance, which informs the product's financial and compliance workflows. As Foldy scales, consent-based and securely structured user data generated through the platform can support increasingly contextual financial and compliance intelligence. We will ensure that any use of user data for AI is subject to appropriate consent, privacy and regulatory controls. |
The defensibility of Foldy's planned AI capabilities will come from the combination of financial and compliance domain expertise, structured financial and compliance workflows, contextual document and tax data, and integration across multiple financial and compliance sources. Rather than relying solely on a general-purpose AI model, Foldy aims to build domain-specific intelligence around Indian taxation, compliance and financial records. |
AI model evaluation will be established as part of product development. Planned evaluation criteria include accuracy of document classification and information extraction, factual consistency, compliance-rule accuracy, response relevance, false-positive and false-negative rates, latency and reliability. Outputs will be validated against defined financial and compliance datasets and reviewed through human-in-the-loop processes before production deployment. |
Foldy is being designed with a DPDP-first approach to data control, including consent-based access to financial and compliance information, controlled data access, secure storage and appropriate user permissions. We plan to apply encryption, access controls, audit logging, data minimisation and secure API practices, with appropriate retention and deletion mechanisms. Security and compliance requirements will be incorporated into the product architecture as the platform progresses toward production. |
Yes. Foldy operates in a highly regulated financial and compliance environment. Key considerations include the Digital Personal Data Protection framework, cybersecurity and data-security requirements, consent and privacy obligations, government and financial-data API requirements, and regulations governing tax and financial information. As the product evolves, we will work with appropriate legal, compliance and cybersecurity professionals to ensure that applicable requirements are incorporated into the platform and operating processes. |
As Foldy scales, the main areas requiring additional engineering attention will be data processing capacity, API and integration throughput, AI inference latency, secure data storage, system reliability and monitoring. The architecture will therefore be designed to scale these components independently, with appropriate controls for security, availability and performance. |
We do not currently have a dedicated in-house AI/ML team. AI capabilities are currently under development, and we plan to build the required technical expertise through our technology development team and specialist AI/ML resources as the product progresses. |
AI datasets are not currently being used in a production AI system. Foldy's planned AI capabilities will leverage structured financial and compliance information generated through the platform, subject to user consent, applicable privacy requirements and appropriate data controls. Any external datasets used during development will be evaluated for appropriate licensing and permitted use. |
We plan to continuously improve model performance through evaluation against representative financial and compliance use cases, monitoring accuracy and error patterns, human review of critical outputs, structured user feedback and periodic model and prompt evaluation. Improvements will be validated before being incorporated into production workflows, particularly for high-impact financial and compliance use cases. |
India initially, with the potential to expand to global markets in later phases. Foldy's initial product is designed around India's taxation, compliance and financial ecosystem, including GST, Income Tax, TDS, ROC and related financial information. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/naHmzTgJaGLQ-Foldy-Pitch-Deck_Pre-Seed-V1.1-1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/naHmzTgJaGLQ-Foldy-Pitch-Deck_Pre-Seed-V1.1-1.pdf |
http://NA%20 |
Yes. Foldy aims to simplify access to and management of financial and compliance information for individuals and businesses in India. By bringing fragmented financial, tax and compliance records into one structured platform, Foldy aims to reduce administrative friction, improve visibility of important financial obligations and help users manage their financial and compliance responsibilities more effectively. |
NA – Not applicable. |
STPI - JAYNAGAR , BENGALURU |
checked |
| Sep 30, 2026 @ 12:50 AM |
Saurabh Awasthi and Amit Pandey |
saurabh@augmen.io |
https://www.linkedin.com/in/saurabh-awasthi-entrepreneur/ |
https://augmen.io |
9717067973 |
Saurabh Awasthi, Co-founder & CEO: IIT Roorkee alumnus with 15+ years in AI, data science and BFSI; leads strategy, GTM, partnerships and financial-inclusion product vision. Amit Pandey, Co-founder & CTO: NMIMS alumnus; leads AI architecture, product engineering, security, integrations and scalable enterprise deployments. |
We worked together at Fractal between 2012 and 2015. Along with our corporate engagement, we worked together in our last startup from 2022 to 2024 before starting Augmen in Feb 2025 |
All |
The team’s biggest strength is the combination of deep BFSI and AI expertise with hands-on execution in real financial-inclusion environments. Saurabh brings domain, GTM and institutional depth, while Amit brings strong technology and product execution—together enabling Augmen to solve complex last-mile finance problems credibly and at scale. |
Augmen |
https://augmen.io |
Mumbai |
Voice – First Market place for Financial Inclusion |
Usability of financial products by the underserved population |
Augmen is a voice-first financial access platform for low-income, low-literacy and underserved users. It converts complex financial journeys—discovery, eligibility, KYC, documentation and form filling—into simple conversations, enabling users to independently access government schemes, credit and, over time, insurance, pension and investment products. |
Augmen’s defensibility comes from combining proprietary voice/video onboarding models, deployed financial-inclusion workflows, real-world interaction data, reusable user profiles, and embedded distribution. Its moat is not just AI, but the integrated stack of technology, field intelligence, institutional trust, and growing access to underserved users and financial providers. |
Revenue |
Revenue |
NBFCs, Banks, MFI, SFBs |
59,300 Crore |
8000 Crore |
1250 Crore |
Commission from financial institutions |
Haqdarshak, Paisabazaar, PolicyBazaar, Jan Samarth |
Partnerships with companies working with underserved (low income low literacy group) for demand and onboarding financial institutions for supply side |
Augmen’s GTM is a B2B2C embedded-distribution model: acquire underserved users through governments, gig platforms, NGOs and MSME ecosystems; onboard dedicated lenders and financial providers; convert users through voice-first DIY journeys; then expand from credit and schemes into insurance, pension and investments using one reusable financial profile. |
To enable every underserved person to make informed financial choices independently and participate fully in the formal financial system. |
Private Limited Company |
0 |
Yes |
For Funding and Strategic Guidance |
Personal Mentorship, Network access and Funding Support |
Yes |
Yes |
Core engine |
Augmen’s AI architecture combines multilingual voice and video AI, document intelligence, KYC and verification models, eligibility engines, and workflow orchestration. Proprietary models convert unstructured conversations and documents into structured, decision-ready applications, while modular APIs, RAG-based knowledge layers, and secure deployment enable integration across governments and financial institutions. |
N/A |
Augmen’s AI defensibility comes from proprietary multilingual voice/video models trained on real underserved-user journeys, combined with document intelligence, KYC, eligibility and workflow orchestration. Live government and lender deployments generate unique interaction and outcome data, continuously improving models while creating a hard-to-replicate advantage in low-literacy financial onboarding at scale across institutions. |
Augmen evaluates AI performance through accuracy, completion, confidence and exception metrics across voice understanding, document extraction, KYC, eligibility and form filling. Production outputs are benchmarked against human-validated samples, monitored for language and user-segment variance, and routed to human review when confidence drops. Deployment feedback continuously improves models and workflows safely. |
Augmen protects user data through privacy-by-design architecture, explicit consent, data minimization, encryption in transit and at rest, role-based access controls, audit logging, and secure retention practices. Its ISO 27001-certified security framework and on-premise deployment options allow institutions to retain control of sensitive data, while limiting unnecessary third-party exposure and access. |
Yes. Key risks include RBI digital-lending, KYC/AML, consent and data-localisation requirements; DPDP Act obligations; restrictions on financial-product solicitation, advisory and distribution; insurance, pension and securities licensing; and evolving AI/accountability rules. Augmen should remain an enablement layer, use regulated partners where required, preserve auditability, and build compliance-by-design from inception at scale. |
At 10× scale, Augmen’s biggest risks are voice latency and inference cost, multilingual accuracy drift, document/KYC exception rates, unreliable third-party APIs, workflow orchestration bottlenecks, and weak observability. The system must scale with automated monitoring, confidence-based fallbacks, human exception handling, load isolation, model versioning, and strict data-security controls. |
Augmen’s in-house AI/ML team develops the core voice, video, document-intelligence, KYC and decisioning capabilities powering the platform. The team focuses on multilingual model performance, workflow intelligence, evaluation, deployment and continuous improvement, while closely integrating AI engineering with product, security and real-world financial-inclusion use cases. |
Owned |
Augmen continuously improves performance through production feedback loops, human-validated samples, model retraining, prompt and workflow optimization, multilingual error analysis, and confidence-based exception review. Outcome data from completed applications, approvals, failures and user drop-offs is fed back into evaluation pipelines, enabling measurable improvements in accuracy, completion rates, latency and system reliability. |
Augmen is India-first, but globally scalable. India provides the ideal proving ground because of its linguistic diversity, underserved population and complex financial-access ecosystem. Once the voice-first DIY model is proven at scale, Augmen can expand to other emerging markets facing similar barriers in literacy, documentation, financial access and inclusion. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/P35q7Jr4j85q-Augmen_Investor_Pitch.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/P35q7Jr4j85q-Augmen_Investor_Pitch.pdf |
https://drive.google.com/file/d/1ofaFyOt-fLx6ttDSk9kd7-zcyG04fNcP/view?usp=sharing |
Augmen is an impact startup because it enables low-income, low-literacy and underserved users to independently access formal financial services. By reducing dependence on agents, simplifying complex financial journeys, and improving access to credit, government schemes, insurance and pensions, Augmen aims to increase financial inclusion, dignity, transparency and economic opportunity. |
No |
SINE, IIT Bombay |
checked |
| Sep 29, 2026 @ 5:03 PM |
Rajagopalan Varadarajan and Vindhya Kumar |
raj@fyvindia.com |
https://www.linkedin.com/in/rajagopalan-varadarajan/ |
https://www.fyvindia.com |
8904606243 |
Rajagopalan Varadarajan, Co-Founder: leads business strategy and technology. Vindhya Kumar, Co-Founder: leads education programme design. |
We met through the National Cadet Corps (NCC) and voluntary organisations over 10 years ago. We have volunteered together in other organisations for about 10 years and have worked together on FYV for the past 1.5 years. Both founders now work full-time on the venture. |
two |
Our biggest strength is combining education and facilitation experience with business and product development. We have built trust through about 10 years of volunteering together and 1.5 years of working together on FYV. Rajagopalan leads business strategy and technology, while Vindhya leads education programme design. Both founders work full-time on the venture. |
FYV Learning Solutions LLP |
https://www.fyvindia.com |
Bengaluru, Karnataka |
Speakery combines community learning clubs and a companion app under development to build communication, critical thinking, leadership and entrepreneurial skills. |
Students and young professionals need sustained opportunities to practise communication, critical thinking and leadership. Traditional classroom learning and one-off workshops often lack personalised feedback, continuous practice and measurable progress. |
Speakery provides community-based clubs where learners practise through weekly discussions, debates, projects and presentations in schools and community spaces. The clubs have served over 100 paying customers, including repeat customers. A companion app prototype is under development and testing internally and with intended users in real Speakery sessions. Planned features include personalised learning paths, AI-assisted speech feedback, mentor feedback, practice challenges and progress tracking. The model combines regular real-world practice, facilitation and ongoing assessment. |
Our differentiation is the combination of recurring community-based practice, facilitator-led learning and a developing digital layer for personalised feedback and progress tracking. Operating clubs and repeat paying customers provide a direct feedback loop for refining the learning experience and app. The founders combine education expertise with business and product development experience. |
Revenue |
two |
Students and young professionals, reached directly through Speakery memberships and through schools, colleges and residential communities; additional target buyers include corporate learning teams and government/CSR skill-development programmes. |
Internal pitch-deck estimate: INR 22,500+ crore annual India market opportunity, covering approximately 15 million potential learners. |
Internal pitch-deck estimate: INR 450-750 crore serviceable market across Karnataka and major urban markets, representing approximately 300,000-500,000 reachable learners. |
Five-year target in the pitch deck: approximately 50,000 active learners and INR 8-12 crore annual revenue potential, through 200+ Speakery clubs and institutional/community partnerships. |
Paid institutional programmes, Speakery community and individual memberships, and corporate training programmes. The planned digital platform adds B2B/B2C subscriptions, assessments, analytics and tools, with events and certifications as additional future revenue streams. |
Alternatives identified in our pitch deck include Toastmasters and Josh Talks in public-speaking communities, Dale Carnegie and NIIT in training, British Council in language learning, and self-paced platforms such as Coursera and Udemy. |
Our acquisition approach combines founder-led outreach to schools, colleges and corporates; partnerships with apartment associations, NGOs and educational institutions; demonstrations and pilot workshops; digital content and community events; and renewals and referrals. |
Validate and refine Speakery through Bengaluru schools and communities, convert demonstration sessions and pilots into annual partnerships and recurring memberships, and expand through a hub-and-spoke club model. Introduce the companion app across clubs as development and testing progress, then scale through institutional, community and government/CSR partnerships across Karnataka and other Indian cities. |
Build an accessible, scalable learning community that helps students and adults develop communication, critical thinking, leadership and entrepreneurial skills through sustained practice. Combine local Speakery clubs, trained facilitators and a personalised digital platform to support measurable progress and expand from Karnataka across India. Our long-term ambition is to create 100,000+ speaking opportunities and a community of 100,000+ members. |
Incorporated as FYV Learning Solutions LLP in India on 26 June 2025; based in Bengaluru, Karnataka. LLPIN: ACP-4375. |
Bootstrapped; no external funding raised. |
one |
We are applying to Avinya to strengthen the AI-assisted learning capabilities of Speakery, refine our companion app through user testing, and build a scalable model alongside our existing community clubs. We seek technical and business mentorship, product validation, suitable funding support, and partnerships across Karnataka. Avinya can help us translate early customer traction into a more robust, measurable and scalable learning product. |
During the four-month programme, we aim to refine and validate the app prototype with intended users, prioritise AI-assisted feedback and personalised learning features, establish clear learning-progress and engagement metrics, strengthen unit economics and go-to-market execution, and develop partnerships for further pilots and rollout. We also aim to improve funding readiness and define a practical roadmap for scaling Speakery across Karnataka. |
Yes, we are open to both hybrid and virtual participation. |
one |
two |
Speakery is at the MVP and early-testing stage, with AI-assisted feedback and personalised learning journeys as key capabilities. The planned architecture uses Gemini and ElevenLabs integrations and is intended to support audio, video, images, transcripts and typed responses. The product combines digital feedback and journey building with facilitated offline and online learning. These integrations and the architecture are being developed and validated; we are not presenting them as a fully deployed production system. |
Our current advantage is direct access to learning-context feedback through Speakery clubs and early app testing, allowing manual comparison and iteration based on real learning experiences. We collect learner submissions with consent and limited retention. We are not claiming a large proprietary training dataset or exclusive rights to learner content. |
Our intended differentiation is the integration of AI feedback and personalised learning journeys with real-world, facilitator-led Speakery clubs. On-ground feedback and manual review help us refine the experience around core life skills. The hybrid model can be adapted for students and professionals, with further language adaptation planned. Our differentiation is in the learning design, delivery model and feedback loop; technical defensibility is still being developed at this early stage. |
At the MVP stage, we review AI-generated feedback through manual comparison and feedback from on-ground Speakery sessions. This qualitative review helps identify mismatches and areas for improvement. |
We obtain consent for collecting learner submissions, including parent or guardian consent for children, and offer deletion on request. Recordings, images and responses are stored temporarily in Google Cloud buckets, accessible only to organisation administrators. Automatic deletion occurs within 30-365 days, with the retention period determined by the use case and requirements of the particular learning challenge or need. |
Relevant risk areas include personal-data protection, handling children's recordings and submissions, consent and retention, and the processing of learner content by third-party AI providers. International expansion will require assessment of destination-market requirements. Our current safeguards include consent, parent or guardian consent for children, restricted administrator access, deletion on request and use-case-based automatic deletion. Provider arrangements and applicable requirements will need review as integrations and markets expand. |
At 10x usage, anticipated constraints include multimedia processing costs and latency, storage and upload capacity, external AI service quotas, and the capacity for manual feedback review. These are prospective risks at our early-testing stage rather than measured production bottlenecks. Scaling will require load testing and validation of cost, response time and feedback quality. |
Development is founder-led, with product management and business analysis experience, including experience at Deloitte and fintech organisations such as Perfios. Engineers and product professionals with technical backgrounds contribute to development, using AI-assisted development tools including Codex and Claude. The current capability is focused on product development and integration; a dedicated AI/ML research team is not being claimed. |
No external datasets are currently used. |
We plan to improve the product through feedback from on-ground Speakery sessions and manual comparison of AI-generated feedback with human review, using identified issues to guide iterative improvements to feedback and personalised learning journeys. |
India first, starting with Bengaluru and expanding across Karnataka and other Indian markets. We subsequently intend to expand the digital product globally. |
|
|
Yes. FYV aims to make practical life-skills development accessible through Speakery clubs and an online/offline learning model. We help learners build communication, critical thinking, leadership and entrepreneurial skills through repeated practice, facilitated activities and personalised feedback. Our community clubs have served over 100 paying customers, while the companion app is in early testing. We intend to adapt the model for students, professionals and additional languages, supporting confidence, employability and community participation. Our immediate focus is India, followed by global expansion of the digital product. |
Woman-co-founded: Vindhya Kumar is a co-founder of FYV Learning Solutions LLP. |
DPIIT × Startup Karnataka - Founder Connect (OFFICIAL) WhatsApp Channel |
checked |
| Sep 29, 2026 @ 2:29 PM |
Sheerin TaJ S |
futuurebrains.ai@gmail.com |
https://www.linkedin.com/company/futurebrains-ai/ |
http://NA |
6362114747 |
Director |
Almost 15 years of work experience i have we saw this opportunity in DDPIT karnataka whats app channel |
2 |
All are professional |
FutureBrains AI LLP |
NA |
Banglore |
FutureBrains AI is an AI-powered education platform that helps schools teach, assess, and personalize learning at scale. |
Schools are struggling to integrate AI into everyday education. Teachers lack structured AI-ready curriculum, practical classroom tools, and sufficient training to create personalized learning experiences, while schools lack a centralized platform to manage AI-enabled teaching, learning, and assessment. |
FutureBrains AI is an AI-powered education platform that helps schools integrate AI into everyday teaching and learning. Teachers can upload lesson plans and generate curriculum-aligned lesson materials, classroom activities, worksheets, quizzes and assessments with AI. Students receive personalized learning activities, practical AI projects and assessments through the platform, while teachers and school administrators can track learning progress and outcomes through dashboards. |
FutureBrains AI is designed specifically for schools rather than as a general-purpose AI chatbot. Our differentiation lies in combining curriculum-aligned AI content generation, teacher assistance, student learning, practical AI projects, assessment, and school-level analytics within one platform. Our long-term defensibility will come from our structured curriculum and teaching resources, school-specific workflows, teacher and student usage data, localized content including Indian regional languages, and partnerships with schools and educators. As we scale, these resources and real-world classroom insights will create a continuously improving education ecosystem that is difficult to replicate through a general-purpose AI tool alone. |
Idea |
Users |
Our primary customers are K–12 private and government schools in Karnataka and India that want to integrate AI into classroom education. Our initial focus is on schools with 200–1,000+ students that need structured AI curriculum, teacher training, AI-enabled teaching tools, student projects, and learning analytic |
AI-enabled K–12 education market across India |
K–12 schools in Karnataka that can adopt AI education solutions |
Initial private and government schools we can acquire in Bengaluru/Karnataka |
Our initial focus will be on recurring annual school subscriptions, with additional revenue generated through teacher training, implementation services and institutional partnerships. |
STEMROBO |
NA |
We will initially focus on converting a small number of Bengaluru schools, use their feedback and outcomes to refine the product, and then expand across Karnataka and India. |
Our long-term vision is to build FutureBrains AI into a leading AI education platform for schools across India. We aim to make AI education accessible, practical and relevant by combining curriculum, teacher enablement, personalized student learning, practical AI projects and measurable learning outcomes in one ecosystem.
Starting in Karnataka, we plan to expand across India and eventually serve schools in international markets, with localized content and support for regional languages. Our goal is to help millions of students develop the skills to understand, use and build with AI, while enabling teachers and schools to confidently integrate AI into everyday education. |
Incoporated |
15 lakh |
Yes |
Need mentorship and onboarding support |
AI enabled schools in india |
Hybrid |
Yes |
Core engine |
FutureBrains AI follows a modular, API-driven AI architecture designed to separate the education platform from the underlying AI models.
1. User & Application Layer
Teachers, students and school administrators access the platform through a web/mobile interface built initially using Bubble.
2. Application & Orchestration Layer
The application manages authentication, school/class/student data, lesson workflows, assignments, assessments and AI requests. An AI orchestration layer validates requests, applies role- and context-specific prompts, and routes requests to the appropriate AI model.
3. AI Model Layer
The platform connects through APIs to foundation models such as OpenAI, Google Gemini or other suitable models. This model-agnostic approach allows us to select models based on accuracy, cost, latency and use case without redesigning the core platform.
4. Education Intelligence Layer
Our proprietary application logic adds curriculum context, grade level, learning objectives, teaching methodology, assessment rules and student learning history to AI requests. This transforms general-purpose AI responses into structured educational outputs.
5. Data & Analytics Layer
Student progress, assessments, learning activities and teacher feedback are stored securely and used to generate dashboards, identify learning gaps and personalize recommendations.
6. Safety & Governance Layer
The architecture includes role-based access, controlled AI interactions, usage limits, content validation, audit logs and protection of student data.
The architecture is designed to scale from an initial school pilot to a multi-school platform while allowing AI models and infrastructure components to be replaced or upgraded independently. |
We currently do not have a proprietary dataset. We plan to build a consent-based, privacy-protected dataset through school deployments, teacher feedback, student learning activities and assessment outcomes as the platform scales. |
Our AI component is not based on building a proprietary foundation model. Instead, our defensibility will come from the education-specific intelligence layer built around existing AI models.
FutureBrains AI will combine curriculum-specific context, structured educational workflows, teacher feedback, assessment patterns and localized content to generate relevant outputs for Indian schools. Over time, real-world classroom usage and feedback will allow us to continuously improve our prompts, workflows, evaluation methods and personalization algorithms.
Our model-agnostic architecture also allows us to adopt newer and more cost-effective AI models without rebuilding the product. The long-term advantage will therefore come from our education-specific workflows, accumulated classroom insights, curriculum knowledge and school integrations rather than dependence on a single AI model. |
We evaluate AI performance using a combination of automated checks and human evaluation.
For educational content, outputs are evaluated for factual accuracy, curriculum relevance, age appropriateness, clarity, completeness and consistency. Teachers review AI-generated lesson plans, activities, quizzes and assessments during pilot deployments, and their feedback is used to improve prompts and workflows.
We also monitor metrics such as response accuracy, failure rate, response time, AI usage patterns and teacher acceptance. For student-facing features, we use controlled prompts, predefined content guidelines and validation checks to reduce inappropriate or unreliable responses.
As the platform scales, we plan to build a standardized evaluation dataset based on approved curriculum content and teacher-reviewed examples, enabling continuous benchmarking of different AI models and selecting models based on quality, cost and reliability. |
FutureBrains AI is being designed with privacy and security as core requirements, particularly because the platform will process data relating to school students.
We will follow applicable Indian data-protection requirements, including the Digital Personal Data Protection Act and Rules. For children, we will implement appropriate parental consent and age-related controls before processing personal data.
Our planned security measures include role-based access for students, teachers and administrators, encryption of data in transit and at rest, secure API authentication, minimal data collection, controlled data retention, audit logging, regular security reviews and restricted access to sensitive information.
Student data will be used only for defined educational purposes and will not be used to train general-purpose AI models without appropriate authorization. Schools and authorized users will have appropriate mechanisms for managing access, data requests and deletion.
As we move from pilot to production, we will formalize our privacy policy, data-processing agreements, security procedures and compliance controls and conduct appropriate security assessments. |
Yes. The primary regulatory risks relate to data privacy, children’s personal data, AI-generated content, and evolving education technology regulations.
Because FutureBrains AI will serve school students, we will need to comply with applicable Indian data-protection requirements, including requirements relating to children’s data and consent. We will also need to ensure that AI-generated educational content is age-appropriate, accurate, curriculum-aligned and subject to appropriate teacher oversight.
We plan to mitigate these risks through privacy-by-design, minimal collection of student data, appropriate consent mechanisms, role-based access, secure data handling, audit controls, human review of educational content and clear AI-use policies.
The regulatory environment for AI is evolving, so we will continuously monitor applicable government regulations and education-sector guidelines as we expand. |
At 10× scale, Bubble can continue to serve as the initial application and administration layer while high-volume AI workloads and specialized services are progressively moved to scalable cloud infrastructure. This allows us to scale without becoming dependent on a single AI provider or technology platform. |
Currently, FutureBrains AI does not have a dedicated in-house AI/ML team. The initial product is being developed using established AI models through APIs, with the core product and education workflows being developed in-hous |
At the current prototype stage, FutureBrains AI does not rely on a proprietary training dataset. We use a combination of publicly available educational resources, appropriately licensed/open educational content, internally created curriculum and instructional materials, and AI model APIs.
Where third-party content is used, we verify and comply with the applicable licensing and usage terms. We do not claim ownership of third-party datasets or content.
As we conduct school pilots, we plan to build our own structured, consent-based dataset from teacher feedback, curriculum interactions and learning outcomes, subject to applicable privacy and data-protection requirements. |
We plan to continuously improve AI performance through a feedback-driven evaluation cycle.
Teacher feedback and human review will be used to evaluate AI-generated lesson plans, activities, quizzes and assessments for accuracy, curriculum alignment, age appropriateness and usefulness. Student outcomes and assessment patterns will help identify areas where AI-generated learning content needs improvement.
We will maintain a library of teacher-approved examples and evaluation cases to benchmark different prompts and AI models. We will continuously monitor accuracy, reliability, latency, cost and safety, and update prompts, workflows and model selection based on these results.
As the platform scales, we plan to build a consent-based education dataset from classroom interactions and feedback to improve personalization and educational recommendations while maintaining appropriate privacy and security controls. |
Yes |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/mN4iRMPQf0qc-FutureBrains_AI_Investor_Pitch_Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/mN4iRMPQf0qc-FutureBrains_AI_Investor_Pitch_Deck.pdf |
http://na |
Yes. FutureBrains AI is a mission-driven, impact-focused startup focused on making quality AI education accessible and practical for school students.
Our mission is to help students develop the skills needed to understand, use and build with AI, while helping teachers integrate AI into everyday classroom learning without requiring them to be AI experts.
We aim to create measurable impact by increasing access to AI education, training teachers, enabling practical AI projects, and helping schools provide personalized learning experiences. We will initially focus on schools in Karnataka, including underserved and government schools through affordable programs and CSR partnerships, before expanding across India.
We plan to measure our impact through metrics such as the number of schools and students reached, teachers trained, student participation, completion rates, learning outcomes and the number of students completing practical AI projects. |
NA |
Startup Karnataka foundation |
checked |
| Sep 29, 2026 @ 5:32 AM |
Ankit Kumar Bansal, Rahul Gupta |
ankit.bansal@qoderepo.com |
http://www.linkedin.com/in/ankit-bansal-1a638a98 |
https://www.qaiss.io,%20https://www.qoderepo.com |
8288889451 |
Ankit Bansal — Co-founder / Technical Architect 14+ years of experience across startups, cybersecurity companies, and MNCs, with expertise in software engineering, technical architecture, cloud, security, and enterprise systems. Has held technical and engineering leadership roles, including Engineering Manager and Technical Architect, and currently focuses on AI security, assurance, evaluation, and agentic systems. Rahul Gupta— Co-founder / Product & Technology Nearly 8 years of experience across product management, SaaS, software development, and quality assurance. An alumnus of IIT-BHU and IISc, Rahul has led products from conception through launch, working across product, engineering, sales, and business teams with a focus on product strategy, execution, and customer needs. |
We are cousin brothers and have known each other throughout our lives. With complementary professional experiences in technology, security, and product, we decided to explore building something together. We started working together on the idea that became QAISS and have been collaborating closely for over 1.5 years. |
1 |
Our biggest strength is the combination of deep technical and security experience with strong product and execution capabilities. We complement each other well - bringing together architecture, AI security, engineering, product strategy, and customer understanding to turn a complex problem into a practical, scalable solution. |
QodeRepo Technologies Private Limited |
https://www.qaiss.io, https://www.qoderepo.com |
Bengaluru |
QAISS is an AI assurance platform that continuously tests AI systems for security, reliability, and compliance before and after deployment. |
Organizations are rapidly deploying generative and agentic AI, but testing these systems for security, reliability, and unintended behavior remains fragmented and largely manual. Teams often lack a unified way to identify vulnerabilities such as prompt injection, data leakage, hallucinations, unsafe outputs, and risks in AI agent workflows before they reach production. |
QAISS is an AI Assurance platform that automates the testing and evaluation of AI systems across security, reliability, RAG, agentic workflows, tools, and policy requirements. It runs structured and adaptive tests, captures evidence, maps findings to relevant frameworks, and provides actionable reports—enabling organizations to continuously validate their AI systems before and after deployment. |
QAISS differentiates itself by taking a **system-level approach to AI assurance**, rather than testing only the underlying model. It brings security testing, RAG evaluation, agent and tool validation, adversarial testing, and policy checks into one platform. Our growing testing methodology, real-world pilot experience, and accumulated evaluation data provide a foundation for building proprietary testing intelligence over time. |
MVP |
Pilots |
Our primary customers are **mid-sized and large enterprises, AI product companies, and organizations developing or deploying generative and agentic AI systems**. We initially focus on technology, financial services, healthcare, retail, and other regulated or AI-intensive sectors where security, reliability, and responsible AI are critical to production adoption. |
US$1.5B — Global AI testing, validation and assurance market. |
US$500M — Enterprise AI testing and assurance opportunity across North America, Europe, India and Singapore, focusing on organizations deploying GenAI and agentic AI. |
US$10–15M — Initial 5-year obtainable market through enterprise customers, AI product companies and AI assurance/testing engagements. |
Enterprise SaaS subscriptions: Usage-based licenses for continuous AI security, evaluation, red teaming, RAG and agent testing. Professional testing services: Paid AI security assessments, red-team engagements, and specialized validation for organizations that need expert-led testing. Enterprise services: Custom integrations, on-premise/private deployments, compliance mapping, and remediation consulting. |
Mingard, Lakera, HiddenLayer |
Our initial customer acquisition strategy is founder-led and focused on direct engagement with organizations developing or deploying generative and agentic AI. We use targeted outreach to CTOs, CISOs, AI and security teams, followed by discovery discussions and pilot assessments to demonstrate QAISS on real-world AI systems. As we validate the market, we plan to scale through industry partnerships, AI/security communities, thought leadership, referrals, and channel partnerships, converting successful pilots into recurring enterprise engagements and SaaS subscriptions. |
Our initial go-to-market strategy is founder-led, enterprise-focused, and pilot-driven. We will target organizations actively developing or deploying generative and agentic AI, particularly technology, financial services, healthcare, and other regulated sectors. We will use targeted outreach to CTOs, CISOs, AI and security leaders to identify active AI deployments and offer focused pilot assessments using QAISS. Successful pilots will be converted into recurring assurance engagements and enterprise subscriptions. As the product and customer base mature, we plan to scale through strategic partnerships with AI development companies, cybersecurity firms, cloud ecosystems, and industry networks, supported by thought leadership, referrals, and targeted digital marketing. |
Our long-term vision is to make AI assurance a standard part of the AI development and deployment lifecycle, similar to how software testing and security became integral to traditional software engineering. We aim to build QAISS into a global platform that continuously evaluates AI systems for security, reliability, safety, and compliance, enabling organizations to adopt AI with greater trust and confidence. |
Private Limited |
NA |
Yes |
For **“Why are you applying to Avinya?”**, I’d make this more personal and specific to QAISS’s current stage. The key is to show that you are not just looking for funding—you need **mentorship, GTM guidance, industry connections, and a stronger path to commercialization**. > We are applying to Avinya because QAISS is at a stage where the right mentorship, industry access, and go-to-market guidance can significantly accelerate our journey. We want to learn from experienced founders and industry leaders, validate our approach with real customers, strengthen our product and business strategy, and build the right foundation to scale QAISS into a global AI assurance company. |
We want to use the programme to **validate our product-market fit, strengthen our go-to-market strategy, and build meaningful industry connections**. We also want to learn from experienced mentors, refine our product based on customer feedback, develop a stronger enterprise sales approach, and establish a clear path toward commercial growth and scale. |
Yes |
Yes |
Automation layer |
QAISS uses a model-agnostic architecture supporting LLMs accessed through APIs and locally hosted models. The stack includes Python and Node.js/TypeScript for testing and orchestration, with REST APIs for model/application integration. We use automated and adaptive test generation, LLM-based evaluation, semantic analysis, and RAG/agent workflow testing. The platform is designed to integrate with frameworks such as LangChain/LangGraph and observability through OpenTelemetry, with support for cloud and self-hosted deployments. |
QAISS is building a proprietary knowledge base from real-world AI testing and pilot engagements. This includes anonymized test results, attack patterns, failure cases, evaluation scenarios, and remediation insights across different AI applications. Over time, this growing dataset will help us improve test generation, risk detection, benchmarking, and adaptive AI security testing. |
QAISS is building defensibility through its proprietary AI testing methodologies, adaptive test-generation techniques, and knowledge base developed from real-world pilot engagements. As testing volume grows, anonymized failure patterns, attack scenarios, evaluation data, and remediation insights will improve our testing intelligence and enable more effective, context-aware assurance. |
QAISS evaluates model performance and reliability using task-specific accuracy, relevance, groundedness, hallucination rate, response consistency, and failure rates. For RAG systems, we assess retrieval relevance and answer faithfulness. For security and agentic systems, we measure attack success rate, policy-violation rate, data-leakage rate, and unsafe tool/action execution. Results are tracked across test runs to identify regressions and improvement over time. |
QAISS is designed with security and privacy as core principles. Customer data and AI inputs are handled with strict access controls and are isolated between engagements. We follow data-minimisation practices and aim to avoid retaining sensitive customer data beyond what is required for testing. The platform is designed to support secure deployments, including private/self-hosted environments, and our roadmap includes alignment with applicable data-protection, AI governance, and security standards. |
Yes. The main regulatory risks for QAISS come from the rapidly evolving AI, data-protection, and cybersecurity regulatory landscape, particularly when serving enterprises across different jurisdictions. Requirements such as the EU AI Act, Singapore AI governance frameworks, and data-protection laws may affect how customers evaluate, test, store, and process AI-related data.
We see these regulations primarily as an opportunity as well as a risk, as they increase the need for structured AI testing and assurance. QAISS will continue adapting its testing methodologies and compliance mappings as regulations evolve. |
At 10x scale, the main challenges would be test execution capacity, cloud/model inference costs, handling large volumes of evaluation data, and maintaining consistent performance across diverse AI architectures. Customer onboarding and technical support could also become bottlenecks. We plan to address these through scalable infrastructure, asynchronous testing, automation, standardized integrations, and a growing engineering team. |
We currently have a technically experienced founding team with expertise in software engineering, security, product development, and AI assurance. The founders are directly involved in AI/ML evaluation, adversarial testing, RAG and agentic AI testing. As QAISS scales, we plan to build a dedicated AI/ML engineering team to deepen our capabilities in evaluation, test generation, and AI security. |
We have our own Product Benchmarking Framework that we have developed inhouse and the datasets we are using is owned by us. |
QAISS continuously improves its AI capabilities through feedback from real-world testing, pilot engagements, and evaluation outcomes. We use test results, failure patterns, attack responses, and evaluator feedback to refine test-generation strategies, evaluation prompts, and risk-detection models. Over time, this feedback loop helps QAISS generate more relevant tests, reduce false positives, and improve the accuracy and consistency of its assessments. |
QAISS is being built for both the Indian and global markets. We see India as an important market for early validation and customer development, while our broader target includes enterprises and AI product companies across North America, Europe, Singapore, and other AI-adopting markets. Our product is designed to be globally applicable and aligned with international AI security, testing, and assurance frameworks. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Jy2HXi3OygvJ-Ship-AI-faster.-With-confidence.-Final.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Jy2HXi3OygvJ-Ship-AI-faster.-With-confidence.-Final.pdf |
https://www.qaiss.io |
Yes. Our mission is to make AI adoption safer, more reliable, and more accessible by making continuous AI assurance a practical part of the development lifecycle. We aim to help organizations identify security, reliability, and responsible-AI risks before they affect users, while also building AI security and testing capabilities through skilled talent and real-world learning. |
No |
Social Media |
checked |
| Sep 28, 2026 @ 8:26 PM |
Aparna Kumari |
kaparna869@gmail.com |
https://www.linkedin.com/company/108812370/admin/page-posts/published/ |
http://www.aicoresystem.in |
+91 97394 63988 |
Computer Science |
We met through our shared interest in technology and cybersecurity and started working together while building AICORESYSTEM. We have been working together for about a year, collaborating on product development, content creation, and growing our AI-powered cybersecurity learning platform. |
1 |
Our biggest strength is our ability to combine technical execution with a strong understanding of the cybersecurity education market. We move quickly, build practical solutions, learn from user feedback, and continuously improve our AI-powered learning platform with a lean and execution-focused approach. |
AICORESYSTEM |
https://www.aicoresystem.in |
Ahmedabad, Gujarat, India |
AICORESYSTEM is an AI-powered cybersecurity learning platform that helps students and professionals gain practical skills through interactive labs, quizzes, and personalized learning. |
India faces a growing cybersecurity talent gap, while many students struggle to gain practical experience beyond theoretical learning. Existing learning platforms often lack hands-on practice, personalized guidance, and industry-focused skill development. This results in graduates who are less prepared for real cybersecurity roles. |
AICORESYSTEM combines AI-powered personalized learning with practical cybersecurity training through a web-based LMS and mobile application. The platform offers structured learning paths, interactive quizzes, hands-on labs, progress tracking, and AI-assisted recommendations to help learners build job-ready cybersecurity skills. |
Our platform combines AI-driven personalization with practical cybersecurity education in a single ecosystem. Unlike traditional learning platforms, we focus on hands-on skill development, adaptive learning recommendations, progress analytics, and a mobile-first experience. Our growing content library, proprietary quiz database, and AI-powered learning roadmap create long-term differentiation. |
MVP |
Users |
Our primary customers are college students, fresh graduates, and early-career professionals who want to build practical cybersecurity and AI skills. We also serve educational institutions and training organizations looking for industry-ready cybersecurity learning solutions. |
$20+ billion globally. The global cybersecurity education and training market is driven by increasing demand for cybersecurity professionals, AI adoption, and digital skill development across industries. |
$1.5+ billion. Our serviceable market includes India's cybersecurity training, AI education, and professional upskilling segment, covering universities, colleges, and working professionals. |
$10 million over the next 5 years. We initially focus on students and professionals in India through our digital-first platform, with expansion into institutional partnerships and enterprise training. |
We operate a freemium SaaS and EdTech model. Revenue comes from premium course subscriptions, certification programs, institutional licensing for colleges, enterprise cybersecurity training, and future AI-powered premium features. |
Our competitors include Coursera, Udemy, TryHackMe, Hack The Box, EC-Council Learning, Great Learning, and Simplilearn. We differentiate ourselves through AI-powered personalized learning combined with practical cybersecurity education. |
We acquire customers through content marketing, LinkedIn, YouTube, Instagram, SEO, referral programs, college outreach, webinars, workshops, and partnerships with educational institutions. |
We will first build strong traction among students through free learning content and practical cybersecurity challenges, convert engaged users into premium subscribers, expand through campus ambassador programs and institutional partnerships, and later scale into enterprise cybersecurity training. |
Our vision is to become India's leading AI-powered cybersecurity learning platform, enabling millions of learners to gain industry-ready skills through personalized AI guidance, practical labs, and globally recognized certification pathways. |
Registered Startup (Partnership Firm) |
Bootstrapped. No external funding raised yet. |
Yes |
We are applying to Avinya to accelerate AICORESYSTEM's growth through mentorship, industry connections, and investor access. We believe the programme can help us strengthen our AI-powered cybersecurity learning platform, validate our product with experts, and prepare for scalable growth. |
During the programme, we aim to improve our AI-powered learning features, increase user acquisition, build strategic partnerships with educational institutions, refine our business model, and become investment-ready for future fundraising. |
yes |
Yes |
Supporting feature |
Our platform combines a web-based LMS and mobile application with an AI-powered recommendation layer. Learner progress, quiz performance, and course interactions are used to generate personalized learning recommendations. We are building an LLM-powered learning assistant that will provide contextual explanations, guided learning paths, and cybersecurity concept support, while our backend manages user data, progress tracking, and analytics. |
We are building a proprietary dataset from learner interactions, quiz performance, learning progress, and cybersecurity-focused assessment content generated within our platform. This domain-specific educational data helps us improve personalization while respecting user privacy. |
Our differentiation comes from combining AI-driven personalization with cybersecurity-specific learning workflows. Instead of offering generic AI responses, we focus on structured learning recommendations, practical assessments, and domain-specific guidance that improves as platform engagement grows. |
We evaluate performance using recommendation relevance, quiz completion rate, learning progression, user engagement, response quality reviews, and accuracy checks for AI-generated educational content. Human review is used for critical learning materials where required. |
We minimize personal data collection, use secure authentication, encrypt sensitive data during transmission, follow role-based access controls, and continuously improve our security practices. Our platform is designed with privacy-by-design principles, and we aim to align with applicable Indian data protection requirements as we scale. |
Potential regulatory considerations include evolving AI governance standards, data protection regulations in India, and compliance requirements for handling learner data. We monitor these developments and design our platform to adapt to future compliance requirements. |
The biggest scaling challenges would be AI inference costs, response latency, and infrastructure capacity for personalized recommendations. We plan to address this through cloud scaling, caching, optimized model selection, and efficient backend architecture. |
Our founding team is actively building the AI roadmap while combining expertise in cybersecurity, software development, backend systems, and product development. We are expanding our AI capabilities through continued development, mentorship, and strategic hiring as we grow. |
We currently use platform-generated educational content, internally developed cybersecurity quizzes, publicly available educational resources where permitted, and user interaction data collected within our platform in accordance with applicable policies. Our proprietary content and learner interaction data are owned by the platform. |
We will improve performance through user feedback loops, continuous evaluation of recommendation quality, expanding cybersecurity-specific training content, improving prompts and workflows, and incorporating anonymized learning insights to refine personalization. |
We are initially focused on India while building a scalable platform that can serve learners globally as our AI-powered cybersecurity education ecosystem expands. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/JbsAKMuDbqzE-AICORESYSTEM_Pitch_Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/JbsAKMuDbqzE-AICORESYSTEM_Pitch_Deck.pdf |
https://lnkd.in/p/ghhr7uu8 |
Yes. AICORESYSTEM is building a mission-driven platform to reduce the cybersecurity skills gap by making practical, industry-relevant learning more accessible. We aim to help students, fresh graduates, and working professionals become job-ready through hands-on learning, AI-assisted guidance, and affordable education, with a long-term goal of strengthening India's cybersecurity workforce. |
NA |
Website |
checked |
| Sep 28, 2026 @ 7:09 PM |
Dr Asutosh Dwivedi |
connect@karmaalab.com |
https://www.linkedin.com/in/ashudwi/ |
https://www.karmaalab.com/ |
+91 8073825735 |
Dr. Ashutosh Dwivedi – Founder and CEO: Deep-tech entrepreneur, researcher and former IIT faculty with 25+ years of experience across AI, Data Science, IoT, NLP, Machine Learning and technology innovation. He leads the product vision, technology strategy and development of HRI. |
The founding team came together through Karmaa Lab's work in AI, data science and technology product development. The team has worked together through the development of Karmaa Lab's technology products, with a shared focus on applying AI to practical business and education challenges. |
2 |
Our biggest strength is the combination of deep technical expertise, product-building experience and a strong understanding of the education-to-employment ecosystem. We combine AI, data science and software engineering capabilities with practical knowledge of recruitment and campus-to-corporate challenges, enabling us to build technology around real user needs. |
Karmaa Lab |
https://www.karmaalab.com/ |
Bengaluru |
AI-powered recruitment and interview screening platform helping organizations identify and assess talent faster. |
Recruiters and placement teams spend significant time screening candidates and conducting repetitive assessments, while students often struggle to prepare for modern hiring processes. Existing processes can be manual, time-consuming and inconsistent. HRI aims to make screening and interview preparation more structured and efficient. |
HRI is an AI-powered screening and interview platform that supports automated assessments, AI-based mock interviews, voice/audio interactions and structured candidate evaluation. It helps organizations streamline screening while enabling students and job seekers to practice modern hiring formats and improve interview readiness. |
HRI combines AI-based candidate screening with interview simulation and structured evaluation in one platform. Its focus on the campus-to-corporate transition connects candidate preparation with real-world hiring workflows, creating value for both candidates and recruiters. |
MVP |
Pilots |
College placement cells, universities, training institutes, recruiters, HR teams and companies hiring entry-level talent, along with students and fresh graduates preparing for interviews and employment. |
NA |
NA |
NA |
B2B SaaS subscriptions and institutional licensing for recruiters, companies, universities and placement/training institutes, with potential usage-based plans for assessments and mock interviews. |
HireVue, Mercer | Mettl, SHL, HackerRank, TestGorilla and other AI-based assessment, recruitment screening and interview platforms. |
We acquire customers through direct B2B outreach to HR teams, recruiters, universities, placement cells and training institutes, supported by product demos, pilot programs, partnerships, LinkedIn outreach, webinars and referrals. |
We will initially focus on B2B partnerships with colleges, placement cells, training institutes and recruiters. We will use pilot deployments to demonstrate value, build institutional references and expand through partnerships, referrals and direct sales. A self-serve student offering can support broader adoption. |
Our long-term vision is to build an intelligent talent assessment and career-readiness ecosystem connecting students, institutions and employers. HRI aims to make screening and interview preparation more efficient, structured and accessible while helping organizations identify suitable talent faster. |
Private Limited Company – Incorporated in India |
No external funding raised |
Yes |
We are applying to Avinya to accelerate HRI's product development, validate product-market fit, strengthen our go-to-market strategy and gain access to mentors, industry networks and potential strategic partnerships. We believe the program can help us move from product development toward scalable adoption. |
We want to validate HRI with target customers, strengthen our product and AI capabilities, establish institutional and industry partnerships, refine our business model and go-to-market strategy, and build a scalable customer acquisition pipeline. We also want to prepare HRI for sustainable growth and future fundraising. |
Yes, we are open to hybrid and virtual participation. |
Yes |
Core engine |
HRI uses AI/ML and generative AI capabilities for candidate assessment, mock interviews, voice/audio interactions and structured evaluation. The architecture supports automated screening and interview simulations, with outputs designed to help candidates and recruiters make better-informed decisions. |
Our advantage comes from building domain-specific datasets around interview questions, candidate responses, assessments and recruitment workflows. As usage grows, these interactions can support continuous refinement of assessment and interview experiences. |
HRI combines AI-powered screening, assessment and interview simulation with a focus on the campus-to-corporate transition. Its integration of candidate preparation and recruitment workflows creates a domain-specific product experience rather than a generic AI tool. |
We evaluate performance through assessment accuracy, consistency of outputs, relevance of generated questions and responses, user feedback, and comparison against expected assessment outcomes. Human review can be incorporated where higher-confidence evaluation is required. |
We follow privacy-by-design principles, limiting data collection to information required for the product workflow and applying appropriate access controls. Candidate and assessment data should be handled according to applicable privacy, security and data-protection requirements. |
HRI operates in the recruitment and education technology space, so key considerations include data privacy, consent, secure handling of candidate information, transparency of AI-assisted assessments and evolving regulations governing AI-based decision-support systems. |
At 10x scale, the main challenges could include infrastructure capacity, inference costs, response latency, data-management volume, quality consistency and monitoring. We plan to address these through scalable infrastructure, usage monitoring, automated evaluation and continuous model optimization. |
Yes. Our team works across AI/ML, data science, software engineering and product development, with expertise applied to AI-based screening, assessment and interview workflows. |
We use domain-specific recruitment and interview data, including assessment content, interview questions, candidate interactions and product-generated data. Data ownership and licensing are managed according to the source and applicable usage rights. |
We will continuously improve performance through user feedback, evaluation of model outputs, human review, error analysis and expansion of domain-specific datasets. Usage patterns and assessment outcomes will help identify areas for refinement while maintaining appropriate privacy controls. |
Both - India initially, with a long-term focus on global markets. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/nTeD7ACTCufn-HRI-Pitch-deck-Avinya-Cohort-3-v1.0.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/nTeD7ACTCufn-HRI-Pitch-deck-Avinya-Cohort-3-v1.0.pdf |
http://. |
Yes. HRI aims to make the transition from education to employment more accessible and effective. We help students and fresh graduates build interview readiness through AI-powered screening, assessments and mock interviews, while helping organizations streamline talent screening. Our goal is to reduce hiring inefficiencies and create better opportunities for emerging talent. |
NA |
NA |
checked |
| Sep 28, 2026 @ 1:30 PM |
Aditya Raj Singh, Prateek Kurkanji |
aditya@kompete.ai |
https://www.linkedin.com/in/aditya-raj-singh-9347b172/ |
https://www.kompete.ai |
8265997050 |
COO, SuperAGI & Contlo |
We know each other from 2008. |
one |
We back each other always |
Kompete.ai |
https://www.kompete.ai |
Bengaluru |
Competitive Intelligence Analyst for B2B companies |
Manual Record keeping of competitors across 100+ sources |
Analyst which can compile info, summarize and tell actionable insights, keeps monitoring for changes |
The intelligence layer that drives the strategy and connects the dots across 100+ sources |
Users |
Pilots |
B2B Companies |
NA |
NA |
NA |
Its subscription pricing with per competitor axis. |
Klue, Crayon |
Founder Outreach |
Target India, US and UK geographies. Industries with a lot of competiton. |
It should be market facing intelligence layer that informs and guides the decision making across teams and creates a |
NA |
NA |
two |
Avinya opens the door to motivated talent who can run a system of agents |
Get to 10 paying customers |
Hybrid |
one |
one |
|
|
|
|
|
|
|
|
|
|
Both |
|
|
|
|
|
|
| Sep 28, 2026 @ 12:24 PM |
Vishwajeet Bhagyawant, Prof. Bakul Rao |
vishwajeet@krishiboot.com |
https://www.linkedin.com/in/vishwajeetbb |
https://krishiboot.com |
+91 8766543498 |
Vishwajeet Bhagyawant, Founder and CEO: B.Tech, IIT Bombay (Metallurgical Engineering, 2018). Building Agriyan full time since 2021 across four hardware generations; owns hardware architecture, the AI stack, field trials, fundraising and partnerships. Prof. Bakul Rao, Co-founder: IIT Bombay faculty, academic and research guidance. |
Vishwajeet is an IIT Bombay alumnus (Metallurgy, 2018) and started building Agriyan in 2021. Prof. Bakul Rao is IIT Bombay faculty and came on board as co-founder as the company took shape inside the IIT Bombay ecosystem; the two have worked together on Agriyan since 2024. Krishiboot is incubated at SINE, IIT Bombay, and the engineering team (R&D engineers Deep Naik and Gautam Mishra, AI/ML engineer Nandita N) has joined since 2025. |
2 |
We build the whole machine ourselves, and we are honest about where it stands.
Five years and four hardware generations have produced an in-house stack end to end: a custom 48V implement controller board, a multispectral camera head at roughly Rs 10,000 bill of materials, and three on-board vision models that share one architecture and fit together on a single Jetson Orin NX. The people who design the models also run the field trials, on cotton, across three soil types, so every model is trained and tested on real Indian smallholder fields rather than lab data.
And we report what is not proven alongside what is. Every figure we quote has a test record, a method and a sample size behind it. |
Krishiboot Technologies Pvt. Ltd. (product: Agriyan) |
https://krishiboot.com |
Mumbai, Maharashtra (incubated at SINE, IIT Bombay) |
Agriyan is an AI-powered autonomous field robot that weeds, sprays and sows Indian smallholder farms plant by plant, sold per acre as a service. |
Indian smallholders spray whole fields on a fixed calendar because nothing affordable can see individual plants, so chemical lands where it is not needed, and weeding and spraying slip because labour is scarce on the day the crop needs it. |
Agriyan is a battery-electric autonomous ground robot, about 250 kg, built for 0.4 to 2 hectare Indian farms. On a single pass it drives the crop row on its own, identifies every crop stem and weed, grades plant health, and actuates the fitted tool at plant resolution instead of treating the whole field. Weeding, spraying and sowing modules share one platform and one compute stack.
Navigation needs no RTK, no LiDAR and no pre-mapping: depth is converted to surface normals to find vertical structures, so the crop rows themselves are the map. A navigation model produces row corridor, plant census, obstacle class and crop-safety keypoints in one 28 ms forward pass. Human detection is a hardcoded controller interrupt the model cannot override.
We do not sell the machine. We own and operate the fleet and bill the FPO per acre per operation, so the farmer pays for field work done rather than for equipment he cannot afford. Every pass adds to a per-plant health record of the farm.
Status, stated plainly: navigation and weeding run on the current platform; precision spraying was validated on our previous platform and the new spray module is designed, not yet fabricated; sowing is not yet built. |
Navigation without RTK, LiDAR or pre-mapping; a Rs 10,000 multispectral sensing head against USD 6,000 to 15,000 for commercial equivalents; three implements on one platform; and a per-plant dataset of Indian smallholder fields that only grows because we operate the fleet ourselves. |
MVP |
Pilots |
Farmer Producer Organisations in the Maharashtra cotton belt, contracting on behalf of member farmers on 0.4 to 2 hectares. Also institutional programmes that hold farmer budgets and a captive farmer base (Ambuja Foundation, Lupin Foundation, WWF-India, World Bank SMART Cotton). From 2029, Custom Hiring Centres buying through the SMAM subsidy channel. |
Rs 72,000 Cr per year: 67M acres of cotton, soybean, grapes and tomato in India, valued at what an acre already spends on labour for sowing, weeding and spraying |
Rs 21,000 Cr per year: 21M acres reachable with the current machine geometry |
Rs 50 Cr annual service revenue by 2031, about 370 machines serving 45,000 acres. Fleet size, not market size, is the binding constraint. |
Robotics-as-a-Service. We own and operate the fleet and bill the FPO per acre per operation (spray, weed, sow), priced inside what the farmer already pays in labour for the same work, and paid on each completed pass against a GPS-tagged record. Plant health monitoring is bundled, not charged. Machine sale opens from 2029 after FMTTI certification. |
Direct: Niqo Robotics (tractor-mounted precision spraying, India) and John Deere See and Spray. Indirect: drone spray operators such as Garuda Aerospace, conventional sprayer makers such as M.I.T.R.A. Agro, and the real incumbent, manual contract labour and local Custom Hiring Centres. |
Through institutions that already hold farmer budgets and a captive farmer base (CSR foundations, WWF-India, World Bank SMART Cotton), and directly through FPO boards within 250 km of our hub. One FPO agreement covers 300 to 2,000 member farms, so we sell to about 10,000 FPOs, not 146 million farmers. |
Cotton in Maharashtra first, the crop whose open rows fit the machine best. 2026: integrated field demonstration plus a same-day labour and chemical baseline. 2027: five machines, first FPO season agreement and first revenue, operated inside a 250 km radius so a breakdown costs hours, not days. 2028: 20 machines across multiple FPOs. 2029: FMTTI certification opens machine sales through the SMAM and CHC subsidy channel. |
Every smallholder field in India farmed plant by plant.
We want Agriyan fleets doing the field work for FPOs across India's cotton, soybean and horticulture belts, with every pass adding to a per-plant record of Indian farms that makes the next spray smaller and the next season more predictable.
The long-term company is three things: a field-operations network that sells work by the acre, a machine maker once the platform is certified, and the most detailed ground truth that exists on how Indian smallholder crops actually grow. |
Private limited company incorporated in India: Krishiboot Technologies Private Limited. DPIIT-recognised, GST-registered in Maharashtra, incubated at SINE, IIT Bombay. |
Rs 1.1 Cr equity from the IIT Bombay 1970 Batch TRA team, invested through SINE, IIT Bombay as CCPS in tranches (latest Rs 24.57 L, Aug 2026). Rs 50 L BIRAC BIG grant sanctioned, agreement pending, not yet disbursed. Now raising a Rs 2 Cr seed round. |
Yes |
Agriyan is AI first: on-board vision models decide every steering and treatment action. We want Avinya's AI mentors to pressure-test our model stack before we scale it, and IIT alumni and investor access to help close our seed round. The Rs 10 L would go into fabricating the precision spray module, the one gap between our validated AI stack and measured chemical savings in the field. |
Three things by Demo Day: the spray module fabricated and validated against a same-day blanket-spray control plot, giving our first measured chemical reduction on the current machine; a concurrent-load benchmark with all AI models running together on one Jetson; and at least one signed FPO letter of intent for kharif 2027, alongside progress on our seed round. |
Yes. We are based in Mumbai, so we would join regular sessions virtually and travel to Bengaluru for Demo Day, reviews and any mandatory in-person sessions. |
Yes |
Core engine |
Everything that steers or treats runs on the robot, on one NVIDIA Jetson Orin NX 16GB, fully offline.
Sensing (KrishiSense): Intel RealSense D435i for RGB, depth and IMU, plus two Sony IMX477 sensors with 720 nm red-edge and 850 nm NIR bandpass filters, hardware-triggered from the RealSense timestamp to 3 to 5 ms sync error.
On-board models:
1. CropNav (navigation, Jetson GPU, 28 ms): one forward pass outputs row corridor mask, plant census, obstacle class and distance, and crop-safety keypoints. It replaces four separate networks, so every steering decision uses four simultaneously current outputs.
2. WeedNet-3D (weed identification and crop avoidance, Jetson DLA, about 50 ms): weed species box and class plus a cotton stem keypoint heatmap for 3D stem localisation.
3. MSCD-Net (per-plant health, Jetson DLA, 60+ FPS): plant box, 11-class disease label, stem keypoint, using the multispectral channels.
All three are one architecture trained three times: a five-stage depthwise separable backbone, depth-derived attention, distance-adaptive convolution, multi-scale pooling, a BiFPN neck and an anchor-free head, trained with a geometric consistency loss. That reuse is why three networks plus spectral computation fit on one edge device.
Off-board: an EfficientNet-B0 leaf-diagnosis model (8 classes with GradCAM++ explanations) serves farmer phone photos from AWS in under a second.
Runtime: ROS 2, offline-first SQLite on NVMe mirroring a cloud PostgreSQL store, synced after each pass via a queue, because about 750 MB per hectare cannot stream over rural 4G. |
Yes, and it is the part no competitor can buy.
We own roughly 5,200 real field frames of Indian smallholder conditions: about 2,000 navigation frames from cotton fields in Nashik, Amravati and Yavatmal, 1,200 cotton weed images across 12 species, and 2,000 leaf images, all labelled in-house with growth-stage stratified labels and weed labels scored against two agronomists (Cohen's kappa 0.86). Alongside that sit about 4,700 simulated frames from Gazebo and a CycleGAN pipeline for cold-starting new crops.
There is no comparable public dataset for Indian smallholder fields at 0.16 hectare plot scale. Because we operate the fleet rather than sell it, every pass adds GPS-tagged, per-plant data to this record, so the advantage compounds with every acre served. |
Four things, in order of how hard they are to copy.
1. The data. Per-plant field data from Indian smallholder farms is collected only by operating in those fields, season after season. Our service model is designed to keep generating it.
2. Geometry-driven perception. Converting depth to surface normals makes detection depend on plant structure rather than colour, which is why one stack works on black Vertisol, red laterite and sandy loam without retraining. Colour-based systems have to be retrained per soil and light condition.
3. Hardware-model co-design. A Rs 10,000 multispectral head, a shared model architecture that fits three networks on one Jetson, and a single-pass navigation network are designed together. Copying the models without the sensor and compute design does not reproduce the cost.
4. IP. Two patent applications in progress with the Indian Patent Office: one on the reconfigurable multi-implement platform, one on the single-pass quad-task navigation network, path lookahead and controller. Neither is granted yet. |
Every figure below has a test record, a method and a sample size. We report navigation in centimetres and detection against human experts.
Navigation: 2.8 cm mean lateral drift against a chalk line (6.1 cm max) over 12 consecutive rows with no operator takeover; 4.2 cm mean on irregular row spacing; 5.8 cm mean on a 15 degree slope.
Stem localisation: 12 mm mean 3D Euclidean error (1 sigma).
Weed identification: per-species precision 0.76 to 0.94, scored against two agronomists at Cohen's kappa 0.86.
Disease classification: mAP 0.83 at IoU 0.5 on board; 89.2% top-1 on the leaf-scan model.
Latency: 28 ms navigation pass, about 50 ms weed model, 60+ FPS health model.
Safety: human detection triggers at 1.5 m and stops in about 0.31 m, with a deliberately low 0.65 threshold.
Reliability in the field is judged by operator takeovers per row and by performance across soil types, not just test-set accuracy.
What we have not yet measured, stated plainly: each latency was measured with that model running alone, and concurrent load with all models together has not been benchmarked; weed removal rate has not yet been scored against a formal protocol. |
The robot captures images of crops and soil, not people, and processes them on the device, fully offline. Data leaves the robot only after a pass, through a queued sync to our cloud store.
We treat farm location and farmer identity as personal data under the Digital Personal Data Protection Act 2023. Data is collected under the FPO service agreement, used to deliver the service and improve our models, and returned to the FPO as a per-farm season report.
On safety, which matters more than data for a field robot: human detection is a hardcoded controller interrupt that no model output can override.
We are a pre-commercial company and will formalise encryption, access control and data-retention policies before the first commercial season in 2027. |
Yes, four, and we plan around each.
1. FMTTI certification is the largest. Selling farm machinery into the SMAM or Custom Hiring Centre subsidy channel needs testing at a Farm Machinery Training and Testing Institute, and no BIS test code exists yet for an autonomous electric ground robot. It does not block our service model, where we own and operate the machines, but it gates machine sales from 2029, so we are starting the process early.
2. Agrochemical handling under the Insecticides Act 1968 and CIB&RC norms: the spray module must meet equipment norms and our operators must follow handling and protective-equipment rules.
3. Autonomous machinery safety: there is no settled Indian standard, so we built human detection as a hard stop and will publish our own safety documentation.
4. Data protection under the DPDP Act 2023 for farm and farmer data. |
Five things, honestly ranked.
1. Edge compute headroom. Each model's latency was measured alone; with all models plus spectral computation running together on one Jetson, we expect contention we have not yet benchmarked.
2. Data pipeline. About 750 MB per hectare per pass cannot stream over rural 4G. At 10x acres the post-pass sync, storage and labelling queue become the bottleneck.
3. Labelling and QA. Expert labels scored against two agronomists do not scale 10x without active learning and a tiered review process.
4. Distribution shift. New crops and row geometries (soybean at 45 cm spacing, for example) and new regions will degrade models trained mostly on Maharashtra cotton.
5. Fleet model management. Versioning, rollback and monitoring of models across tens of machines in the field, with limited connectivity.
These are exactly what we want to work on with Avinya's mentors before we scale, not after. |
Yes. Nandita N is our AI/ML engineer, working on the vision models and training pipeline. Vishwajeet designed the model architecture, the single-pass navigation network and the multispectral sensing head, and still runs the field trials that generate the data. Deep Naik and Gautam Mishra, R&D engineers, cover embedded systems and robot integration, which is where models meet actuators. Prof. P. Udmale of IIT Bombay mentors the agronomy interface. |
Mostly our own. About 5,200 real field frames (navigation, cotton weeds across 12 species, leaves) were collected by us on farms in Maharashtra and labelled in-house, so we own them outright. We also use about 4,700 simulated Gazebo frames we generated ourselves.
Two external inputs, disclosed plainly: ImageNet-pretrained weights for backbone initialisation, and the public PlantVillage dataset as one domain in our CycleGAN cold-start pipeline for leaf images. We are confirming the licence terms for both before commercial deployment, and our real-field data is steadily replacing the PlantVillage share. |
The fleet is the data engine. Every pass logs GPS-tagged, time-synced RGB, depth and multispectral frames with the model's decisions attached.
1. Hard-case mining: frames where confidence is low, where an operator intervened, or where models disagree are queued first for labelling.
2. Expert labelling with agreement checks against agronomists, keeping inter-rater agreement measured, not assumed.
3. Retraining on a seasonal cycle, with growth-stage stratified test sets so a model cannot improve on one stage while regressing on another.
4. Field validation before release: every new model is re-run against the same centimetre-level navigation and precision tests before it goes on a machine.
5. New crops start from the CycleGAN cold-start pipeline and are then replaced with real field data as soon as the fleet operates there.
One gap we are closing now: validating the plant-health model on real field photos rather than simulation. |
India first. The machine is sized for 0.4 to 2 hectare plots and 0.16 hectare plot-scale data, which is where Indian agriculture actually is, and where Western autonomous machines do not fit on price or geometry. Our first market is cotton in Maharashtra, then other Indian cotton and soybean belts.
Later, the same smallholder economics apply across South and Southeast Asia and parts of Africa, but we will not look outside India until the service model is proven here. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/KOrzo6ccsYyw-Krishiboot-Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/KOrzo6ccsYyw-Krishiboot-Pitch-Deck.pdf |
https://drive.google.com/file/d/173o19Jp6b8RHoEjF8G92Yc3xiCTMp82V/view |
Yes. Agriyan exists to cut agrochemical use at source and take the spray tank off the farm worker's back.
Cotton covers roughly 5% of India's cropped land but has historically accounted for close to half of its pesticide use, nearly all of it applied by blanket spraying. Chemical that misses its target runs off into water, drifts, and is inhaled by the person carrying the tank. Treating the plant instead of the field means that chemical is never applied.
The impact we will measure, each against a same-day farmer-practice control plot: kilograms of pesticide active ingredient avoided per acre, spray water saved, diesel and CO2 avoided by a battery-electric machine replacing tractor or engine sprayers, herbicide displaced by mechanical weeding, and hours of human chemical exposure eliminated.
For the farmer, our fee sits inside what he already pays in labour, so his gain is the chemical bill: about Rs 5,400 per acre on our modelled cotton case, which we are now measuring in the field. The category benchmark is about 50 to 60% chemical reduction (John Deere See and Spray, Niqo Robotics); we will quote our own number only once we have measured it on this machine. |
NA |
SINE IIT Bombay |
checked |
| Sep 28, 2026 @ 12:19 PM |
Vishwajeet Bhagyawant, Prof. Bakul Rao |
vishwajeet@krishiboot.com |
https://www.linkedin.com/in/vishwajeetbb |
https://krishiboot.com |
+91 8766543498 |
Vishwajeet Bhagyawant, Founder and CEO: B.Tech, IIT Bombay (Metallurgical Engineering, 2018). Building Agriyan full time since 2021 across four hardware generations; owns hardware architecture, the AI stack, field trials, fundraising and partnerships. Prof. Bakul Rao, Co-founder: IIT Bombay faculty, academic and research guidance. |
Vishwajeet is an IIT Bombay alumnus (Metallurgy, 2018) and started building Agriyan in 2021. Prof. Bakul Rao is IIT Bombay faculty and came on board as co-founder as the company took shape inside the IIT Bombay ecosystem; the two have worked together on Agriyan since 2024. Krishiboot is incubated at SINE, IIT Bombay, and the engineering team (R&D engineers Deep Naik and Gautam Mishra, AI/ML engineer Nandita N) since 2025. |
2 |
We build the whole machine ourselves, and we are honest about where it stands.
Five years and four hardware generations have produced an in-house stack end to end: a custom 48V implement controller board, a multispectral camera head at roughly Rs 10,000 bill of materials, and three on-board vision models that share one architecture and fit together on a single Jetson Orin NX. The people who design the models also run the field trials, on cotton, across three soil types, so every model is trained and tested on real Indian smallholder fields rather than lab data.
And we report what is not proven alongside what is. Every figure we quote has a test record, a method and a sample size behind it. |
Krishiboot Technologies Pvt. Ltd. (product: Agriyan) |
https://krishiboot.com |
Mumbai, Maharashtra (incubated at SINE, IIT Bombay) |
Agriyan is an AI-powered autonomous field robot that weeds, sprays and sows Indian smallholder farms plant by plant, sold per acre as a service. |
Indian smallholders spray whole fields on a fixed calendar because nothing affordable can see individual plants, so chemical lands where it is not needed, and weeding and spraying slip because labour is scarce on the day the crop needs it. |
Agriyan is a battery-electric autonomous ground robot, about 250 kg, built for 0.4 to 2 hectare Indian farms. On a single pass it drives the crop row on its own, identifies every crop stem and weed, grades plant health, and actuates the fitted tool at plant resolution instead of treating the whole field. Weeding, spraying and sowing modules share one platform and one compute stack.
Navigation needs no RTK, no LiDAR and no pre-mapping: depth is converted to surface normals to find vertical structures, so the crop rows themselves are the map. A navigation model produces row corridor, plant census, obstacle class and crop-safety keypoints in one 28 ms forward pass. Human detection is a hardcoded controller interrupt the model cannot override.
We do not sell the machine. We own and operate the fleet and bill the FPO per acre per operation, so the farmer pays for field work done rather than for equipment he cannot afford. Every pass adds to a per-plant health record of the farm.
Status, stated plainly: navigation and weeding run on the current platform; precision spraying was validated on our previous platform and the new spray module is designed, not yet fabricated; sowing is not yet built. |
Navigation without RTK, LiDAR or pre-mapping; a Rs 10,000 multispectral sensing head against USD 6,000 to 15,000 for commercial equivalents; three implements on one platform; and a per-plant dataset of Indian smallholder fields that only grows because we operate the fleet ourselves. |
MVP |
Pilots |
Farmer Producer Organisations in the Maharashtra cotton belt, contracting on behalf of member farmers on 0.4 to 2 hectares. Also institutional programmes that hold farmer budgets and a captive farmer base (Ambuja Foundation, Lupin Foundation, WWF-India, World Bank SMART Cotton). From 2029, Custom Hiring Centres buying through the SMAM subsidy channel. |
Rs 72,000 Cr per year: 67M acres of cotton, soybean, grapes and tomato in India, valued at what an acre already spends on labour for sowing, weeding and spraying |
Rs 21,000 Cr per year: 21M acres reachable with the current machine geometry |
Rs 50 Cr annual service revenue by 2031, about 370 machines serving 45,000 acres. Fleet size, not market size, is the binding constraint. |
Robotics-as-a-Service. We own and operate the fleet and bill the FPO per acre per operation (spray, weed, sow), priced inside what the farmer already pays in labour for the same work, and paid on each completed pass against a GPS-tagged record. Plant health monitoring is bundled, not charged. Machine sale opens from 2029 after FMTTI certification. |
Direct: Niqo Robotics (tractor-mounted precision spraying, India) and John Deere See and Spray. Indirect: drone spray operators such as Garuda Aerospace, conventional sprayer makers such as M.I.T.R.A. Agro, and the real incumbent, manual contract labour and local Custom Hiring Centres. |
Through institutions that already hold farmer budgets and a captive farmer base (CSR foundations, WWF-India, World Bank SMART Cotton), and directly through FPO boards within 250 km of our hub. One FPO agreement covers 300 to 2,000 member farms, so we sell to about 10,000 FPOs, not 146 million farmers. |
Cotton in Maharashtra first, the crop whose open rows fit the machine best. 2026: integrated field demonstration plus a same-day labour and chemical baseline. 2027: five machines, first FPO season agreement and first revenue, operated inside a 250 km radius so a breakdown costs hours, not days. 2028: 20 machines across multiple FPOs. 2029: FMTTI certification opens machine sales through the SMAM and CHC subsidy channel. |
Every smallholder field in India farmed plant by plant.
We want Agriyan fleets doing the field work for FPOs across India's cotton, soybean and horticulture belts, with every pass adding to a per-plant record of Indian farms that makes the next spray smaller and the next season more predictable.
The long-term company is three things: a field-operations network that sells work by the acre, a machine maker once the platform is certified, and the most detailed ground truth that exists on how Indian smallholder crops actually grow. |
Private limited company incorporated in India: Krishiboot Technologies Private Limited. DPIIT-recognised, GST-registered in Maharashtra, incubated at SINE, IIT Bombay. |
Rs 1.1 Cr equity from the IIT Bombay 1970 Batch TRA team, invested through SINE, IIT Bombay as CCPS in tranches (latest Rs 24.57 L, Aug 2026). Rs 50 L BIRAC BIG grant sanctioned, agreement pending, not yet disbursed. Now raising a Rs 2 Cr seed round. |
Yes |
Agriyan is AI first: on-board vision models decide every steering and treatment action. We want Avinya's AI mentors to pressure-test our model stack before we scale it, and IIT alumni and investor access to help close our seed round. The Rs 10 L would go into fabricating the precision spray module, the one gap between our validated AI stack and measured chemical savings in the field. |
Three things by Demo Day: the spray module fabricated and validated against a same-day blanket-spray control plot, giving our first measured chemical reduction on the current machine; a concurrent-load benchmark with all AI models running together on one Jetson; and at least one signed FPO letter of intent for kharif 2027, alongside progress on our seed round. |
Yes. We are based in Mumbai, so we would join regular sessions virtually and travel to Bengaluru for Demo Day, reviews and any mandatory in-person sessions. |
Yes |
Core engine |
Everything that steers or treats runs on the robot, on one NVIDIA Jetson Orin NX 16GB, fully offline.
Sensing (KrishiSense): Intel RealSense D435i for RGB, depth and IMU, plus two Sony IMX477 sensors with 720 nm red-edge and 850 nm NIR bandpass filters, hardware-triggered from the RealSense timestamp to 3 to 5 ms sync error.
On-board models:
1. CropNav (navigation, Jetson GPU, 28 ms): one forward pass outputs row corridor mask, plant census, obstacle class and distance, and crop-safety keypoints. It replaces four separate networks, so every steering decision uses four simultaneously current outputs.
2. WeedNet-3D (weed identification and crop avoidance, Jetson DLA, about 50 ms): weed species box and class plus a cotton stem keypoint heatmap for 3D stem localisation.
3. MSCD-Net (per-plant health, Jetson DLA, 60+ FPS): plant box, 11-class disease label, stem keypoint, using the multispectral channels.
All three are one architecture trained three times: a five-stage depthwise separable backbone, depth-derived attention, distance-adaptive convolution, multi-scale pooling, a BiFPN neck and an anchor-free head, trained with a geometric consistency loss. That reuse is why three networks plus spectral computation fit on one edge device.
Off-board: an EfficientNet-B0 leaf-diagnosis model (8 classes with GradCAM++ explanations) serves farmer phone photos from AWS in under a second.
Runtime: ROS 2, offline-first SQLite on NVMe mirroring a cloud PostgreSQL store, synced after each pass via a queue, because about 750 MB per hectare cannot stream over rural 4G. |
Yes, and it is the part no competitor can buy.
We own roughly 5,200 real field frames of Indian smallholder conditions: about 2,000 navigation frames from cotton fields in Nashik, Amravati and Yavatmal, 1,200 cotton weed images across 12 species, and 2,000 leaf images, all labelled in-house with growth-stage stratified labels and weed labels scored against two agronomists (Cohen's kappa 0.86). Alongside that sit about 4,700 simulated frames from Gazebo and a CycleGAN pipeline for cold-starting new crops.
There is no comparable public dataset for Indian smallholder fields at 0.16 hectare plot scale. Because we operate the fleet rather than sell it, every pass adds GPS-tagged, per-plant data to this record, so the advantage compounds with every acre served. |
Four things, in order of how hard they are to copy.
1. The data. Per-plant field data from Indian smallholder farms is collected only by operating in those fields, season after season. Our service model is designed to keep generating it.
2. Geometry-driven perception. Converting depth to surface normals makes detection depend on plant structure rather than colour, which is why one stack works on black Vertisol, red laterite and sandy loam without retraining. Colour-based systems have to be retrained per soil and light condition.
3. Hardware-model co-design. A Rs 10,000 multispectral head, a shared model architecture that fits three networks on one Jetson, and a single-pass navigation network are designed together. Copying the models without the sensor and compute design does not reproduce the cost.
4. IP. Two patent applications in progress with the Indian Patent Office: one on the reconfigurable multi-implement platform, one on the single-pass quad-task navigation network, path lookahead and controller. Neither is granted yet. |
Every figure below has a test record, a method and a sample size. We report navigation in centimetres and detection against human experts.
Navigation: 2.8 cm mean lateral drift against a chalk line (6.1 cm max) over 12 consecutive rows with no operator takeover; 4.2 cm mean on irregular row spacing; 5.8 cm mean on a 15 degree slope.
Stem localisation: 12 mm mean 3D Euclidean error (1 sigma).
Weed identification: per-species precision 0.76 to 0.94, scored against two agronomists at Cohen's kappa 0.86.
Disease classification: mAP 0.83 at IoU 0.5 on board; 89.2% top-1 on the leaf-scan model.
Latency: 28 ms navigation pass, about 50 ms weed model, 60+ FPS health model.
Safety: human detection triggers at 1.5 m and stops in about 0.31 m, with a deliberately low 0.65 threshold.
Reliability in the field is judged by operator takeovers per row and by performance across soil types, not just test-set accuracy.
What we have not yet measured, stated plainly: each latency was measured with that model running alone, and concurrent load with all models together has not been benchmarked; weed removal rate has not yet been scored against a formal protocol. |
The robot captures images of crops and soil, not people, and processes them on the device, fully offline. Data leaves the robot only after a pass, through a queued sync to our cloud store.
We treat farm location and farmer identity as personal data under the Digital Personal Data Protection Act 2023. Data is collected under the FPO service agreement, used to deliver the service and improve our models, and returned to the FPO as a per-farm season report.
On safety, which matters more than data for a field robot: human detection is a hardcoded controller interrupt that no model output can override.
We are a pre-commercial company and will formalise encryption, access control and data-retention policies before the first commercial season in 2027. |
Yes, four, and we plan around each.
1. FMTTI certification is the largest. Selling farm machinery into the SMAM or Custom Hiring Centre subsidy channel needs testing at a Farm Machinery Training and Testing Institute, and no BIS test code exists yet for an autonomous electric ground robot. It does not block our service model, where we own and operate the machines, but it gates machine sales from 2029, so we are starting the process early.
2. Agrochemical handling under the Insecticides Act 1968 and CIB&RC norms: the spray module must meet equipment norms and our operators must follow handling and protective-equipment rules.
3. Autonomous machinery safety: there is no settled Indian standard, so we built human detection as a hard stop and will publish our own safety documentation.
4. Data protection under the DPDP Act 2023 for farm and farmer data. |
Five things, honestly ranked.
1. Edge compute headroom. Each model's latency was measured alone; with all models plus spectral computation running together on one Jetson, we expect contention we have not yet benchmarked.
2. Data pipeline. About 750 MB per hectare per pass cannot stream over rural 4G. At 10x acres the post-pass sync, storage and labelling queue become the bottleneck.
3. Labelling and QA. Expert labels scored against two agronomists do not scale 10x without active learning and a tiered review process.
4. Distribution shift. New crops and row geometries (soybean at 45 cm spacing, for example) and new regions will degrade models trained mostly on Maharashtra cotton.
5. Fleet model management. Versioning, rollback and monitoring of models across tens of machines in the field, with limited connectivity.
These are exactly what we want to work on with Avinya's mentors before we scale, not after. |
Yes. Nandita N is our AI/ML engineer, working on the vision models and training pipeline. Vishwajeet designed the model architecture, the single-pass navigation network and the multispectral sensing head, and still runs the field trials that generate the data. Deep Naik and Gautam Mishra, R&D engineers, cover embedded systems and robot integration, which is where models meet actuators. Prof. P. Udmale of IIT Bombay mentors the agronomy interface. |
Mostly our own. About 5,200 real field frames (navigation, cotton weeds across 12 species, leaves) were collected by us on farms in Maharashtra and labelled in-house, so we own them outright. We also use about 4,700 simulated Gazebo frames we generated ourselves.
Two external inputs, disclosed plainly: ImageNet-pretrained weights for backbone initialisation, and the public PlantVillage dataset as one domain in our CycleGAN cold-start pipeline for leaf images. We are confirming the licence terms for both before commercial deployment, and our real-field data is steadily replacing the PlantVillage share. |
The fleet is the data engine. Every pass logs GPS-tagged, time-synced RGB, depth and multispectral frames with the model's decisions attached.
1. Hard-case mining: frames where confidence is low, where an operator intervened, or where models disagree are queued first for labelling.
2. Expert labelling with agreement checks against agronomists, keeping inter-rater agreement measured, not assumed.
3. Retraining on a seasonal cycle, with growth-stage stratified test sets so a model cannot improve on one stage while regressing on another.
4. Field validation before release: every new model is re-run against the same centimetre-level navigation and precision tests before it goes on a machine.
5. New crops start from the CycleGAN cold-start pipeline and are then replaced with real field data as soon as the fleet operates there.
One gap we are closing now: validating the plant-health model on real field photos rather than simulation. |
India first. The machine is sized for 0.4 to 2 hectare plots and 0.16 hectare plot-scale data, which is where Indian agriculture actually is, and where Western autonomous machines do not fit on price or geometry. Our first market is cotton in Maharashtra, then other Indian cotton and soybean belts.
Later, the same smallholder economics apply across South and Southeast Asia and parts of Africa, but we will not look outside India until the service model is proven here. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/SXJZxo70O6Tz-Krishiboot-Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/SXJZxo70O6Tz-Krishiboot-Pitch-Deck.pdf |
https://drive.google.com/file/d/173o19Jp6b8RHoEjF8G92Yc3xiCTMp82V/view |
Yes. Agriyan exists to cut agrochemical use at source and take the spray tank off the farm worker's back.
Cotton covers roughly 5% of India's cropped land but has historically accounted for close to half of its pesticide use, nearly all of it applied by blanket spraying. Chemical that misses its target runs off into water, drifts, and is inhaled by the person carrying the tank. Treating the plant instead of the field means that chemical is never applied.
The impact we will measure, each against a same-day farmer-practice control plot: kilograms of pesticide active ingredient avoided per acre, spray water saved, diesel and CO2 avoided by a battery-electric machine replacing tractor or engine sprayers, herbicide displaced by mechanical weeding, and hours of human chemical exposure eliminated.
For the farmer, our fee sits inside what he already pays in labour, so his gain is the chemical bill: about Rs 5,400 per acre on our modelled cotton case, which we are now measuring in the field. The category benchmark is about 50 to 60% chemical reduction (John Deere See and Spray, Niqo Robotics); we will quote our own number only once we have measured it on this machine. |
NA |
SINE, IIT Bombay |
checked |
| Sep 28, 2026 @ 12:12 PM |
Vishwajeet Bhagyawant, Prof. Bakul Rao |
vishwajeet@krishiboot.com |
https://www.linkedin.com/in/vishwajeetbb |
https://krishiboot.com |
+91 8766543498 |
Vishwajeet Bhagyawant, Founder and CEO: B.Tech, IIT Bombay (Metallurgical Engineering, 2018). Building Agriyan full time since 2021 across four hardware generations; owns hardware architecture, the AI stack, field trials, fundraising and partnerships. Prof. Bakul Rao, Co-founder: IIT Bombay faculty, academic and research guidance. |
Vishwajeet is an IIT Bombay alumnus (Metallurgy, 2018) and started building Agriyan in 2021. Prof. Bakul Rao is IIT Bombay faculty and came on board as co-founder as the company took shape inside the IIT Bombay ecosystem; the two have worked together on Agriyan since 2024. Krishiboot is incubated at SINE, IIT Bombay, and the engineering team (R&D engineers Deep Naik and Gautam Mishra, AI/ML engineer Nandita N) plus accounts and compliance lead Pramila Tawade have joined since 2025. |
2 |
We build the whole machine ourselves, and we are honest about where it stands.
Five years and four hardware generations have produced an in-house stack end to end: a custom 48V implement controller board, a multispectral camera head at roughly Rs 10,000 bill of materials, and three on-board vision models that share one architecture and fit together on a single Jetson Orin NX. The people who design the models also run the field trials, on cotton, across three soil types, so every model is trained and tested on real Indian smallholder fields rather than lab data.
And we report what is not proven alongside what is. Every figure we quote has a test record, a method and a sample size behind it. |
Krishiboot Technologies Pvt. Ltd. (product: Agriyan) |
https://krishiboot.com |
Mumbai, Maharashtra (incubated at SINE, IIT Bombay) |
Agriyan is an AI-powered autonomous field robot that weeds, sprays and sows Indian smallholder farms plant by plant, sold per acre as a service. |
Indian smallholders spray whole fields on a fixed calendar because nothing affordable can see individual plants, so chemical lands where it is not needed, and weeding and spraying slip because labour is scarce on the day the crop needs it. |
Agriyan is a battery-electric autonomous ground robot, about 250 kg, built for 0.4 to 2 hectare Indian farms. On a single pass it drives the crop row on its own, identifies every crop stem and weed, grades plant health, and actuates the fitted tool at plant resolution instead of treating the whole field. Weeding, spraying and sowing modules share one platform and one compute stack.
Navigation needs no RTK, no LiDAR and no pre-mapping: depth is converted to surface normals to find vertical structures, so the crop rows themselves are the map. A navigation model produces row corridor, plant census, obstacle class and crop-safety keypoints in one 28 ms forward pass. Human detection is a hardcoded controller interrupt the model cannot override.
We do not sell the machine. We own and operate the fleet and bill the FPO per acre per operation, so the farmer pays for field work done rather than for equipment he cannot afford. Every pass adds to a per-plant health record of the farm.
Status, stated plainly: navigation and weeding run on the current platform; precision spraying was validated on our previous platform and the new spray module is designed, not yet fabricated; sowing is not yet built. |
Navigation without RTK, LiDAR or pre-mapping; a Rs 10,000 multispectral sensing head against USD 6,000 to 15,000 for commercial equivalents; three implements on one platform; and a per-plant dataset of Indian smallholder fields that only grows because we operate the fleet ourselves. |
MVP |
Pilots |
Farmer Producer Organisations in the Maharashtra cotton belt, contracting on behalf of member farmers on 0.4 to 2 hectares. Also institutional programmes that hold farmer budgets and a captive farmer base (Ambuja Foundation, Lupin Foundation, WWF-India, World Bank SMART Cotton). From 2029, Custom Hiring Centres buying through the SMAM subsidy channel. |
Rs 72,000 Cr per year: 67M acres of cotton, soybean, grapes and tomato in India, valued at what an acre already spends on labour for sowing, weeding and spraying |
Rs 21,000 Cr per year: 21M acres reachable with the current machine geometry |
Rs 50 Cr annual service revenue by 2031, about 370 machines serving 45,000 acres. Fleet size, not market size, is the binding constraint. |
Robotics-as-a-Service. We own and operate the fleet and bill the FPO per acre per operation (spray, weed, sow), priced inside what the farmer already pays in labour for the same work, and paid on each completed pass against a GPS-tagged record. Plant health monitoring is bundled, not charged. Machine sale opens from 2029 after FMTTI certification. |
Direct: Niqo Robotics (tractor-mounted precision spraying, India) and John Deere See and Spray. Indirect: drone spray operators such as Garuda Aerospace, conventional sprayer makers such as M.I.T.R.A. Agro, and the real incumbent, manual contract labour and local Custom Hiring Centres. |
Through institutions that already hold farmer budgets and a captive farmer base (CSR foundations, WWF-India, World Bank SMART Cotton), and directly through FPO boards within 250 km of our hub. One FPO agreement covers 300 to 2,000 member farms, so we sell to about 10,000 FPOs, not 146 million farmers. |
Cotton in Maharashtra first, the crop whose open rows fit the machine best. 2026: integrated field demonstration plus a same-day labour and chemical baseline. 2027: five machines, first FPO season agreement and first revenue, operated inside a 250 km radius so a breakdown costs hours, not days. 2028: 20 machines across multiple FPOs. 2029: FMTTI certification opens machine sales through the SMAM and CHC subsidy channel. |
Every smallholder field in India farmed plant by plant.
We want Agriyan fleets doing the field work for FPOs across India's cotton, soybean and horticulture belts, with every pass adding to a per-plant record of Indian farms that makes the next spray smaller and the next season more predictable.
The long-term company is three things: a field-operations network that sells work by the acre, a machine maker once the platform is certified, and the most detailed ground truth that exists on how Indian smallholder crops actually grow. |
Private limited company incorporated in India: Krishiboot Technologies Private Limited. DPIIT-recognised, GST-registered in Maharashtra, incubated at SINE, IIT Bombay. |
Rs 1.1 Cr equity from the IIT Bombay 1970 Batch TRA team, invested through SINE, IIT Bombay as CCPS in tranches (latest Rs 24.57 L, Aug 2026). Rs 50 L BIRAC BIG grant sanctioned, agreement pending, not yet disbursed. Now raising a Rs 2 Cr seed round. |
Yes |
Agriyan is AI first: on-board vision models decide every steering and treatment action. We want Avinya's AI mentors to pressure-test our model stack before we scale it, and IIT alumni and investor access to help close our seed round. The Rs 10 L would go into fabricating the precision spray module, the one gap between our validated AI stack and measured chemical savings in the field. |
Three things by Demo Day: the spray module fabricated and validated against a same-day blanket-spray control plot, giving our first measured chemical reduction on the current machine; a concurrent-load benchmark with all AI models running together on one Jetson; and at least one signed FPO letter of intent for kharif 2027, alongside progress on our seed round. |
Yes. We are based in Mumbai, so we would join regular sessions virtually and travel to Bengaluru for Demo Day, reviews and any mandatory in-person sessions. |
Yes |
Core engine |
Everything that steers or treats runs on the robot, on one NVIDIA Jetson Orin NX 16GB, fully offline.
Sensing (KrishiSense): Intel RealSense D435i for RGB, depth and IMU, plus two Sony IMX477 sensors with 720 nm red-edge and 850 nm NIR bandpass filters, hardware-triggered from the RealSense timestamp to 3 to 5 ms sync error.
On-board models:
1. CropNav (navigation, Jetson GPU, 28 ms): one forward pass outputs row corridor mask, plant census, obstacle class and distance, and crop-safety keypoints. It replaces four separate networks, so every steering decision uses four simultaneously current outputs.
2. WeedNet-3D (weed identification and crop avoidance, Jetson DLA, about 50 ms): weed species box and class plus a cotton stem keypoint heatmap for 3D stem localisation.
3. MSCD-Net (per-plant health, Jetson DLA, 60+ FPS): plant box, 11-class disease label, stem keypoint, using the multispectral channels.
All three are one architecture trained three times: a five-stage depthwise separable backbone, depth-derived attention, distance-adaptive convolution, multi-scale pooling, a BiFPN neck and an anchor-free head, trained with a geometric consistency loss. That reuse is why three networks plus spectral computation fit on one edge device.
Off-board: an EfficientNet-B0 leaf-diagnosis model (8 classes with GradCAM++ explanations) serves farmer phone photos from AWS in under a second.
Runtime: ROS 2, offline-first SQLite on NVMe mirroring a cloud PostgreSQL store, synced after each pass via a queue, because about 750 MB per hectare cannot stream over rural 4G. |
Yes, and it is the part no competitor can buy.
We own roughly 5,200 real field frames of Indian smallholder conditions: about 2,000 navigation frames from cotton fields in Nashik, Amravati and Yavatmal, 1,200 cotton weed images across 12 species, and 2,000 leaf images, all labelled in-house with growth-stage stratified labels and weed labels scored against two agronomists (Cohen's kappa 0.86). Alongside that sit about 4,700 simulated frames from Gazebo and a CycleGAN pipeline for cold-starting new crops.
There is no comparable public dataset for Indian smallholder fields at 0.16 hectare plot scale. Because we operate the fleet rather than sell it, every pass adds GPS-tagged, per-plant data to this record, so the advantage compounds with every acre served. |
Four things, in order of how hard they are to copy.
1. The data. Per-plant field data from Indian smallholder farms is collected only by operating in those fields, season after season. Our service model is designed to keep generating it.
2. Geometry-driven perception. Converting depth to surface normals makes detection depend on plant structure rather than colour, which is why one stack works on black Vertisol, red laterite and sandy loam without retraining. Colour-based systems have to be retrained per soil and light condition.
3. Hardware-model co-design. A Rs 10,000 multispectral head, a shared model architecture that fits three networks on one Jetson, and a single-pass navigation network are designed together. Copying the models without the sensor and compute design does not reproduce the cost.
4. IP. Two patent applications in progress with the Indian Patent Office: one on the reconfigurable multi-implement platform, one on the single-pass quad-task navigation network, path lookahead and controller. Neither is granted yet. |
Every figure below has a test record, a method and a sample size. We report navigation in centimetres and detection against human experts.
Navigation: 2.8 cm mean lateral drift against a chalk line (6.1 cm max) over 12 consecutive rows with no operator takeover; 4.2 cm mean on irregular row spacing; 5.8 cm mean on a 15 degree slope.
Stem localisation: 12 mm mean 3D Euclidean error (1 sigma).
Weed identification: per-species precision 0.76 to 0.94, scored against two agronomists at Cohen's kappa 0.86.
Disease classification: mAP 0.83 at IoU 0.5 on board; 89.2% top-1 on the leaf-scan model.
Latency: 28 ms navigation pass, about 50 ms weed model, 60+ FPS health model.
Safety: human detection triggers at 1.5 m and stops in about 0.31 m, with a deliberately low 0.65 threshold.
Reliability in the field is judged by operator takeovers per row and by performance across soil types, not just test-set accuracy.
What we have not yet measured, stated plainly: each latency was measured with that model running alone, and concurrent load with all models together has not been benchmarked; weed removal rate has not yet been scored against a formal protocol. |
The robot captures images of crops and soil, not people, and processes them on the device, fully offline. Data leaves the robot only after a pass, through a queued sync to our cloud store.
We treat farm location and farmer identity as personal data under the Digital Personal Data Protection Act 2023. Data is collected under the FPO service agreement, used to deliver the service and improve our models, and returned to the FPO as a per-farm season report.
On safety, which matters more than data for a field robot: human detection is a hardcoded controller interrupt that no model output can override.
We are a pre-commercial company and will formalise encryption, access control and data-retention policies before the first commercial season in 2027. |
Yes, four, and we plan around each.
1. FMTTI certification is the largest. Selling farm machinery into the SMAM or Custom Hiring Centre subsidy channel needs testing at a Farm Machinery Training and Testing Institute, and no BIS test code exists yet for an autonomous electric ground robot. It does not block our service model, where we own and operate the machines, but it gates machine sales from 2029, so we are starting the process early.
2. Agrochemical handling under the Insecticides Act 1968 and CIB&RC norms: the spray module must meet equipment norms and our operators must follow handling and protective-equipment rules.
3. Autonomous machinery safety: there is no settled Indian standard, so we built human detection as a hard stop and will publish our own safety documentation.
4. Data protection under the DPDP Act 2023 for farm and farmer data. |
Five things, honestly ranked.
1. Edge compute headroom. Each model's latency was measured alone; with all models plus spectral computation running together on one Jetson, we expect contention we have not yet benchmarked.
2. Data pipeline. About 750 MB per hectare per pass cannot stream over rural 4G. At 10x acres the post-pass sync, storage and labelling queue become the bottleneck.
3. Labelling and QA. Expert labels scored against two agronomists do not scale 10x without active learning and a tiered review process.
4. Distribution shift. New crops and row geometries (soybean at 45 cm spacing, for example) and new regions will degrade models trained mostly on Maharashtra cotton.
5. Fleet model management. Versioning, rollback and monitoring of models across tens of machines in the field, with limited connectivity.
These are exactly what we want to work on with Avinya's mentors before we scale, not after. |
Yes. Nandita N is our AI/ML engineer, working on the vision models and training pipeline. Vishwajeet designed the model architecture, the single-pass navigation network and the multispectral sensing head, and still runs the field trials that generate the data. Deep Naik and Gautam Mishra, R&D engineers, cover embedded systems and robot integration, which is where models meet actuators. Prof. P. Udmale of IIT Bombay mentors the agronomy interface. |
Mostly our own. About 5,200 real field frames (navigation, cotton weeds across 12 species, leaves) were collected by us on farms in Maharashtra and labelled in-house, so we own them outright. We also use about 4,700 simulated Gazebo frames we generated ourselves.
Two external inputs, disclosed plainly: ImageNet-pretrained weights for backbone initialisation, and the public PlantVillage dataset as one domain in our CycleGAN cold-start pipeline for leaf images. We are confirming the licence terms for both before commercial deployment, and our real-field data is steadily replacing the PlantVillage share. |
The fleet is the data engine. Every pass logs GPS-tagged, time-synced RGB, depth and multispectral frames with the model's decisions attached.
1. Hard-case mining: frames where confidence is low, where an operator intervened, or where models disagree are queued first for labelling.
2. Expert labelling with agreement checks against agronomists, keeping inter-rater agreement measured, not assumed.
3. Retraining on a seasonal cycle, with growth-stage stratified test sets so a model cannot improve on one stage while regressing on another.
4. Field validation before release: every new model is re-run against the same centimetre-level navigation and precision tests before it goes on a machine.
5. New crops start from the CycleGAN cold-start pipeline and are then replaced with real field data as soon as the fleet operates there.
One gap we are closing now: validating the plant-health model on real field photos rather than simulation. |
India first. The machine is sized for 0.4 to 2 hectare plots and 0.16 hectare plot-scale data, which is where Indian agriculture actually is, and where Western autonomous machines do not fit on price or geometry. Our first market is cotton in Maharashtra, then other Indian cotton and soybean belts.
Later, the same smallholder economics apply across South and Southeast Asia and parts of Africa, but we will not look outside India until the service model is proven here. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/6S0IRJyLlGdK-Krishiboot-Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/6S0IRJyLlGdK-Krishiboot-Pitch-Deck.pdf |
https://drive.google.com/file/d/173o19Jp6b8RHoEjF8G92Yc3xiCTMp82V/view |
Yes. Agriyan exists to cut agrochemical use at source and take the spray tank off the farm worker's back.
Cotton covers roughly 5% of India's cropped land but has historically accounted for close to half of its pesticide use, nearly all of it applied by blanket spraying. Chemical that misses its target runs off into water, drifts, and is inhaled by the person carrying the tank. Treating the plant instead of the field means that chemical is never applied.
The impact we will measure, each against a same-day farmer-practice control plot: kilograms of pesticide active ingredient avoided per acre, spray water saved, diesel and CO2 avoided by a battery-electric machine replacing tractor or engine sprayers, herbicide displaced by mechanical weeding, and hours of human chemical exposure eliminated.
For the farmer, our fee sits inside what he already pays in labour, so his gain is the chemical bill: about Rs 5,400 per acre on our modelled cotton case, which we are now measuring in the field. The category benchmark is about 50 to 60% chemical reduction (John Deere See and Spray, Niqo Robotics); we will quote our own number only once we have measured it on this machine. |
NA |
SINE, IIT Bombay |
checked |
| Sep 25, 2026 @ 2:28 PM |
Dheerendra Kumar , Ayush Mandhana , Krishna kant Sharma |
developer.dheerendra@gmail.com |
https://www.linkedin.com/in/dheerendra-dev/ |
https://creonnect.com |
08840092746 |
Dheerendra Kumar — CEO Senior Software Engineer at Peoplebox (YC S22) with 5+ years of experience building B2B products. Leads Creonnect’s overall product vision, business strategy, fundraising, and execution. Ayush — CTO Software Engineer at Intangles Labs, with 4 years of experience building scalable products. Leads the technology and engineering architecture at Creonnect. Krishna Sharma — COO Brings 7+ years of experience in B2B SaaS and leads operations and execution. |
Ayush and I worked together at my previous company, where we built products together and developed a strong understanding of each other’s working style and technical strengths. We have known and worked with each other for several years.
Krishna is a childhood friend, so we have known each other for many years. We have a long-standing personal relationship and understand each other’s strengths, which has helped us work effectively together on the company.
The rest of the team has come together through our professional network , College friends and shared experience in technology, product, AI, and B2B SaaS. |
1 |
Our biggest strength is that the entire team genuinely believes in the vision and is committed to building it together. Everyone on the team has a full-time role, but we still spend our evenings and time after work building QuantumRealm because we believe in what we are creating. We have a strong culture of mutual trust and respect—everyone’s ideas and decisions are valued, and we give each other ownership. Our different strengths across engineering, AI, product, operations, and growth allow us to move quickly while staying aligned on one vision. |
QuantumRealm AI Labs Private Limited |
https://creonnect.com/ |
Bengaluru |
AI-powered infrastructure for the creator economy that automates brand–creator discovery, outreach, and campaign execution. |
Brands still spend a lot of time manually finding the right creators, reaching out, following up, managing campaigns, and tracking results across multiple tools or agencies. This makes creator marketing difficult and expensive to scale. At the same time, creators struggle to discover relevant brand opportunities and manage collaborations. **Creonnect solves this by using AI agents to automate the workflow from creator discovery and matching to outreach, campaign execution, and performance tracking. |
Creonnect is an AI-powered creator marketing platform that automates the workflow from a brand’s campaign brief to creator discovery, matching, outreach, campaign management, and performance tracking. AI agents shortlist relevant creators, manage outreach and follow-ups, and help track campaign execution, while creators can discover brand opportunities and manage collaborations. Our goal is to make creator marketing faster, scalable, and measurable for brands while creating more opportunities for creators. |
Our solution is unique because we are building an AI-agent layer for creator marketing, rather than just another creator marketplace or discovery tool. AI agents automate the workflow from campaign brief to creator matching, outreach, follow-ups, campaign tracking, and performance analysis. Over time, the system can learn from creator participation and campaign outcomes, improving matching and execution. Our growing creator network, campaign outcome data, and agent workflows can create a compounding advantage as more campaigns run through the platform. |
Users |
Users, Pilots, Signups |
Our target customers are both brands and creators. Brands include D2C brands, startups, and growing businesses that want to find the right creators, run campaigns, automate outreach, and measure results. Creators include micro, mid-sized, and emerging creators who are looking for relevant brand collaborations and better ways to manage their partnerships and monetization opportunities. We want to build a two-sided platform where brands can run creator campaigns more efficiently and creators can discover more relevant opportunities. |
$33B |
$10B |
$50M |
Our revenue model is primarily brand-led. Creators are free during the beta phase to build supply and encourage adoption. * Brand subscriptions: Monthly plans for brands based on their campaign and usage needs. * Campaign commission: 2.5% commission on campaign spend processed through Creonnect. * AI credits: Pay-as-you-go AI credit packs for additional AI-powered workflows and usage. * Creator subscriptions: After the beta phase, we plan to introduce optional creator plans at ₹149, ₹399, and ₹999 per month. Our initial revenue focus is entirely on brands through subscriptions, campaign commissions, and AI credits. |
Qoruz KlugKlug HashFame CREDO/M Traditional influencer marketing agencies Creonnect differentiates itself by focusing on an AI-agent-driven workflow that goes beyond creator discovery—covering creator matching, outreach, follow-ups, campaign execution, deliverable tracking, and performance analysis in one platform. |
Brand acquisition: Direct outreach to D2C brands and marketing teams, supported by founder-led sales and industry networking. Creator acquisition: Targeted social campaigns, creator communities, referrals, and partnerships to build a strong creator network. Pilot campaigns: Work closely with early brands to run campaigns, demonstrate results, and convert successful pilots into recurring customers. Partnerships: Build relationships with agencies, marketing partners, and ecosystem organizations that can introduce brands and creators to Creonnect. Performance marketing: As we validate CAC and conversion, invest in targeted digital advertising and content-led acquisition. Referral and network effects: Successful campaigns and creator participation can generate referrals and strengthen the brand-creator network. |
Brand acquisition: Use founder-led sales and direct outreach to D2C brands and marketing teams to onboard early customers. Creator acquisition: Build the creator network through social media, creator communities, referrals, partnerships, and targeted campaigns. Pilot campaigns: Run initial campaigns with selected brands, demonstrate measurable results, and convert successful pilots into recurring customers. Partnerships: Partner with influencer agencies, marketing partners, and ecosystem organizations to expand distribution. Performance marketing: Once CAC and conversion rates are validated, scale through targeted digital advertising and content marketing. Network effects: As more brands and creators join and complete successful campaigns, referrals and campaign data will strengthen the platform and support organic growth. |
Our long-term vision is to make Creonnect the most trusted workspace for the creator economy—a single platform where brands and creators can discover each other, build partnerships, manage collaborations, and grow together.
We want Creonnect to become the trusted brand behind creator-brand relationships globally, powered by AI that makes discovery, communication, campaign execution, payments, and performance measurement simpler and more transparent. |
QuantumRealm AI Labs Private Limited is an incorporated private limited company in India, operating Creonnect as its product. |
2500000 |
Yes |
We are applying to Avinya because we believe Creonnect can become the most trusted workspace for the creator economy. We have built the initial creator side and are now focused on scaling brand adoption and our AI-powered automation layer. Avinya can help us with the right ecosystem, mentorship, industry connections, and access to customers and partners as we move from early product validation to larger-scale adoption. We see this as an opportunity to accelerate our growth while learning from experienced founders and the broader startup community. |
During the programme, we want to move Creonnect from early product validation to stronger market adoption. Our key goals are to onboard more brands and creators, validate repeatable customer acquisition, launch and improve our AI-powered automation layer, and run successful pilot campaigns with measurable outcomes. We also want to learn from experienced mentors, build strategic partnerships, and strengthen our go-to-market strategy. By the end of the programme, we aim to have stronger product-market fit, measurable traction, and a clear roadmap for scaling Creonnect globally as the most trusted workspace for the creator economy. |
Virtual |
Yes |
Core engine |
Creonnect uses a modular AI-agent architecture built around LLMs, structured creator data, and workflow automation. The system collects creator and campaign data, processes it through AI agents, and uses the outputs to automate key workflows.
The architecture includes:
Data Layer: Stores creator profiles, audience data, campaign requirements, engagement metrics, and campaign outcomes.
AI/LLM Layer: Uses LLMs to understand campaign briefs, creator profiles, and communication context.
Matching Engine: Matches creators with campaigns based on category, audience, campaign goals, budget, and other requirements.
Agent Layer: AI agents handle creator discovery, outreach, follow-ups, communication, deliverable tracking, and campaign analysis.
Workflow Layer: Connects agents with the campaign lifecycle while keeping human approval for important actions.
Learning Layer: Uses campaign outcomes, responses, engagement, and performance data to continuously improve matching and agent workflows.
The goal is to create an AI-powered system that can progressively automate the end-to-end brand–creator collaboration workflow. |
Our proprietary data advantage is expected to come from the structured creator and campaign data generated through Creonnect. As more creators and brands use the platform, we collect data around creator profiles, campaign requirements, outreach responses, collaborations, deliverables, engagement, and campaign outcomes.
This data can help our AI improve creator–brand matching, outreach, campaign execution, and performance predictions over time. The advantage compounds as more campaigns run through Creonnect, creating a feedback loop between real campaign outcomes and our AI workflows.
We are still at an early stage, so we view this as a developing data advantage rather than an established moat. |
Our AI component becomes defensible through the combination of proprietary workflow data, specialized agent workflows, and continuous learning from real campaign outcomes.
As more campaigns run through Creonnect, we build structured data on creator–brand matching, outreach responses, engagement, deliverables, and campaign performance. This feedback loop can improve our matching and automation models over time.
Our defensibility is therefore not just the underlying LLM, which competitors can access, but the domain-specific data, workflows, agent orchestration, and campaign feedback accumulated within Creonnect. |
We evaluate model performance using both AI-specific and business-level metrics. Key metrics include:
Matching accuracy: relevance of recommended creators to campaign requirements.
Outreach performance: response rate, positive response rate, and conversion to collaboration.
Agent reliability: task completion rate, workflow success rate, and error rate.
Deliverable accuracy: accuracy of campaign tracking and status updates.
Performance analysis: accuracy of AI-generated campaign insights compared with actual results.
User feedback: brand and creator feedback on AI recommendations and workflows.
Business impact: campaign conversion, time saved, CAC, retention, and campaign outcomes.
We also use human review for important AI decisions and continuously evaluate results using real campaign data. |
We follow a privacy-by-design approach, with access to data limited to what is required for the platform’s functionality. Creator and brand data is stored securely, with role-based access controls and appropriate authentication and authorization.
We minimize unnecessary collection of personal data and aim to use data only for clearly defined product and business purposes. For AI workflows, we focus on structured and relevant data rather than exposing unnecessary personal information to models.
As we scale, we will continue strengthening security controls, data retention policies, consent mechanisms, audit logging, and compliance with applicable data-protection requirements. |
Yes. The main regulatory risks for Creonnect are related to data privacy, influencer advertising disclosures, and platform/API policies.
As a creator marketing platform, we need to ensure campaigns follow applicable advertising and consumer-protection requirements, including clear disclosure of paid or other material relationships between brands and creators.
On the AI and data side, we need to comply with applicable data-protection requirements and ensure appropriate consent, access controls, data minimization, and security practices.
We plan to build compliance into the product through campaign disclosure workflows, audit logs, permission controls, and clear data-handling policies, while monitoring changes in regulations as we expand into new markets. |
At 10x scale, the main challenges would be AI inference costs, latency, data quality, and reliability of agent workflows. More creators and campaigns would also increase the volume of matching, outreach, and campaign events that the system needs to process.
We are designing the architecture to handle this through modular agents, asynchronous workflows, caching, structured data pipelines, and scalable cloud infrastructure. We would also monitor model accuracy, latency, cost per workflow, failure rates, and human intervention rates.
The biggest risk is maintaining consistent AI performance and cost efficiency as the volume of campaigns and interactions grows. |
Yes. We have an in-house AI/ML and engineering team with strong startup and B2B product experience.
Our Head of AI, Sharath, leads our AI initiatives, with expertise in LLMs, Generative AI, AI agents, RAG, and prompt engineering. He is focused on building the AI layer for creator discovery, creator–brand matching, outreach, campaign management, deliverable tracking, and performance analysis.
Our engineering team brings 5+ years of experience building and scaling B2B products and SaaS systems across startups. Krishna, our COO and a second-time founder, brings 7+ years of engineering and startup experience, with strong understanding of building products, teams, and business operations.
The broader team has experience across Python, TypeScript, Node.js, cloud infrastructure, scalable B2B systems, and AI-enabled product development. This allows us to build, test, deploy, and iterate on our AI systems entirely in-house. |
We currently use a combination of first-party and publicly available data. Our creator data is primarily collected through creators who join and use Creonnect, making the core profile and campaign interaction data first-party and owned by us, subject to applicable consent and privacy requirements.
We also use campaign and engagement data generated through our platform, including creator profiles, campaign requirements, outreach responses, collaboration activity, deliverables, and campaign outcomes. This data becomes increasingly valuable as more campaigns run through Creonnect.
For external information, we use publicly available data and third-party APIs/services where permitted under their respective terms. We do not rely on unlicensed proprietary datasets for our core AI workflows. |
We plan to build self-improvement into our AI agents through continuous feedback from real campaign outcomes. The system will learn from creator responses, brand feedback, successful and unsuccessful matches, outreach performance, engagement, and campaign results.
These signals will be used to identify patterns, improve prompts and agent workflows, refine creator–brand matching, and evaluate model performance. We will maintain human oversight for critical decisions while allowing the system to continuously improve through data-driven evaluation and feedback loops. |
We are building for both India and global markets. India is our initial market, where we are focused on D2C brands, startups, and creators to validate the product, build our creator network, and establish repeatable customer acquisition.
Once we validate the model in India, we plan to expand into other major creator markets, including Southeast Asia, the US, and other international markets. Our AI-powered architecture is designed to support different markets, languages, creator ecosystems, and campaign workflows as we scale.
Our long-term vision is to make Creonnect the most trusted workspace for the global creator economy. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Zs0s9O7dBgBZ-Creonnect_Investor_Pack_Final.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Zs0s9O7dBgBZ-Creonnect_Investor_Pack_Final.pdf |
http://N/A |
Yes. Creonnect is mission-driven around making the creator economy more accessible, transparent, and sustainable for both creators and businesses.
We want to help creators, including emerging creators from smaller cities, discover relevant brand opportunities and build more sustainable income through collaborations. At the same time, we want to make creator marketing accessible to smaller and growing brands by reducing the cost and manual effort required to discover creators and run campaigns.
Our long-term mission is to build Creonnect as the most trusted workspace for the creator economy, where technology and AI reduce barriers, improve transparency, and create more opportunities for creators and brands to work together. |
No. I do not identify as part of an underrepresented group in tech/startups. |
None |
checked |
| Sep 24, 2026 @ 3:02 PM |
Somasundaram N (Founder & CEO), Arun Somasundaram (Co-Founder, Product & AI) |
arun@finixel.com |
https://www.linkedin.com/in/arun-somasundaram-83b54b147/ |
http://www.finixel.com |
09886789533 |
Somasundaram N — Founder & CEO. IIT alumnus; 40+ years in automotive sourcing and operations across [GM, Mitsubishi, Eicher. Owns domain, buyers and commercials. Arun Somasundaram — Co-founder, Product & AI. Technical University of Munich, Applied ML. Owns the product, the data pipeline and the scoring engine. |
Father and son, working together on this business since April 2025. |
All |
We are not learning this industry during the program. Soma has spent a career inside Automotive sourcing with an operator network in OEMs and Tier 1s. This company has run a B2B supplier platform before and suppliers paid a real enrolment fee to join it. Arun builds the product and the AI. Domain access is the half a competitor cannot hire quickly. |
Finixel Technologies Private Limited |
www.finixel.com |
Bangalore |
AI Native Customer Discovery Platform |
A capable supplier's hardest problem is not making the part — it is finding the buyer who is sourcing it right now and knowing what would disqualify them before they spend months chasing it. Every platform in this market was built so buyers can find suppliers. The supplier is left listing themselves and waiting. |
The supplier pastes a website or uploads documents. We build a structured plant profile from public record in seconds and they confirm it. That plant is then scored against live buyer programs on seven weighted dimensions, showing which single gap puts them under a specific buyer's threshold — and we hand them a dated action plan to close it. When they clear it, we run the outreach. One enrolment, re-matched continuously as new sourcing needs open. |
1. Our supplier-focused AI solution works on distilled expertise from domain experts. 2. Our scoring system is evidence-based and auditable, while still having the speed of AI. 3. Our data curation and supplier qualification process. |
MVP |
Pilots |
Mid-size Suppliers in Electric Vehicle, Automotive and Engineering Industries, already exporting or wanting to do so. |
500 B USD |
12.5 B USD |
180 M USD |
Supplier enrolment — a one-time fee per plant, Success commission on closed business, negotiated per deal. Buyer-side retainers from OEMs and Tier 1s who want a qualified shortlist against a specific program. |
Static Directories and Deal agents (IndiaMART, ThomasNet), Managed manufacturing networks — Zetwerk, Moglix, Xometry Both were built so buyers can find suppliers. In Finixel's case the supplier is our customer. |
Professional contacts, networking and participating in Industry associations |
Professional networking |
To be the lead player in enabling business for aspiring suppliers in the Engineering vertical. |
Private Limited |
Bootstrapped |
Yes |
1. Soma is an IIT Madras Alumni and there is great reputation of Avinya to scale businesses. 2. Access to market. 3. AI Mentorship and peer-networking |
1. Obtain live sourcing needs from real buyers on the platform, utilizing the cohort and alumni network. 2. Indian suppliers enrolled and matched against those needs, converting the warm list from our earlier platform. 3. One closed, commission-eligible deal, to prove our business model. 4. Investor-ready data room and first conversations with Indian VCs. 5. We would also like the AI architecture reviewed by someone who has scaled one. |
Yes |
Yes |
Core engine |
Agentic automations (Cursor, Claude, etc.). NoSQL database. Node.js app server. |
The curation process is our current advantage, with our expert team |
The distilling of our team of experts into agentic models. |
We rely on human expert feedback, as well as customer feedback. The metrics used for now are primarily accuracy-based. |
We will follow all local and global laws regarding privacy, compliance, and security, including Agentic automations to find and resolve any exploits found in any packages we use. |
No. We will be compliant to all laws. |
1.Data concurrence 2. Token usage costs per user, and in running backend applications |
Yes. Arun has expertise in Applied ML and is using his AI expertise in building this product. |
No datasets. Our database at the moment is primarily from curating publicly available data, as well as user provided data. |
We will distill all usage metrics and expensive model insights, as well as user and expert feedback continuously back into our models. |
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/RYuILZcysHzY-Finixel_Avinya_Pitch_Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/RYuILZcysHzY-Finixel_Avinya_Pitch_Deck.pdf |
https://www.youtube.com/watch?v=WM3PKfQWFMw |
Yes. India's certified supplier base is among the largest in the world, but export orders go to the firms with sales reach, not the best plants. Finixel gives any qualified MSME the buyer access that today only large suppliers have. We will report suppliers qualified, introductions made, orders won and export value added, by district, starting in Karnataka. |
NA |
Ramaswamy Balasubramanian, Business Mentor, IIT Madras DST GDC Incubate Program |
checked |
| Sep 23, 2026 @ 1:39 PM |
T. Sreekanth (Co-Founder & CEO), Sambasiva Rao Suda (Co-Founder & CTO). |
contact@optficial.ai |
https://in.linkedin.com/company/optficiallabs |
https://optficial.ai/ |
8328653091 |
T. Sreekanth (Co-Founder & CEO): Former IAS officer with extensive leadership in public administration, large-scale rural governance, institutional distribution, and public policy execution. Sambasiva Rao Suda (Co-Founder & CTO): Enterprise software and AI systems architect specializing in scalable microservices, low-latency machine learning pipelines, and distributed data systems. Dr. Asadi Srinivasulu (Chief Advisor): Senior AI researcher, academic leader, and data science advisor specializing in applied machine learning and ethical AI governance. |
Our founding team and advisory board came together through shared work across public-sector technology implementations, enterprise systems engineering, and applied academic AI research. We have collaborated for over 2 years aligning deep technical architecture with mission-critical institutional workflows. |
All |
The intersection of ground-level public policy execution and scalable enterprise AI engineering. Having leadership with a former IAS background gives us unparalleled institutional access, trust, and distribution insight into regulated sectors, while our technical bench brings deep systems architecture and research rigor. |
Optficial Labs Pvt Ltd |
https://optficial.ai/ |
Hyderabad, Telangana (Offices in Gachibowli), India |
Optficial Labs builds purpose-built vertical AI systems for frontline workflows in agriculture (BhoomiAI), healthcare (Mendly), and education (dhimantAI) that strengthen human judgment. |
Generic AI models act as superficial wrappers, hallucinate, and fail in high-stakes, context-heavy domains where trust, accuracy, and local compliance matter. In agriculture, smallholders lack actionable, vernacular guidance; in healthcare, doctors burn out from manual EHR data entry and fragmented clinical coordination; in education, AI encourages rote answering instead of conceptual reasoning. |
We build domain-specific vertical AI engines embedded directly into real-world operational workflows:
BhoomiAI (Agriculture): Localized farm intelligence delivering micro-climate risk alerts, image-based crop disease diagnosis, and vernacular voice guidance for smallholder farmers.
Mendly (Healthcare): An ambient clinical operations copilot converting multilingual doctor-patient consultations into structured SOAP notes and coordinating care across clinic staff.
dhimantAI (Education): A patient Socratic AI tutor that prompts students step-by-step to think through problems rather than dumping answers, with teacher gap-diagnostic analytics. |
We do not deploy generic foundation model wrappers. Our defensibility stems from: Deep workflow embedding (solving real operational bottlenecks, not just text generation). Ground-truth domain data across Indian regional languages and rural dialects. "Human-in-the-loop" trust design that assists rather than replaces human experts. |
MVP |
Users, Revenue, Pilots |
BhoomiAI: Farmer Producer Organizations (FPOs), agritech enterprises, rural credit/crop insurance institutions, and state agricultural departments. Mendly: Tier-2/3 primary care clinics, day-care nursing homes, diagnostic networks, and independent practitioners. dhimantAI: K-12 private and state educational institutions, learning centers, and school networks. |
$18B+ combined Indian market across agritech advisory, healthcare IT/clinical transcription, and AI-enabled education. |
$2.8B addressing Tier-2/3 clinics, smallholder FPO ecosystems, and regional private/budget school chains. |
$45M serviceable obtainable market over 3 years targeting high-density operational hubs in South India (Telangana, Andhra Pradesh, Karnataka). |
B2B SaaS monthly/annual subscription per clinic seat or school license. Volume-based consumption tiers (per-consultation ambient transcription or per-acre farm advisory). Institutional pilot and enterprise deployment contracts. |
BhoomiAI (vs. CropIn, Plantix, Fasal): Unlike hardware-heavy IoT setups or commerce-first apps, we provide a pure-software vernacular voice advisory that eliminates literacy and cost barriers for smallholders. Mendly (vs. Nuance DAX, Abridge, HealthPlix): Global scribes are cost-prohibitive and fail on Indian linguistic code-switching, while legacy EHRs require tedious typing. We offer affordable, Indic-fluent ambient transcription linked to complete 7-role clinic workflow coordination. dhimantAI (vs. Khanmigo, Doubtnut, ChatGPT): Generic EdTech dumps direct answers. We use guided Socratic questioning to enforce active reasoning, paired with institutional learning-gap analytics for teachers. |
Direct B2B institutional sales, public-private partnerships, FPO and cooperative network distribution, and clinical association tie-ups, leveraged by our founding team's deep public and enterprise networks. |
Our Go-To-Market (GTM) strategy follows our core operational framework: Listen → Prove → Grow. Because Optficial Labs builds for high-trust, regulated sectors (Agriculture, Healthcare, and Education), we do not rely on generic digital advertising or low-retention consumer acquisition. Instead, our GTM is anchored on B2B2C institutional distribution, where we partner with trusted aggregators who already hold the relationship with the end-users. |
To become India's foundational vertical AI studio, powering mission-critical human services with trustworthy, vernacular, and context-aware intelligence that elevates human capability. |
CIN: U62090KA2023PTC171421 |
We are currently 100% bootstrapped and self-funded by our founders. To date, we have not taken external institutional venture funding. We have funded our initial product development, AI architecture, and field pilots through internal capital, and we are now applying to raise our first external pre-seed / accelerator check to scale operations. |
Yes |
We are building specialized AI tools for three areas that genuinely matter: farming, healthcare, and education. General AI tools like ChatGPT are too broad and often guess answers when they shouldn’t. That doesn't work when a farmer’s livelihood or a patient’s health is on the line. We are applying to Avinya to take what we’ve built out of the lab and into the real world. Having IIT mentors will help us fine-tune our voice tools so they work smoothly in regional Indian languages and on patchy village internet. Partnering with the Karnataka ecosystem gives us the ground-level access we need to test these tools directly in clinics and farming communities. |
Over the 4 months, our goal is to take our vertical AI systems out of the sandbox and turn them into battle-tested production tools: Optimize our Indic speech stack: Work with IITACB mentors to quantize our ASR and NLP pipelines for regional dialects (Kannada, Telugu, Hindi), targeting sub-2-second latency and offline audio caching on low-end Android hardware. Validate in high-friction environments: Run live field pilots for BhoomiAI with Karnataka FPOs to prove our pest/weather alerts reduce input costs, and deploy Mendly in busy local clinics to ensure ambient SOAP note extraction requires minimal clinician correction. Harden infra for Demo Day: Use the ₹10L seed check for dedicated inference compute and pilot telemetry, entering Demo Day with verified Word Error Rates (WER), high daily retention, and proven workflow fit. |
Yes, fully open to a hybrid model. |
Yes |
Core engine |
Our AI architecture uses a decoupled, 4-tier pipeline designed for low latency, regional dialects, and zero-hallucination tolerance:
Perception & Edge Ingestion: Client-side noise filtering and audio chunking feed fine-tuned Indic ASR models (handling regional dialects and code-switching in Kannada, Telugu, Hindi) alongside lightweight convolutional vision models for mobile crop disease detection. Audio buffers locally for spotty rural networks.
Context & Hybrid RAG: We extract domain entities (symptoms, crops, soil metrics) and run hybrid retrieval (dense vector embeddings + BM25 keyword matching) strictly against verified databases (ICAR agronomy, ICD-10/SNOMED clinical ontologies, curriculum graphs).
Task-Specific Orchestration: We route requests to fine-tuned, self-hosted Small Language Models (SLMs) on dedicated inference engines (vLLM) to keep inference under 2 seconds and control compute costs. State machines enforce Socratic pedagogical pacing in dhimantAI and structured SOAP parsing in Mendly.
Deterministic Guardrails & HITL: All model outputs pass strict JSON schema validation before presentation. The system generates draft notes or recommendations—the clinician, teacher, or farmer always retains final sign-off authority. |
Acoustic Edge in Indic Dialects: Field audio capturing rural dialects and clinical code-switching (Kannada, Telugu, Hindi-English) that standard models fail on.
Structured Operational Ontologies: Proprietary decision graphs mapping hyper-local crop disease cycles, 7-role clinic coordination triggers, and student prerequisite gap trees.
HITL Correction Flywheel: Continuous fine-tuning data from real-world clinician note edits, agronomist validations, and teacher feedback loops. |
Our defensibility comes down to three structural engineering choices that prevent generic foundation models or thin API wrappers from easily replicating our stack:
Dialect-Level Acoustic & Code-Switching Pipelines:
Off-the-shelf models fail when a doctor alternates between English clinical terms and Telugu symptoms, or when a farmer speaks in an unstandardized rural dialect. We own specialized phonetic pre-processors, localized language vocabularies, and fine-tuned ASR weights that run with high accuracy where public APIs return gibberish.
Domain-Restricted Hybrid RAG on Verified Ontologies:
We do not rely on standard open-web vector dumps. Our retrieval engines are hard-coupled to verified domain ontologies (ICD-10/SNOMED-CT clinical taxonomies, ICAR research packages, state board learning graphs). By combining dense embeddings with sparse keyword validation (BM25) and strict JSON schema checks, our models operate without the hallucinations common to general-purpose competitors.
Closed HITL Operational Data Flywheel:
Every correction a doctor makes on a generated SOAP note or an agronomist confirms on a pest alert feeds back into our evaluation pipeline. Competitors without our on-the-ground pilot touchpoints cannot access this granular, high-conviction ground truth to fine-tune task-specific Small Language Models (SLMs).
Workflow Lock-In Over Chat Interfaces:
We don't sell text boxes; we sell embedded operational glue. In Mendly, transcribing ambient speech automatically triggers downstream actions across triage, nursing, and pharmacy workflows. In BhoomiAI, advice links directly to FPO member management and aggregated input logistics. Replicating the AI prompt doesn't replicate the multi-stakeholder operational integration. |
We evaluate performance across three operational layers:
Perception: Low Word Error Rate (WER) on dialect-dense Indic audio, sub-2s streaming latency, and high F1-score on mobile crop vision models.
Groundedness & RAG: Context precision, zero-tolerance schema validation, and <0.1% hallucination rate against verified clinical and agronomic datasets.
Field Reality (HITL): Clinician Correction Rate (<8% manual edits on ambient SOAP notes), agronomist advisory acceptance, and session completion on spotty rural connectivity. |
We enforce privacy through strict technical boundaries:
PII Redaction at the Edge: Real-time tokenization of identifiers before data reaches our models; audio is processed ephemerally.
Data Sovereignty: Self-hosted SLMs in isolated VPCs with AES-256 encryption at rest, TLS 1.3 in transit, and a strict zero-data-leakage policy (no public API training).
Regulatory Compliance: Built ground-up for India’s DPDP Act, ABDM standards for clinic data, and strict child privacy for K-12 learners. |
We address regulatory exposure through strict software boundary design:
Healthcare (CDSCO / SaMD): Mendly operates purely as a documentation scribe and clinic coordinator with mandatory clinician sign-off, avoiding Class B/C medical device licensing.
Data Privacy (DPDP Act): Built for DPDP compliance using ephemeral audio processing, real-time PII scrubbing, and strict no-profiling protections for K-12 students.
Agriculture: BhoomiAI restricts crop advisories to ICAR-verified treatments with strict dosage boundaries to eliminate agronomic liability. |
Scaling our system by an order of magnitude won’t break our business logic, but it will put severe pressure on three specific layers of our technical infrastructure:
1. Concurrent Audio Streaming & GPU Compute Bottlenecks
The Failure Mode: Real-time Indic ASR inference for Mendly (ambient doctor-patient consultations) and BhoomiAI (farmer voice queries) is compute-heavy. Scaling 10x means handling hundreds of parallel multi-tenant audio streams simultaneously. Without architectural adjustments, GPU queue latency will spike past our 2-second SLA, driving up cloud inference costs and exhausting dedicated GPU memory (VRAM).
The Fix: Move from running full-precision models in monolithic instances to quantized Small Language Models (SLMs) and INT8/FP8 ASR models served via optimized inference engines like vLLM or TensorRT-LLM with dynamic continuous batching. For less time-sensitive tasks (like post-op nursing notes or aggregate agronomic summaries), we shift workloads to asynchronous background queues (Celery/Redis).
2. Edge Ingestion & Rural Connection Drops
The Failure Mode: When 10x more farmers use BhoomiAI in rural areas with spotty 2G/3G connectivity or high packet loss, naive bidirectional WebSocket audio streaming will fail repeatedly, leading to dropped calls, truncated audio buffers, and incomplete RAG queries.
The Fix: Implement client-side progressive audio compression (Opus codec at low bitrates), deterministic client-side audio chunking with local SQLite caching, and an asynchronous retry mechanism that seamlessly resumes data transfer as soon as signal recovers.
3. Vector Database Throughput & Hybrid Search Latency
The Failure Mode: Scaling to millions of historical patient interaction tokens, hyper-local farm soil/weather records, and curriculum graphs will bottleneck our hybrid retrieval pipeline (dense vector search + BM25 keyword matching). High concurrent query volumes on vector indices (like HNSW) degrade latency and consume significant RAM.
The Fix: Decouple hot and cold data using tiered indexing: partition vector collections by domain, tenant, and geographic region (e.g., regional agro-climatic zones or specific clinic clusters) so each search traverses a compact, localized partition rather than a global index. |
Yes. Led by Co-Founder & CTO Sambasiva Rao Suda (scalable systems & inference architecture) and Chief Advisor Dr. Asadi Srinivasulu (IIIT-Allahabad alumnus, 20+ years in AI/NLP research). Our in-house engineering team covers:
Fine-tuning Indic ASR models for regional dialect code-switching.
Building hybrid RAG pipelines backed by domain ontologies (ICD-10/ICAR).
Edge-optimized vision models for mobile crop diagnostics.
Hardened MLOps with sub-2s inference SLAs. |
We use a tiered data strategy: we combine verified, open-license public domain truth with internally curated, proprietary field interaction data. We do not scrape unverified web data.
1. Agriculture (BhoomiAI)
Agronomic Guidelines & Packages of Practices (Licensed / Public Domain): Sourced from the Indian Council of Agricultural Research (ICAR) and State Agricultural Universities (SAUs) under open-access public research licenses. Covers crop life cycles, pest thresholds, and regional treatment protocols.
Weather & Remote Sensing (Open / API Licensed): Meteorological telemetry from India Meteorological Department (IMD) open feeds and multispectral satellite bands (Sentinel-2 via Copernicus open-access).
Crop Pathology & Dialect Audio (Owned):
Proprietary mobile-captured leaf pathology images annotated directly by local agricultural extension officers.
Field audio recordings of farmer queries in rural Telugu and Kannada captured during pilot operations.
2. Healthcare (Mendly)
Medical Ontologies & Vocabularies (Licensed / Open Standards): Standardized clinical reference terminologies, including ICD-10 and SNOMED-CT (licensed via India’s National Release Centre under MoHFW/ABDM), alongside open-source pharmacology registries for generic drug names and contraindications.
Ambient Clinical Transcripts & Code-Switching Data (Owned): Internally curated synthetic and de-identified pilot clinical interaction audio, mapping Indian clinician-patient code-switching (e.g., Telugu-English, medical Hinglish) directly to standardized SOAP schemas.
3. Education (dhimantAI)
Curriculum Taxonomies & Learning Objectives (Licensed / Open Access): Standardized NCERT, CBSE, and State Board curriculum frameworks mapped under educational fair use and open government educational licensing.
Socratic Misconception & Reasoning Graph (Owned): Internally designed dependency trees that map student error patterns to underlying conceptual gaps across middle and secondary math and science. |
We continuously improve models through a closed three-step MLOps flywheel:
HITL Correction Deltas: Logging manual doctor edits on SOAP notes and agronomist overrides on crop alerts as high-conviction supervised fine-tuning data.
Active Learning on Uncertainty: Isolating low-confidence dialect audio and borderline leaf pathology images for targeted internal annotation.
Automated Regression CI/CD: Benchmarking every model update against golden test sets for sub-2s latency, WER, and zero-hallucination schema conformance before production rollout. |
India-first, scaling to the Global South.
We are hardening our architecture against India’s complex realities: multilingual code-switching, patchy connectivity, and high-volume, resource-constrained frontlines. Solving for low-bandwidth rural farms and overburdened Tier-2/3 clinics creates a battle-tested blueprint directly exportable to emerging markets across Southeast Asia and Africa facing identical infrastructure and workforce gaps. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/tjo05atTxL3C-Optficial-Pitchdesk.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/tjo05atTxL3C-Optficial-Pitchdesk.pdf |
https://drive.google.com/drive/folders/1CvsDTOUGNN9E_-y3ONBtW98wnlfaNy2x |
Yes. Optficial Labs is built around three foundational pillars: Earth, Body, and Mind. We deploy localized, voice-first AI into high-friction frontline environments to solve systemic resource shortages:
BhoomiAI: Lowers smallholder farm input costs via vernacular pest/weather intelligence.
Mendly: Cuts clinician burnout and elevates care quality in busy Tier-2/3 OPDs.
dhimantAI: Delivers scalable Socratic tutoring to underserved learners. |
Yes (Non-traditional background & regional focus). Led by an ex-IAS officer and veteran systems architect operating out of South Indian non-metro clusters, our founding team breaks the standard tech-hub founder mold—bringing deep grassroots administrative insight and field-level operational access directly to frontline Indian problems. |
https://iitacb.org/about-us/#:~:text=IITACB'smentoringprogrammesleveragetheexpertiseof,Centreahubofmentorshipandinnovation. |
checked |
| Sep 23, 2026 @ 12:00 PM |
Devyani Parameshwar & Raghavendra Badaskar |
devyani.p@merakal.in |
http://www.linkedin.com/in/devyani-parameshwar |
https://bills2score.com/ |
9742011577 |
Devyani leads business development, partnerships and daily operations, whilst Raghav leads technology and product development. Both collaborate on the product strategy. |
We were ex colleagues 15 years ago, and have been working together now for 3 years. |
two |
We rarely agree upfront as we think differently, and push each others thinking. |
Mera Kal |
https://bills2score.com/ |
Bangalore |
Credit builder solution for new to credit customers. |
Digitally active customers are excluded from formal finance by lenders owing to the credit score gate. |
Credit line on UPI restricted to paying utility bills for new to credit customers to utilise and simultaneously build their credit profile. There are > 110mn UPI active customers who are NTC |
Acquiring rejected NTC customers, restricting the credit line and gamifying score build reinforce each other and has not been done before. Offering respect to rejects. |
two |
Signups, Testimonials |
NTC customers who are UPI active; basis research, 35% are small business owners, 57% are salaried, 7% are students and new to the job market (interns). 78% have been rejected for credit on account of having no bureau history. |
120mn |
80mn |
10mn |
Customer subscription to build scores, loan facilitation fee for those who graduate to larger ticket loans post score build. |
Secured card services, though not meeting the same need entirely. |
Via lenders and loan marketplaces who reject 30-90% of applicants, often due to no credit history. |
B2B2C, SEO/AEO |
To build a platform that continuously innovates on products and enablers to push lenders to include more marginal customers more affordably. |
Incorporated as a private limited co |
1.5Cr |
one |
We require support to de-risk a new model and co-opt lending partners to innovate. |
Get to go live with a lender partner |
Yes, though happy to engage physically where possible as we are Bangalore based. |
one |
two |
x |
x |
xx |
xxx |
xx |
xx |
xx |
xxxx |
xx |
ss |
India to begin with. There are over 3 dozen other emerging markets that face a similar challenge, where the model could be tailored to the country context to solve the same challenge. |
|
|
We’re building Mera Kal to reimagine financial services for underserved India |
Woman founder, serving the under-served. |
Chittaranjan and Chaitali Savdi |
|
| Sep 22, 2026 @ 4:10 PM |
Mr. Sunil Kandpal |
care@icbrwellness.com |
https://www.linkedin.com/in/sunil-kandpal-digitalmarketing |
https://www.icbrwellness.com/ |
7042788829 |
Co-Founder & COO, Sunil Kandpal leads product development, technology, and operations at ICBR Wellness. He is responsible for developing and managing the digital platform, overseeing technology-related activities, coordinating product execution, and supporting the company's operational growth. |
The founding team came together through professional and entrepreneurial interactions and shared an interest in building a scalable digital healthcare platform. We have worked together on developing ICBR Wellness, with complementary responsibilities across business strategy, product, technology, operations, and partnerships. Our collaboration has strengthened through the product-development and market-validation journey. |
2 |
Our biggest strength is the complementary skill set of the founding team. We combine business strategy, product and technology, operations, partnerships, and fundraising capabilities. This allows us to approach ICBR Wellness from both the customer and business perspective while maintaining strong execution focus. We work closely across functions, adapt quickly to market feedback, and are committed to building and scaling the platform together. |
ICBR |
https://www.icbrwellness.com/ |
New delhi |
ICBR Wellness is a digital holistic healthcare platform connecting users with verified practitioners for personalized consultations, wellness services, products, and ongoing care. |
ICBR Wellness is solving the problem of fragmented and difficult-to-access holistic healthcare. Users often struggle to find trusted practitioners, access personalized consultations, and maintain continuity of care. Holistic healthcare services, practitioners, products, and follow-ups are often spread across different providers and offline channels, making the overall wellness journey inconvenient and unstructured. ICBR Wellness aims to address this by bringing verified practitioners, consultations, personalized wellness plans, wellness services, products, and ongoing follow-up into one digital platform, making holistic healthcare more accessible, convenient, and organized. |
ICBR Wellness provides a digital holistic healthcare platform that connects users with verified Ayurvedic and holistic healthcare practitioners. The platform enables users to discover practitioners, access online, in-clinic, or at-home consultations, receive personalized wellness plans, access wellness services and products, and maintain ongoing follow-up through one integrated platform.
The solution addresses the fragmented nature of holistic healthcare by bringing practitioner discovery, consultations, personalized care, services, products, and continuity of care into a single digital ecosystem. The current focus is on expanding the practitioner network, validating the platform with users, and building a scalable model for holistic healthcare delivery. |
ICBR Wellness differentiates itself by bringing verified Ayurvedic and holistic healthcare practitioners, consultations, personalized wellness plans, wellness services, products, and ongoing follow-up into one integrated platform. Unlike solutions focused primarily on consultations, products, fitness, or individual wellness services, ICBR is designed around the complete holistic healthcare journey. Our potential defensibility comes from building a trusted practitioner network, structured user-care journeys, practitioner-user relationships, and accumulated platform data and feedback as adoption grows. The combination of multiple care-delivery options—online, in-clinic, and at-home—also supports a broader and more convenient user experience. |
MVP |
Pilots |
ICBR Wellness primarily targets urban, health-conscious consumers aged 25–45, especially working professionals and entrepreneurs in Tier 1 and Tier 2 cities across India. They seek trusted Ayurvedic and holistic healthcare, personalized wellness guidance, convenient consultations, and ongoing follow-up. Secondary customer segments include families, wellness-conscious individuals, and consumers interested in Ayurveda, yoga, naturopathy, nutrition, and other holistic wellness services. In the future, ICBR Wellness also plans to serve corporate customers through employee wellness programs. |
₹1,50,000 Crore |
₹15,000 Crore |
₹150 Crore |
ICBR Wellness follows a multi-stream revenue model: revenue will be generated through practitioner onboarding/platform fees, commissions on consultations, subscription-based wellness programs, and wellness product sales. As the platform scales, we also plan to introduce premium memberships and corporate wellness programs. This model combines transaction-based and recurring revenue while allowing the business to monetize both healthcare services and the broader wellness ecosystem. |
Jiva Ayurveda, NirogStreet, HealthifyMe, cult.fit, Ayushakti |
ICBR Wellness plans to acquire customers through a digital-first and practitioner-led approach. We will use social media, educational wellness content, targeted digital marketing, SEO, referrals, and wellness communities to reach health-conscious consumers. Our onboarded practitioners will also help generate early users through their existing networks and referrals. We will build partnerships with wellness providers, healthcare institutions, and relevant communities to expand reach. Initial pilot programs and customer feedback will help identify the most effective acquisition channels before scaling them. |
ICBR Wellness will follow a digital-first, practitioner-led go-to-market strategy. We will initially focus on urban, health-conscious consumers in Tier 1 and Tier 2 cities and build awareness through social media, educational wellness content, SEO, targeted digital marketing, and wellness communities. We will leverage our onboarded Ayurvedic practitioners to generate early users through practitioner referrals and their existing networks. In parallel, we will develop partnerships with wellness providers, healthcare institutions, and relevant communities to expand reach. The initial phase will focus on practitioner onboarding, pilot user acquisition, and market validation. Customer feedback, engagement, retention, and acquisition economics will be tracked to identify effective channels. Once validated, we will scale the most efficient acquisition channels and expand into additional cities and future corporate wellness partnerships. |
Our long-term vision is to build a trusted, scalable digital ecosystem for holistic and preventive healthcare in India, making personalized care accessible across cities and communities. We aim to connect users with a strong network of verified practitioners and provide an integrated journey covering consultations, personalized wellness plans, wellness services, products, and ongoing follow-up.
Over time, ICBR Wellness plans to expand across India, strengthen its practitioner network, develop corporate wellness offerings, and use technology and data-driven personalization to improve the user experience. Our goal is to establish ICBR Wellness as a comprehensive platform for accessible, personalized, and continuous holistic healthcare. |
Private ltd |
0 |
Yes |
We are applying to Avinya to access structured mentorship, healthcare-sector expertise, market validation support, and strategic networks that can help us move from early product development to commercialization. We particularly seek guidance on customer acquisition, practitioner-network expansion, business-model refinement, partnerships, and fundraising. Avinya’s ecosystem can help ICBR Wellness validate its model, strengthen execution, and build a scalable holistic healthcare platform. |
Validate product-market fit through structured user and practitioner feedback. Expand our verified practitioner network and strengthen our service delivery model. Refine our business and revenue model and establish clear unit economics. Develop an effective go-to-market and customer acquisition strategy. Build healthcare and strategic partnerships for pilots and market access. |
Yes. We are open to both hybrid and virtual participation. This flexibility would allow our team to actively engage in mentorship sessions, workshops, expert interactions, and programme activities while continuing product development and market-validation activities. |
Yes |
Supporting feature |
The AI architecture is currently in the planned development stage. We envision a modular architecture consisting of a secure data layer, user and practitioner inputs, a personalization/recommendation layer, and an application layer integrated with the existing digital platform. Relevant user-provided and practitioner-validated information would be processed to generate personalized wellness insights, with human oversight. The system will be designed with privacy, access controls, monitoring, and scalability in mind. AI/ML models and infrastructure will be selected and implemented during the next stage of product development based on validated use cases and data availability. |
At the current stage, ICBR Wellness does not yet have a proprietary data advantage because the platform is still in the early product-development and market-validation phase. As the platform scales, we plan to build a structured, consent-based dataset from practitioner interactions, user wellness preferences, consultation journeys, service usage, and feedback. This data can help us improve personalization, understand user needs, and optimize the platform over time. |
Our AI component is currently in the planned development stage and therefore does not yet have a proven defensibility advantage. As we develop it, our differentiation will come from combining structured practitioner knowledge, user wellness data collected with consent, practitioner feedback, and longitudinal user interactions to support personalized wellness recommendations. Over time, the resulting proprietary dataset, domain-specific workflows, practitioner network, and continuous product learning can create a stronger competitive barrier than generic AI technology alone. |
The AI component is currently in the planned development stage, so model performance metrics are not yet available. Once implemented, we plan to evaluate reliability using metrics such as accuracy, precision, recall, recommendation relevance, user feedback, and practitioner validation. We will also monitor consistency, error rates, and safety through ongoing testing and human oversight before deploying AI-generated recommendations in the platform. |
ICBR Wellness is committed to protecting user and practitioner data through consent-based data collection, controlled access, secure data storage, and appropriate authentication and access controls. As the platform develops, we will implement applicable Indian data-protection, healthcare, and privacy requirements, including appropriate consent, data minimization, retention, and user-data handling practices. Sensitive information will be accessed only by authorized personnel for legitimate service purposes. We will also conduct security testing and periodic reviews as the platform scales. |
Yes. As a digital holistic healthcare platform, ICBR Wellness may face regulatory requirements related to healthcare service delivery, practitioner verification, Ayurvedic practice, teleconsultation, consumer protection, advertising claims, and collection and handling of personal health data. We plan to address these risks through appropriate practitioner verification, clear service and health-related disclaimers, informed user consent, data-protection measures, and ongoing legal and regulatory review as the platform expands. |
The AI component is not yet deployed, so there is currently no production AI system to stress-test at 10× scale. When implemented, the main scaling risks would be increased infrastructure and processing costs, slower response times, data-quality issues, and maintaining recommendation consistency and reliability as the volume and diversity of user data increase. We plan to address these through scalable cloud infrastructure, monitoring, testing, data-quality controls, and human oversight. |
No. ICBR Wellness does not currently have a dedicated in-house AI/ML team. The AI/ML component is planned for a future stage of product development. Currently, the founding team focuses on product development, technology, operations, business strategy, and partnerships, while specialized AI/ML expertise can be brought in through external technical resources or future hiring as the AI component is developed. |
We are not currently using any proprietary or licensed datasets for AI/ML, as the AI component is still in the planned development stage. The current platform primarily uses operational and user-provided information required for delivering services, subject to appropriate consent and privacy controls. As AI development progresses, we plan to use consent-based, relevant platform data and appropriately licensed or publicly available datasets where required, with clear ownership and usage rights. |
As the AI component is still in the planned development stage, we do not yet have production model-performance data. Once implemented, we plan to continuously improve performance through practitioner-validated feedback, user feedback, monitoring of recommendation relevance and accuracy, regular testing, data-quality checks, and periodic model evaluation. We will also maintain human oversight and update the system based on observed errors, changing user needs, and validated domain knowledge. |
ICBR Wellness is initially focused on the Indian market, particularly Tier 1 and Tier 2 cities, where we aim to validate the platform, build our practitioner network, and establish product-market fit. Once the model is validated and scalable, we plan to expand into international markets where Ayurveda and holistic wellness have growing demand, adapting the platform to local regulations and customer needs. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/eHzp2ZRklACZ-ICBR-final-Deck.pptx.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/eHzp2ZRklACZ-ICBR-final-Deck.pptx.pdf |
https://drive.google.com/file/d/1pFUtO-SLhmvkJZ3-9heMPB0GtTn5yugw/view?usp=drive_link |
Yes. ICBR Wellness is mission-driven with a focus on improving access to personalized, preventive, and holistic healthcare. Our mission is to make trusted holistic wellness services more accessible by connecting users with verified practitioners and bringing consultations, personalized care, wellness services, products, and follow-up into one platform.
We aim to create impact by improving access to practitioner-led care, supporting preventive wellness, enabling continuity of care, and creating digital access for consumers beyond major healthcare centres. As we scale, we also aim to create opportunities for qualified practitioners to reach and serve a broader user base. |
Yes. Our founding team includes women entrepreneurs, and we are building a healthcare technology startup from India. We aim to create an inclusive platform that improves access to holistic healthcare and creates opportunities for healthcare practitioners. |
through linkedin |
checked |
| Sep 22, 2026 @ 3:51 PM |
Dipendu Biswas |
biswas.dipendu@gmail.com |
https://www.linkedin.com/in/dipendubiswas/ |
https://tarisa.vercel.app/lms.html |
+91 9980796314 |
Dipendu is a Technology Architect and does product development. He has 26 years of experience in digital transformations for Capital One, Nationwide Insurance, Credit Suisse and ANZ Bank. He has professionally practiced technology architecture for Accenture, EY, Nokia and Wipro. Karatoa looks after talent acquisition and public relations. She has 5 years of experience in Information Technology training and education. |
We met through networking. We have been working together for over 6 years. |
2 |
We have pitched to angel investors and venture capitalists and convinced them on the investment opportunity. We have received a soft commitment from an investor. We have received recognition for our venture development by Government of Karnataka, Government of India, Bank of Baroda and Reserve Bank of India. Press releases have covered us and our startup, which are social proofs. |
Tarisa Technologies |
https://fin.tarisa.co.in |
Bengaluru |
Software as a Service to determine the creditworthiness of small and medium businesses using alternative data sources. |
Micro, small and medium enterprises form a large portion in their contribution to the economy. They may not have credit history with the bureaus. They are often financially excluded as financial institutions find it prohibitive in terms of cost for lending in small ticket sizes. Financial analysts have to process volumes of disparate information about MSME and hence it takes lot of time. |
We have developed a platform for underwriters to assess financial health of small and medium businesses interactively. Biases of gender, economic status, industry and community are checked and removed. Our service determines the probability of default in repayment of loan by a borrower accurately. Small ticket sized loans are assessed in real-time. Large ticket sized loans are assessed in due course of time. Our platform learns about simple cash flows of businesses quickly. It requires adequate time to learn about complex cash flows of businesses. |
While business analytics is the base for processing huge volumes of data. It has to be extended with inferencing capabilities of AI for high productivity in loan processing. |
MVP |
Testimonials |
Financial institutions, underwriters and analysts |
489 billion USD |
2 billion USD |
150 million USD |
B2B subscription and Annual Maintenance Contract |
Lendingkart, Kaleidofin, Bluevine, Ondeck |
Founder sales |
Few financial institutions as design partners. Exhibits for financial industry trade shows. |
Become the best in providing simplest of solutions with technology. Evolve an ecosystem of trust in business to business interoperability. |
Private Limited Company |
NA |
Yes |
Continuous venture acceleration |
GTM |
Hybrid |
Yes |
Core engine |
Our technology stack for AI uses open source components like Apache Spark, Zeppelin notebook, Hadoop distributed file system, R packages and Python libraries. We use open source pre-trained models. However, we have intellectual property in algorithms to enhance data science, machine learning and neural networks. Base models are enhanced with loan book data with proprietary algorithms to make it specific to financial institutions. |
Initial training dataset is 70% of financial history available at any point in time. Data is annotated by data scientists for samples and then annotations are automated. Production models are available for check bounce, account balance shortfall (monthly, quarterly), repayment default, tax invoices, annual returns and tax returns. |
Banking and financial services require compliance with regulations. Hence, neural networks in our service have sufficient depth so as to provide explainability. |
A sample set from production is used for simulations and performance analysis. Predictions of loan default will be measured against repayment made within due date or loan underwritten. Probability of default is directly proportional to the rate of interest charged to borrower. Probability crossing defined thresholds will lead to rejection in loan sanctions. 5% inaccuracy will be tolerated. |
We use RSA for encryption. OAuth2 for authentication and authorization. ISO 20022 for payment transactions. API are REST compliant for interoperability. |
We will comply with NIST guidelines and GDPR regulation apart from DPDPA. Our SaaS will be certified by AWS auditor partners. We will apply for CERT-IN. |
We require 8 GPU for 91 hours per month for pre-training models. Fine tuning of models require 1 GPU for 8 hours.
The computation is required for 25 hours per week. The capacity costs USD 98.32 per hour. So, the estimated cost is USD 9832.00 per month. Hence our computation for machine learning is scaled down.
We are currently using credits from AWS for our Software as a Service as the cost of expensive computation is not covered by revenue or venture capital.
Beyond the timeline of this program, we will require finances to hire additional software engineers. We will raise investment from Venture Capital funds after this program. |
Dipendu Biswas has 26 years of experience in digital transformations for Capital One, Nationwide Insurance, Credit Suisse and ANZ Bank. He has professionally practiced technology architecture for Accenture, EY, Nokia and Wipro. 10 years of experience in providing technology solutions to Banks, Custodians, Financial Asset Management Companies, Insurers. Developed ISO 20022 based payment rails for Banks. Part of development team for enterprise security framework of State Street. Provided architecture assessment framework for commercial banking in Lloyds Bank. Part of mobile banking project for ANZ Bank. Developed content management system for National Australia Bank. Developed tools for analytics of Capital One Bank. Built microservices for Nationwide Insurance. Provided consulting to Credit Suisse for analytics in Wealth Management. Provided advisory as part of Cloud Centre of Excellence to group companies of AIA.
Extensive experience in Architecture, Design, Development and Implementation of Enterprise Solutions. Applied open source frameworks for corporate solutions. Practiced Agile methodologies. Good experience with application and web servers: Websphere, Weblogic, JBoss, Tomcat, Apache. Good experience with Spring, Struts, Axis, Hibernate, Google Web Toolkit, Ant, Maven, Docker, Kubernetes. Familiar with tools like JProbe, Soap UI, Eclipse, CVS, ClearCase, Rational Software Architect. Consulting assignments in Banking and Telecom domains. Collaborative development with teams across geographies. Good experience with Natural Language Processing, Blockchain, AWS, GCP. Developed software in NodeJS, Java, MongoDB, SQLlite, Android, AngularJS, Ethereum, Web 3.0, Python, Pytorch, Spark, Hadoop.
Karatoa Biswas has 5 years of experience in Information Technology training and education. |
Financial history is sourced from various data sources like core banking systems, credit bureaus, account aggregators. They are run through data quality processes and transformed before being saved in data lake. Databases like MongoDB, MySQL, Apache Hive are used for staging data. |
Machine learning is tuned periodically to reduce inaccuracies in predictions. |
MEITY invited us to the Global Partnership on Artificial Intelligence Working Group Convening in 2023. MEITY recognised our work in supporting the outputs of Responsible AI Working group. We were amongst 10 gamechangers in AI who were sponsored by MEITY for exhibit at Global Partnership for Artificial Intelligence at Bharat Mandapam in 2023. We were amongst top 30 startups selected by IIT Kharagpur and Yes Bank. We were selected by nasscom amongst 20 startups for G20 Digital Innovation Alliance where we exhibited our fintech. We were amongst 8 Indian and 10 Swiss startups in the Global Innovation Alliance Market Access Program. We were top 30 startups in GCC event by Gulf Islamic Investments and IIT Bombay. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/1TQXicKpQMrq-Underwriter-copilot.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/1TQXicKpQMrq-Underwriter-copilot.pdf |
https://www.youtube.com/watch?v=yoFuHPAC6YI |
We are on a mission to become the best in providing simplest of solutions with technology. We use AI to determine the creditworthiness of financially underserved population. Our solution also helps financial institutions to determine the probability of default by borrowers so that they can price the risk appropriately and not deny the finance. Financial institutions can subscribe to our Software as a Service on monthly or yearly plans. We provided live demonstration of innovation in financial services to leadership of Bank of Baroda amongst 10 other teams across India. We were a top 10 finalist in voice assisted financial services in local languages in a competitive event hosted by Reserve Bank innovation hub in India. We have demonstrated prototype for retail transactions in offline mode to Reserve Bank of India in a competitive event amongst 8 other solution providers in India who were shortlisted. International Financial Services Centres Authority of India has recognised us as a winner in Longevity Economy Cohort Program. We won fintech category at IIT Madras E Summit where Kotak Mahindra Bank was a sponsor. We won promising startup award for national export excellence from export council in India. We won Tally MSME Honours as a Next Gen Icon for identifying a market gap and introducing a solution which was sponsored by DBS bank. We were selected into a business accelerator of ICICI Bank and Beyond Seed with cohort size of 38 startups. Jury appreciation for business excellence at the BML Munjal Awards 2026 of Hero Enterprise. We were amongst top 15 fintechs shortlisted for digital adoption by State Bank of India. We were recognized as top 10 AI gamechanger in an event of OECD. See press release here
https://pib.gov.in/PressReleasePage.aspx?PRID=1984364 |
Woman co-founder |
ACSEL, GoK, KDEM |
checked |
| Sep 22, 2026 @ 1:07 PM |
Khyati Jain & Vanshika Modi |
contact@cycleofsamsara.com |
https://www.linkedin.com/in/khyati-jain1/ |
http://www.cycleofsamsara.com |
9874704678 |
Khyati Jain — CEO & Co-Founder: Background in luxury brand PR, D2C brand building and brand strategy. Leads business, growth, brand, operations and overall strategy at Cycle of Samsara. Vanshika Modi — CPTO & Co-Founder: Leads product, technology, UI/UX and customer journey. Focuses on building the technology and product infrastructure powering COS. |
We met in London while studying and became flatmates. We started working together through our shared interest in fashion and entrepreneurship and went on to build Cycle of Samsara together. We have worked together as co-founders since the inception of COS. |
2 |
Our biggest strength is the combination of brand, business and growth expertise with product, technology and UX. We complement each other closely, allowing us to move quickly from identifying a customer problem to building, testing and executing the solution. |
Cycle of Samsara |
www.cycleofsamsara.com |
Kolkata |
Cycle of Samsara is India's trust-first marketplace for verified pre-owned premium and luxury fashion. |
India's luxury and premium fashion resale market is large but unorganised — plagued by fake listings, inconsistent quality, and payment risk, largely conducted informally over WhatsApp. High-value fashion needs stronger trust infrastructure than generic peer-to-peer resale platforms provide, and no funded Indian player has solved this at scale. |
Cycle of Samsara provides an end-to-end, trust-first resale experience: seller KYC and verification, curated listing approval, AI-assisted plus physical authentication, managed logistics, and escrow-based payments that protect both buyer and seller until delivery is confirmed. |
We combine curated luxury inventory with a built-in trust layer, rather than operating as an open classifieds marketplace. In Q3 2026 alone we authenticated 51 items and stopped ₹2.23L of bad GMV with zero counterfeits delivered across 130 paid orders — a defensibility competitors relying on open listings can't easily replicate. Our House of Samsara pillar (owned vintage supply, influencer drops, designer deadstock collaborations) adds a further inventory edge, and our authentication API is a path to licensing this trust layer as B2B infrastructure. |
Revenue |
Users, Revenue, Pilots, Testimonials |
Urban Indian women aged 22–40 in metro cities who buy ethnic, occasion, and designer wear in the ₹3,000–₹1,00,000 range — value-conscious about spending on high-cost occasion wear worn only once or twice, and increasingly open to verified pre-owned as a smarter alternative to new purchases. On the supply side: sellers holding underused designer and luxury pieces looking for a trusted resale channel. |
~$9–10B (India's apparel resale market) |
~$1.5–2B (India's premium, luxury and occasion-wear resale segment specifically) |
Estimated at ~₹4.11 Cr GMV by FY27–28 based on current growth trajectory (69% QoQ GMV growth, 270+ verified sellers, expanding to tier-2 cities) — this is a projection rather than a bottom-up SOM calculation, worth flagging if the reviewer probes it closely. |
25% commission on transactions, plus authentication and delivery/service fees. Future streams include premium listings, cleaning & restoration, and seller visibility services. Asset-light — no inventory purchased. |
Luxepolis (luxepolis.com), Ziniosa (ziniosa.com), and Circle (circlestore.in) in India; Vestiaire Collective and The RealReal internationally as aspirational benchmarks. |
Primarily organic — 600K+ monthly Instagram reach at zero paid spend, plus micro-influencer collaborations in fashion and sustainability niches and a referral programme. Currently near-zero CAC; targeting ₹200–300 CAC via Meta Ads post-raise as paid acquisition scales. |
Phased: organic-led launch and seller onboarding (complete), followed by targeted Meta Ads and influencer-led acquisition in metro cities, a curated Bridal Edit to capture wedding-industry resale supply, SEO-led content, and eventual tier-2 city expansion alongside scaled AI authentication. |
To become India's trust layer for fashion resale — expanding Cycle of Samsara's own marketplace across categories and cities while licensing our Samsara Authentication API as B2B infrastructure to other resale and reselling platforms, making verified, trust-first resale the default way India buys and sells pre-owned fashion. |
Styleise Outfitters LLP; DPIIT-recognised; based in Kolkata |
₹0 external funding raised to date — entirely bootstrapped, with ₹5,00,000 in founder investment deployed. Currently raising a ₹3.5 Cr Seed round (open, no cheques closed yet). |
Yes |
We're applying to access Avinya's mentorship network, technical infrastructure support, and seed capital to strengthen our AI-assisted authentication system — the core technology layer protecting trust in our marketplace. While our business is a consumer fashion marketplace rather than a traditional AI/deep-tech startup, our authentication engine is a genuine applied-AI problem (image and pattern recognition for verifying luxury and designer item genuineness), and we see real value in Avinya's technical mentorship for hardening and scaling that system. |
Over the 4-month programme, we aim to scale our automated authentication screening to cover a larger share of our catalogue, reduce reliance on manual physical verification, and refine detection accuracy — directly supporting our path to operating breakeven by February 2027. |
Hybrid & Virtual |
Yes |
Automation layer |
Our authentication system combines an AI-based image-screening layer with manual physical verification. The image-screening model analyzes listing photos to flag likely counterfeit or inconsistent items before publication, acting as a first-line filter; items above certain value thresholds or flagged by the model proceed to physical authentication by our team. This hybrid AI-plus-human architecture lets us scale screening volume while keeping a human check on higher-stakes items. |
We're accumulating a growing dataset of verified transactions and authenticated items — 270+ verified sellers and 51 authenticated items to date — pairing listing images with confirmed authenticity outcomes from real marketplace activity. This labeled, transaction-linked dataset is early-stage but grows with every order, and is not something a generic classifieds platform accumulates. |
At our current stage, defensibility comes less from model sophistication and more from the compounding data loop: every transaction adds verified, labeled training data that a new entrant starting from zero doesn't have. Combined with our seller network and trust-layer workflow (KYC, escrow, authentication) as an integrated system, this is harder to replicate than the underlying model architecture alone. |
We are formalizing structured evaluation metrics (precision/recall on flagged listings, false-positive rate) as authentication volume scales; today, model outputs are validated against our physical authentication step, giving us a real-world accuracy check on every flagged item before it reaches a buyer. |
Seller KYC data, buyer information, and listing images are stored within our platform infrastructure with access limited to authorised team members. We are aligning our data handling practices with India's Digital Personal Data Protection (DPDP) Act requirements as we scale. |
Two areas we're actively managing: GST invoicing treatment for resale transactions (Margin Scheme vs. full value, currently being finalised with our CA), and DPDP Act compliance for KYC and personal data handling as our user base grows. |
Our current model relies on a relatively small labeled dataset; at 10x volume, we'd expect false-positive/false-negative rates to become more visible and the manual authentication step to become a bottleneck unless automated screening coverage expands proportionally. Scaling requires both a larger training dataset and infrastructure investment in the screening pipeline — which is the core use case for this programme's support. |
Yes — Manikant Pandey, our Full Stack AI Engineer, built and maintains our authentication and platform AI systems in-house. |
We use our own proprietary dataset — listing images and authentication outcomes generated through platform transactions. We do not rely on third-party licensed datasets for our authentication model. |
As authentication volume grows, we continuously feed new verified transaction outcomes back into the model to expand its training data and improve detection accuracy. Programme funding and support would help us formalise this feedback loop and scale automated first-line screening across a larger share of our catalogue. |
India to begin with then expand to south east Asia |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/4yKeVdBWyFCi-Cycle-Of-Samsara-Pitch-Deck_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/4yKeVdBWyFCi-Cycle-Of-Samsara-Pitch-Deck_compressed.pdf |
https://youtu.be/2z-SjaIjivw |
Yes. Cycle of Samsara is building circular-economy infrastructure for India's fashion industry — extending the lifecycle of premium and luxury garments that would otherwise sit unused after one or two wears, reducing textile waste and the environmental footprint of new production. Beyond sustainability, we're also solving a trust and access problem: making high-quality, verified fashion more affordable and accessible while giving sellers a legitimate, transparent way to monetise underused pieces in a market that today runs largely on unorganised, WhatsApp-based reselling. |
Yes — both co-founders are women in a startup ecosystem where women-led ventures remain significantly underrepresented, particularly in receiving institutional funding. |
NA |
checked |
| Sep 21, 2026 @ 1:28 PM |
Prakash Gupta |
prakash@scanzer.ai |
https://www.linkedin.com/in/prakash-k-gupta/ |
https://scanzer.ai |
+91 9980196769 |
Founder & CEO |
Our five-member core team has been built through long-standing professional relationships.
Four of us, including me, worked together for atleast 6 years at Oracle on the same product and developed a strong working relationship across engineering, research, infrastructure and product development. The fifth member was my colleague at Krowd for 5 years, where she led security for financial and banking clients.
Today, the core team comprises the Founder & CEO, VP Engineering, Head of Security & Infrastructure, Head of Research, and Principal Engineer. Collectively, we have worked together across these two organizations for many years, giving us an established foundation of trust, complementary expertise and experience building complex enterprise technology together. |
All |
Our biggest strength is the combination of deep technical expertise and long-standing trust.
We bring together product leadership, engineering, research, security and infrastructure experience, with several of us having already built complex enterprise technology together over many years. This allows us to move quickly across the full stack, from research and deep technology to production-grade systems and enterprise deployment.
Just as importantly, we know how each other works. That established trust lets us challenge ideas openly, make decisions quickly and stay focused on solving difficult problems rather than building the team from scratch. |
Scanzer |
https://scanzer.ai |
Bengaluru |
Scanzer is building independent trust infrastructure for AI, helping enterprises observe, control and prove how AI systems behave from intent to execution. |
As AI moves from generating answers to making decisions, executing workflows and initiating transactions, enterprises increasingly depend on systems they cannot fully observe or independently verify. Existing monitoring and governance tools largely rely on application logs, model outputs or vendor-reported metrics, leaving gaps in understanding what actually happened during execution, why systems diverged, and whether behaviour can be reproduced or audited. This creates significant security, compliance, operational and financial risk, particularly in regulated sectors such as banking, payments and healthcare. |
Scanzer provides an independent assurance layer for AI systems and their supporting infrastructure.
Our technology captures execution-level evidence, detects meaningful behavioural or environmental divergence, evaluates trust continuously, and creates audit-ready records that enterprises can use to investigate incidents, compare environments, enforce policies and verify system behaviour.
This enables organisations to independently assess AI models, applications and technology providers without relying solely on self-reported performance or traditional monitoring. |
Scanzer is differentiated by its ability to generate independent evidence across the AI execution stack, connecting application behaviour with runtime and infrastructure-level signals rather than observing only model inputs and outputs. Our defensibility comes from proprietary technology for execution intelligence, reproducibility, divergence detection and trust evaluation, supported by a growing patent portfolio with an initial priority filing in December 2025. The technology is designed to work across models, applications and infrastructure environments, creating a vendor-neutral assurance layer that becomes more valuable as enterprises operate increasingly complex and autonomous AI systems. |
MVP |
Users, Pilots |
Our primary customers are regulated enterprises and technology providers deploying AI in high-consequence environments. Initial focus is on banks, payment service providers, card and commerce networks, fintechs and healthcare technology companies. Within these organisations, our buyers include CIOs, CISOs, CTOs, Heads of AI/AI Governance, Risk and Compliance leaders, and platform engineering teams. Over time, we also see customers across cloud infrastructure, AI platforms, autonomous systems and semiconductor ecosystems. |
$560 B |
$224 B |
$560M (0.1 % of TAM) |
Primarily recurring B2B enterprise software revenue. Customers pay annual platform/licence fees based on deployment scope, environments and capabilities, supplemented by implementation and usage-based/API revenue where appropriate. As the platform scales, additional revenue models include OEM/technology licensing, embedded assurance, transaction-linked pricing for payment ecosystems, and licensing of selected capabilities to infrastructure and semiconductor partners. |
The competitive landscape includes AI observability and evaluation companies such as Arthur AI and Fiddler, AI governance platforms such as Credo AI, Holistic AI and OneTrust, and AI security companies such as Protect AI and HiddenLayer. Most existing solutions focus on specific layers such as model governance, application security or observability. Scanzer is building an independent assurance layer spanning AI behaviour, execution, runtime and underlying infrastructure, with evidence designed for reproducibility, investigation and continuous assurance. |
We use a founder-led, design-partner-driven enterprise sales model. We engage directly with banks, payment companies, AI companies and infrastructure providers through industry networks, conferences, accelerators, technology partnerships and senior enterprise introductions. Early engagements begin with a clearly defined assurance or security problem, followed by a pilot or design-partner deployment that demonstrates measurable value before expanding into continuous enterprise usage. |
Our GTM follows a focused land-and-expand strategy. We enter regulated enterprises through high-value problems where independent assurance is immediately relevant, initially banking, payments and AI-driven financial infrastructure. Examples include continuous assurance of AI and fraud-detection providers, runtime security, AI-agent trust and privacy/compliance assurance. We use focused pilots and design partnerships to establish measurable outcomes, then expand across additional AI systems, providers and environments within the organisation. In parallel, we are developing distribution through technology providers, financial ecosystem partners and infrastructure companies that can embed or offer Scanzer's assurance capabilities to their own customers. |
Our vision is for Scanzer to become the independent trust infrastructure for autonomous systems.
As software evolves from applications that execute predefined instructions to AI systems that perceive, decide and act autonomously, enterprises will need an independent layer that can determine what happened, whether it should have happened, whether it can be reproduced, and whether it can be trusted.
We aim to make this trust layer ubiquitous across models, applications, runtime environments and infrastructure, creating a common foundation for assuring autonomous systems from intent through execution to silicon. |
Incorporated - Jan 2024 |
~50L, Friends and Family |
Yes |
Reference from Chittaranjan Sawadi (Alum from this program) |
Paid Pilots, Product Market Fit, Preparation for next round |
Yes |
Yes |
Core engine |
Scanzer uses a model-agnostic architecture that observes AI execution, evaluates behaviour and consistency, generates trust signals, and maintains audit-ready evidence across application, runtime and infrastructure layers. The underlying evaluation methods are proprietary and part of our IP portfolio. |
Our primary advantage today is proprietary technology rather than a large owned dataset. Over time, deployments generate a growing corpus of execution patterns, divergences and reliability signals that can strengthen benchmarking and assurance. |
Our defensibility comes from proprietary execution intelligence, cross-layer evidence, reproducibility and continuous assurance. This is supported by our patent portfolio and strengthened by deployment-generated knowledge over time. |
We evaluate task success, consistency, reproducibility, behavioural divergence, policy adherence, latency, failure rates, robustness and traceability, with metrics adapted to each customer use case. |
We follow data minimisation, customer isolation, controlled access and configurable retention. Where required, Scanzer can operate within customer-controlled environments. The platform is being designed for requirements such as DPDP, GDPR and sector-specific regulation. |
Yes. AI, privacy and cybersecurity regulations are evolving rapidly. These create compliance requirements, but also increase demand for independent assurance, auditability and evidence. |
The main challenges will be execution volume, evidence storage, infrastructure diversity and maintaining low overhead. Our modular architecture is designed to scale ingestion, evaluation and evidence processing independently. |
Yes. Our core team combines AI research, engineering, security, infrastructure and distributed-systems expertise, with several members having worked together for many years at Oracle and Krowd. |
We use internally generated workloads, controlled test scenarios, public or appropriately licensed datasets, and customer/design-partner data where permitted. Our technology does not depend on a large third-party training dataset. |
Through controlled experiments, deployment feedback and accumulated execution evidence across models, workloads and environments. |
Both. India is an important initial market, particularly in banking and payments, while the platform is being designed from the outset for global regulated enterprises. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/QHCHE6uSVzFn-Scanzer_Angel_Investment_Snapshot.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/QHCHE6uSVzFn-Scanzer_Angel_Investment_Snapshot.pdf |
https://youtu.be/KdugziR4cLc |
Yes. Scanzer is building independent trust infrastructure for AI and autonomous systems.
As AI increasingly influences financial transactions, healthcare, commerce and other critical workflows, organisations need to know not only whether an AI system performs well, but whether its behaviour can be independently verified, controlled and audited.
Scanzer helps enterprises make AI systems more observable, reproducible and accountable, with evidence across software, runtime and infrastructure layers. Our goal is to reduce risks such as fraud, unsafe AI behaviour, operational failures and unverified automated decisions, particularly in regulated and high-impact sectors.
Our broader mission is to help make autonomous technology trustworthy enough to be deployed responsibly at scale, so that the benefits of AI can grow without compromising security, accountability or public trust. |
No |
Chittaranjan Savadi |
checked |
| Sep 19, 2026 @ 10:17 PM |
Shiv Shankar; RathaRamanan Natarajan |
shiv@easeworkai.com |
https://www.linkedin.com/in/shiv-shankar-mktg/ |
https://easeworkai.com |
9901931337 |
Shiv Shankar (CEO & Co-founder) MBA, Mechanical Engineer with 16+ years in global manufacturing. Led sales, marketing, and business development roles, working closely with enterprise customers across regions. Brings deep understanding of procurement workflows, customer pain points, and how buying decisions are made in large organizations. At EaseworkAI, leads GTM, customer acquisition, and product direction with a strong focus on real-world usability and outcomes. RathaRamanan Natarajan (CPTO & Co-founder) Mechanical Engineer with 20+ years of experience in procurement and supply chain transformation. Has led large-scale enterprise initiatives, delivering next-gen procurement solutions and driving digital transformation programs. Deep expertise in building AI/ML-driven systems for real-world enterprise use cases. At EaseworkAI, leads product architecture, AI development, and platform scalability, ensuring the solution is both technically robust and aligned with customer workflows. |
Met at a Startup mixer in late 2023 and have been building Easework AI since. |
All |
Our biggest strength is the combination of deep industry expertise, enterprise go-to-market experience and strong product and technology capabilities within the founding team. We understand the real-world complexity of industrial procurement and supply chains, while having the ability to build and deploy AI products that integrate into enterprise environments and deliver measurable business outcomes. |
Easework Technologies Pvt. Ltd. |
https://easeworkai.com |
Hassan, Karnataka |
Agentic AI for procurement and supply chain in manufacturing |
Large enterprises still run critical procurement and supply chain processes across fragmented ERP systems, spreadsheets, emails and manual workflows. This creates high operational effort, slow cycle times, inconsistent decisions, limited visibility into spend and missed savings opportunities, especially in complex industrial categories. Easework AI addresses this by using AI agents to understand enterprise and category context and execute these processes end to end. |
Easework AI is an agentic AI platform that automates enterprise procurement and supply chain workflows end to end. AI agents combine deep category intelligence with structured and unstructured enterprise data to understand context, make decisions and execute processes across sourcing, supplier management, purchasing, invoicing and payments, while integrating with existing ERP systems. This helps enterprises reduce manual effort, accelerate cycle times, improve procurement decisions and capture measurable cost savings. |
Our differentiation comes from combining deep procurement category intelligence, enterprise data handling and end-to-end AI execution. Our proprietary procurement knowledge graph captures category-specific relationships across specifications, suppliers, pricing, alternatives and negotiation strategies, while our data layer can harmonize structured and unstructured enterprise data. This enables AI agents to execute complex, connected procurement workflows with business context, rather than simply generating recommendations. Over time, category intelligence, workflow knowledge and customer-specific process context improve the system’s ability to deliver reliable outcomes in complex enterprise environments. |
Revenue |
Revenue |
Procurement and Supply chain leadership at Manufacturing and industrial enterprises |
USD 22Bn |
USD 4Bn |
USD 1.2Bn |
Platform license + usage-based billing |
Lio, Procure.AI, Didero |
LinkedIn, Industry Events, Referrals |
Sales-led with land and expand. 4-week PoC followed by production contract. |
Build our own procurement small language model (SML). There are no procurement-specific language models. This gives us greater control over accuracy and costs. |
Incorporated in Karnataka, India |
USD 50,000 |
Yes |
We are applying to ACSEL to accelerate the next phase of Easework AI’s growth by strengthening our product, enterprise GTM and access to the right industry and investor networks. We have moved beyond the prototype stage and are deploying our AI agents with large enterprises, and we now want to scale from successful enterprise deployments to a repeatable growth model. ACSEL’s focus on AI, access to mentors and investors, and Karnataka’s strong technology and enterprise ecosystem make it particularly relevant to our journey. We also see the program as an opportunity to contribute our experience in industrial AI and enterprise procurement while learning from other founders and ecosystem leaders. |
During the program, we want to achieve three things: Strengthen our enterprise GTM: Build a repeatable sales and customer acquisition model for large industrial enterprises and sharpen our positioning, pricing and sales strategy. Accelerate product scale: Improve our AI agents, category intelligence and enterprise deployment capabilities to make implementations faster, more reliable and scalable across customers. Build strategic relationships: Connect with mentors, industry leaders, potential customers and investors who can help us expand our enterprise footprint and prepare for our next stage of growth. |
Hybrid |
Yes |
Core engine |
Easework uses a **multi-agent, enterprise AI architecture** built around a procurement knowledge layer, an enterprise data layer and an execution layer. The system ingests structured and unstructured data from ERP systems, documents, emails and other enterprise sources, applies taxonomy, harmonization and contextualization, and connects this information to our procurement knowledge graph and category intelligence.
AI agents use this context to reason, make decisions and execute multi-step procurement workflows through our integration and orchestration layer. The architecture supports multiple LLMs, including customer-provided or private models, with human-in-the-loop controls, business rules and validation mechanisms for reliable enterprise execution. The platform can be deployed in customer VPCs, private cloud or on-premises environments, allowing sensitive enterprise data to remain within the customer’s controlled environment. |
Our data advantage comes primarily from **procurement-domain knowledge and customer-specific process context**, rather than simply owning a large generic dataset. Easework’s procurement knowledge graph captures relationships across categories, specifications, suppliers, pricing, alternatives, negotiation strategies and workflows, while our platform continuously learns from customer-approved workflow outcomes and interactions.
This creates a growing layer of **contextual, category-specific intelligence** that helps our AI agents make better procurement decisions and execute workflows more reliably. Customer data remains isolated and controlled by the customer and is not used to train shared models. |
Our AI is defensible because it combines **procurement-specific category intelligence, enterprise data contextualization and workflow execution**, rather than relying on a generic LLM. Our proprietary knowledge graph captures relationships between specifications, suppliers, pricing, alternatives, policies and negotiation strategies, while our data layer handles fragmented structured and unstructured enterprise data.
The AI agents then apply this context to execute multi-step procurement workflows with business rules, validation and human oversight. As we deploy across categories and workflows, the accumulated category knowledge, workflow patterns and execution feedback improve reliability and create a domain-specific intelligence layer that is difficult to replicate simply by accessing the same underlying LLMs. |
We evaluate AI performance at both the model and workflow level, with a strong focus on reliable enterprise execution. Key metrics include field-level and task-level confidence scores, extraction and classification accuracy, decision/recommendation accuracy, task completion rate, hallucination/error rate, human override rate, response latency and end-to-end workflow success rate.
Confidence scoring is applied at the individual data-field and task level, allowing the system to identify low-confidence outputs and route them for validation or human review. For enterprise deployments, we also measure automation/touchless rate, cycle-time reduction, manual effort reduction and financial impact against the customer baseline. Agents use business rules, validation checks and human-in-the-loop controls, with performance tested on customer-specific data and workflows before production. |
Easework is designed with enterprise data privacy and security as core architectural principles. Customer data can remain within the customer’s controlled environment through on-premises, private cloud or VPC deployment, and customer data is not used to train shared AI models. We support customer-controlled LLM options, including BYO LLM keys and private LLMs, giving enterprises greater control over how their data is processed.
Our architecture includes role-based access controls, encryption, secure API integrations, auditability, human-in-the-loop controls and data isolation. We also apply data minimization and controlled access across the AI and workflow layers, with the objective of meeting enterprise security and compliance requirements, including SOC 2 and ISO 27001 as part of our security roadmap. |
The primary regulatory considerations are AI governance, data privacy, cybersecurity and sector-specific enterprise requirements across the countries where our customers operate. Regulations such as the EU AI Act and applicable data protection laws may impose requirements around transparency, data handling, risk management and human oversight, particularly as AI becomes more deeply embedded in business processes.
We mitigate these risks through human-in-the-loop controls, audit trails, role-based access, data isolation, customer-controlled deployment options and support for private or customer-provided LLMs. Because Easework is primarily an enterprise procurement and workflow automation platform, rather than a system making decisions in highly regulated domains such as healthcare or credit, we currently expect the regulatory risk to be manageable, while continuing to adapt our architecture and controls as AI regulations evolve. |
At 10× scale, the biggest challenges would be **enterprise data and integration complexity, AI inference cost and latency, and maintaining reliability across a much larger number of workflows and categories**. Each enterprise has different ERP systems, data structures, taxonomies and business rules, so scaling implementation without increasing engineering effort proportionally is a key challenge.
We are addressing this through our **in-house integration middleware, reusable data and taxonomy layers, modular agent architecture, model-agnostic LLM approach, confidence scoring and automated validation**. Our goal is to make integrations and agent deployment increasingly repeatable, while maintaining measurable accuracy, reliability and cost efficiency as transaction volumes and customers grow. |
Yes. We have an in-house AI/ML and product engineering team with expertise in agentic AI, LLM orchestration, NLP, information extraction, knowledge graphs, enterprise data processing and AI-driven workflow automation. The team has experience building AI systems that work with both structured and unstructured enterprise data, including taxonomy, data harmonization, contextual retrieval, confidence scoring and human-in-the-loop validation.
Our team also has deep procurement and supply chain domain expertise, allowing us to combine AI/ML capabilities with real-world enterprise workflows rather than building generic AI applications. We follow a model-agnostic approach, allowing integration with leading and private/customer-provided LLMs while maintaining control over the application, data and workflow layers. |
We use a combination of customer-provided enterprise data, proprietary procurement knowledge and licensed/publicly available data sources, depending on the use case. Customer data includes structured ERP data and unstructured documents such as specifications, invoices, contracts and supplier information, and remains isolated within the customer environment.
Our proprietary procurement knowledge graph and category intelligence are developed in-house using domain expertise, curated information and customer-approved data and workflows. We do not use customer data to train shared foundation models, and where external datasets or models are used, we comply with the applicable licensing and usage terms. |
We continuously improve performance through a closed feedback loop across data, models and workflow outcomes. Field-level and task-level confidence scores, human validations, overrides, errors and workflow outcomes help us identify where agents need improvement, while customer-approved data and feedback are used to refine category intelligence, prompts, retrieval, rules and agent workflows.
We also evaluate models against customer-specific test sets and production benchmarks, tracking accuracy, task completion, hallucination/error rates, automation rates and business outcomes. Our model-agnostic architecture allows us to adopt better LLMs and specialized models as they become available without rebuilding the application and workflow layer. |
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qB0Ywx1Aa850-Easework_Investor_Pitch-Aug-26-India.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qB0Ywx1Aa850-Easework_Investor_Pitch-Aug-26-India.pdf |
https://youtu.be/63xXUrgl8qA |
Yes. Easework AI is focused on helping India and the US strengthen and scale local manufacturing while navigating increasing global competitive and supply chain pressures. Manufacturers need to improve cost competitiveness, productivity and supply chain resilience while managing complex procurement operations and global sourcing dynamics.
Our mission is to use AI to make industrial procurement and supply chain operations significantly more efficient, helping manufacturers reduce costs, improve decision-making and respond faster to changing market conditions. By enabling companies to get more value from their existing people, data and systems, we aim to contribute to stronger, more competitive and resilient manufacturing ecosystems in both India and the US. |
NA |
NSRCEL |
checked |
| Sep 19, 2026 @ 9:52 PM |
Shiv Shankar; RathaRamanan Natarajan |
shiv@easeworkai.com |
https://www.linkedin.com/in/shiv-shankar-mktg/ |
https://easeworkai.com |
9901931337 |
Shiv Shankar (CEO & Co-founder) MBA, Mechanical Engineer with 16+ years in global manufacturing. Led sales, marketing, and business development roles, working closely with enterprise customers across regions. Brings deep understanding of procurement workflows, customer pain points, and how buying decisions are made in large organizations. At EaseworkAI, leads GTM, customer acquisition, and product direction with a strong focus on real-world usability and outcomes. RathaRamanan Natarajan (CPTO & Co-founder) Mechanical Engineer with 20+ years of experience in procurement and supply chain transformation. Has led large-scale enterprise initiatives, delivering next-gen procurement solutions and driving digital transformation programs. Deep expertise in building AI/ML-driven systems for real-world enterprise use cases. At EaseworkAI, leads product architecture, AI development, and platform scalability, ensuring the solution is both technically robust and aligned with customer workflows. |
Met at a startup mixer in Bangalore. Have been working together in Easework AI for 3+ years. |
All |
Our biggest strength is the combination of deep industry expertise, enterprise go-to-market experience and strong product and technology capabilities within the founding team. We understand the real-world complexity of industrial procurement and supply chains, while having the ability to build and deploy AI products that integrate into enterprise environments and deliver measurable business outcomes. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Sep 19, 2026 @ 4:14 PM |
Kriti Kumar, Kshitij Babbar |
kk@peopleaxis.ai |
https://www.linkedin.com/in/kriti-k/ |
https://www.peopleaxis.ai/ |
9930843210 |
Kriti Kumar - Co-founder & CEO, Kshitij Babbar - Co-founder & COO |
We met through a common reference (mutual ex-colleague) in April 2025 and started working almost instantly. Formed PeopleAxis Pvt Ltd in June 2025. |
All |
Our biggest strength is the combination of deep domain understanding and the ability to turn it into a commercially viable technology product.
Between the founders, we bring nearly four decades of combined experience across HR, business building and go-to-market. One side of the team understands workforce problems from inside organisations: how attrition, engagement and people decisions actually happen while the other brings the execution and commercial lens required to build, sell and scale a B2B enterprise product.
This gives us an important advantage: we are not using AI to solve a theoretical HR problem. We have lived the problem, know where traditional HR analytics falls short and are building PeopleAxis around decisions companies struggle to make today.
We have also demonstrated our ability to move quickly from insight to execution building VibeX, engaging enterprise customers, running pilots and continuously refining the product based on real-world feedback.
What binds the team together is a strong bias for action, resilience and a shared ambition to build PeopleAxis into the workforce intelligence layer enterprises rely on to understand risk, decide what action to take and measure whether that action actually worked. |
PeopleAxis Private Limited |
https://www.peopleaxis.ai/ |
Bengaluru |
PeopleAxis is building VibeX, a workforce intelligence platform that helps organisations understand emerging people risks, take timely action and measure outcomes to prevent avoidable workforce costs. |
Organisations invest heavily in hiring and developing employees, but they often lack a reliable way to identify and address emerging workforce problems. Relevant information is spread across HR, performance, compensation and employee-feedback systems. Reporting these inputs separately does not necessarily show which concerns are distinctive, how strong the evidence is or what a manager should do next. As a result, retention action may arrive too late or focus on the wrong issue. An avoidable exit can lead to recruitment, onboarding, lost productivity, tacit knowledge loss and management time. The problem is therefore not simply a lack of data: it is the gap between available information, a sound decision, timely action and evidence that the action worked. |
VibeX starts with approved workforce data and checks its quality before producing risk signals. The analysis separates the risk score from confidence in the available evidence and considers whether a pattern is individual, team-level or organisation-wide. Explainable outputs help HR understand why attention may be needed. Trigger-based listening can add fresh context and test a hypothesis without treating a survey response as proof of resignation intent. The action layer is designed to turn approved recommendations into accountable tasks, reminders and follow-ups. An outcome layer then tracks delivery and subsequent changes. Together, these steps form a repeatable process: understand the concern, validate the evidence, act within the right authority and evaluate the result. |
Our differentiation is the connection between intelligence, validation and execution. VibeX does not treat a risk score as a complete answer. It shows the evidence and its limitations, considers whether a concern is individual or systemic, and uses appropriately governed listening to test the interpretation. The next step is not merely another report: the platform is being developed to support approved interventions, assign ownership and measure results. This creates a feedback loop from workforce data to action and back to evidence. The technical and commercial value lies in making the whole process reliable and usable within enterprise HR workflows, rather than in adding a generic AI interface to existing data. PeopleAxis is building defensibility around the parts of workforce decision support that require sustained domain and product work: data contracts, risk and confidence logic, population-aware explanations, privacy boundaries, integration into HR workflows and rigorous evaluation. Enterprise adoption also depends on implementation quality and the trust earned through useful results. Over time, appropriately permissioned and governed learning from deployments may improve the product's understanding of which interventions are useful in which contexts. This is a development strategy, not a claim that the company already owns a unique large dataset or a patent portfolio. The strongest evidence of defensibility will be reliable performance, customer adoption and a workflow that is difficult to reproduce well. |
Users |
Pilots |
The initial customer focus is mid-sized and large enterprises where workforce size, replacement cost and fragmented people data make better retention decisions valuable. CHROs and HR heads are key buyers and sponsors, while CFOs and business leaders assess economic value and organisational priorities. HRBPs use the platform to investigate concerns and coordinate interventions within their assigned scope. Managers participate in authorised actions rather than receiving unrestricted employee-risk access. The initial go-to-market focus is India, with scope to expand after repeatable deployment and customer evidence are established. The strongest early customers are those willing to provide usable data, sponsor a pilot and evaluate outcomes against an agreed baseline. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Sep 19, 2026 @ 11:41 AM |
Ranjan Kumar Baisak |
ranjan@xhield.tech |
https://linkedin.com/in/ranjanbaisak |
https://xhield.tech |
9880398951 |
CEO, Cyber security, engineering development |
3 years |
two |
3 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Sep 18, 2026 @ 10:38 PM |
RAKESH KUMAR |
sangeetachilshetty@gmail.com |
http://AGSYNC%20 |
http://agsync.in |
+919108257239 |
RAKESH KUMAR CEO, SANGEETA CBO, DR SUSHANT CHIDRI PRODUCT MANAGER, ROHIT BANDE CTO AND still we do have 5+full time team on the same |
we are here from since the college days and then thought about the idea in the end was very curious on certain things and looked for the work and git the ideas with the curated problem and started working on the same |
two |
we do have 8 members in the team with the expers in the domain |
AGSYNC PVT LTD |
AGSYNC.IN |
KALABURAGI |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Sep 18, 2026 @ 9:41 PM |
George Johnson |
george.johnson@datamatrixai.co.in |
https://www.linkedin.com/in/georgejohnsonin/ |
http://www.datamatrixai.co.in |
7829440444 and 8847822146 |
Co - Founders |
We were together in the cohort of IISc GenAI and Prompt Engineering course and formed our company in April 2025. |
two |
The two co-founders complement each other having decades of experience in Technology and Buisness. George was P&L head in his role during corporate stint and Radha was in Delivery function. Together we are able to understand the customers pain areas and develop efficient AI based solutions and keep up with changing market needs. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Sep 18, 2026 @ 2:08 PM |
Umang Joshi |
umang@thewealthink.com |
https://www.linkedin.com/in/umangjoshi4/ |
https://www.thewealthink.com |
08625809781 |
Founder & CEO; 4 generation old family business in finance, Robotics engineering graduate, worked in tech industry for 10 years serving customers in India, UAE, and US, 3 years in Wealth Management, Ex-growth partner at KamayaKya Wealth Management (Pune) |
N/A |
one |
Deep finance and tech expertise, fast in shipping products and testing with leading PMS and AIF funds in Mumbai and PAN India |
Wealthink |
https://www.thewealthink.com |
Bengaluru |
Desktop app for private AI in Wealth Management |
Increasing coverage and throughput of investment research analyst in PMS and AIF firms |
A desktop app which computes locally with sovereign AI models for any investment research work and makes collaboration easy. People have been using AI chatbots in silos but that doesn't result and tangible ROI in the finance industry. We prove that cross team collaboration on a single platform is the way to actually increase the throughput for them. We are giving SMBs the technology to work like an institutional hedge fund in Wall Street. |
Domain specific AI harness which specifically solves financial tasks and keeps all session records locally on the machine which keeps compliance in check and annual SEBI audits can be 1 click activity. Which solves the biggest bottleneck for growth for our customers. |
Users |
one |
Medium size SEBI regulated PMS and AIF funds |
USD 200B |
NA |
NA |
SaaS + Managed Inference |
Chatgpt Work, Hebbia, Finpilot |
Warm connections so far, later it will be content and brand led acqusition |
Rolling out first product which works out of the box for individual analysts, then get pilots in firms where we have at least 4 users onboard and solve for their compliance, and repeat for every referral. A good product has strong word of mouth engagement among the peers in this industry. |
Build an organisation brain for wealth management industry. It will make the data and knowledge work cheaper, thereby encourage more people to take up a license and start business in this domain. True financial utopia can only be reached if the tax payer to money manager ratio is 1:500 at least like developed economies (unlike today which is 1:50,000 in India) |
Incorporated |
0 |
one |
Recommended by friend, funding to hire GTM and engineering team |
Connections into Karnataka wealth manager communities for early adopters |
Yes |
one |
one |
Opus 5, GPT-5.6, Posthog, Langfuse, Bun, Vite inference package, Pincone vector db, docker, mongo, AWS EC2, Azure codesign, Cloudflare CDN |
Stores client sessions in their machine and uses that for all future work |
Harness agnostic architecture which keeps the user flexible to use any model or agent based on their token budget. Our flagship product is very competitive in pricing compared to others with similar capabilities. |
We use langfuse and posthog for full content telemetry and evaluate on proprietary method to analyse the performance. |
All user data resides on their laptop. Nothing streamed to us. We will apply for SOC2 certification after getting some traction and funds. |
None |
An overhead will be to maintain multiple MCP and harness libraries across various versions of them to keep it operational for different customers. On international expansion we will have to setup data residencies in local economies which will add to dedicated devOps team. |
Yes, team has previously worked on model evals for meta and openAI |
NA |
NA |
Both |
|
|
What is the difference between the two? |
NA |
|
|
| Sep 18, 2026 @ 12:10 PM |
Saurabh Bassi |
saurabh@atgworldD |
https://www.linkedin.com/in/saurabhbassi/ |
https://www.interviewgod.ai/ |
9900036467 |
Saurabh Bassi — Founder & CEO: Saurabh leads IG's overall strategy, product vision and business development, bringing an engineering and enterprise technology background developed through building and operating Banao Technologies. Core Team: IG is supported by an 8-member full-time team operating across the broader Banao ecosystem, including IG, Banao, Vikaas and Vidya. The team brings complementary capabilities across product, engineering, sales, technology and operations, enabling IG to build, deploy and continuously improve its AI-powered hiring platform. The wider team is further supported by approximately 30–35 interns contributing across the group's different verticals. |
IG is founder-led by Saurabh Bassi and operates as part of the broader Banao ecosystem. The core team works across multiple verticals including Banao, Vikaas, Vidya and IG, bringing together capabilities across engineering, product, sales and operations. IG itself emerged from Banao's enterprise software and internal hiring experience and was subsequently productized as a dedicated hiring platform. |
one |
Our biggest strength is the combination of deep engineering capability, enterprise software experience and strong execution across product, AI, sales and implementation. IG is built on Banao Technologies' decade of experience solving complex enterprise technology problems, while the dedicated team brings expertise across AI/ML, product, engineering and enterprise sales. This allows us to build, deploy and continuously improve the product while staying close to real customer problems. |
InterviewGod Hirings LLP |
https://www.interviewgod.ai/ |
Bengaluru |
InterviewGod is the hiring decision infrastructure for enterprises. AI sourcing, structured interviews, integrity detection, and post-hire tracking — in one platform. |
Enterprise hiring decisions are often fragmented, inconsistent and difficult to audit. Recruiters and hiring teams spend significant time screening and interviewing candidates, while different interviewers may evaluate candidates differently and critical decision signals can be lost across the hiring process. This creates delays, inconsistent candidate evaluation and limited visibility into why hiring decisions were made. |
InterviewGod is an AI-powered hiring decision infrastructure platform that standardizes and automates key stages of enterprise hiring. It provides AI-powered call screening, structured assessments and AI interviews, with configurable evaluation criteria and interview integrity signals. The platform helps organizations turn candidate interactions into structured, comparable decision signals while keeping the final hiring decision with human interview panels. It also supports post-hire tracking to connect hiring decisions with outcomes over time. |
IG combines an AI-powered hiring decision layer with the implementation and enterprise technology capabilities of the broader Banao ecosystem. Unlike point solutions that address only one part of recruitment, IG connects screening, assessment, structured interviewing and integrity signals within one decision infrastructure. Its defensibility is strengthened by real-world exposure to diverse enterprise hiring workflows, configurable evaluation frameworks and the ability to combine product, AI and implementation expertise. Over time, accumulated hiring decision and outcome data can further strengthen the platform's intelligence |
Users |
Pilots |
Mid-to-large enterprises with 500+ employees, primarily across India, the GCC and the US, with significant or high-volume hiring requirements. Our primary buyers are CHROs, Talent Acquisition Directors, HR VPs, Talent Heads and CEOs who need to make hiring more structured, consistent, scalable and auditable. We focus particularly on enterprises in workforce-intensive sectors such as manufacturing, healthcare, financial services, retail/e-commerce, logistics and technology. |
NA |
NA |
NA |
IG follows a usage-based B2B SaaS model primarily priced per interview, with pricing varying by the hiring workflow and configuration. Customers can use individual AI screening, assessment or interview modules, or combine them into a complete hiring flow. We typically enter organizations through a paid pilot covering a defined role and candidate set, after which successful pilots can expand into recurring, higher-volume enterprise usage. |
HireVue, Eightfold, Mercer Mettl, iMocha, Paradox and Mercor. |
We acquire customers through a combination of targeted enterprise outbound, referrals, strategic partnerships, events and inbound digital channels. Our primary motion is targeted outreach to senior HR and business decision-makers within our enterprise ICP, supported by warm introductions through the broader Banao ecosystem. Prospects are typically taken through a consultative discovery process and converted through a focused paid pilot before expanding into broader hiring workflows and enterprise usage. |
Our go-to-market strategy is enterprise-focused and land-and-expand. We target mid-to-large organizations with significant hiring volumes and position IG as Hiring Decision Infrastructure rather than another point hiring tool. We enter through a focused hiring use case or paid pilot, demonstrate measurable improvements in screening and evaluation efficiency, and then expand into additional roles, hiring workflows and business units. Our initial priority markets are India, the GCC and the US, with the broader Banao ecosystem providing enterprise relationships, implementation capabilities and cross-sell opportunities. |
Our long-term vision is to become the intelligence and decision infrastructure layer for enterprise hiring. We aim to move hiring from fragmented, subjective processes toward structured, evidence-based and auditable decision-making. Over time, IG will connect hiring signals with post-hire outcomes, allowing organizations to understand which signals and decisions consistently lead to successful hires. Our goal is to become the system enterprises rely on not only to conduct hiring, but to continuously improve how they make hiring decisions. |
Registered — InterviewGod Hirings LLP, incorporated in India on 26 February 2026. |
0 |
|
We are applying to Avinya to accelerate IG's transition from early enterprise pilots to scalable adoption. We want to leverage Avinya's AI-focused mentorship, industry network and access to the IIT ecosystem to strengthen our product, refine our enterprise go-to-market strategy, deepen our AI and responsible-AI capabilities, and build stronger pathways to enterprise customers and strategic partners. The program's Bengaluru ecosystem is particularly relevant to us as a Karnataka-based AI startup. |
During the programme, we aim to strengthen IG's product and AI capabilities, convert early pilots into repeatable enterprise deployments, refine our pricing and go-to-market model, and expand our enterprise customer pipeline. We also want to establish stronger industry and investor relationships, validate our product across additional hiring use cases, and develop a scalable path for expansion across India and international markets. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Sep 17, 2026 @ 1:08 AM |
Parakh Sharma, Rajhans Hasmukh Sharma |
parakh@axalonsystems.com |
http://www.linkedin.com/in/parakh-sharma |
http://axalonsystems.com |
9413552887 |
Rajhans Sharma – Co-Founder & CEO: Full-time founder and hardware systems lead with 14 years of experience in high-voltage electrical systems and field engineering. Parakh Sharma – Co-Founder, AI & Flight Systems: Technical founder responsible for AI/ML, computer vision, autonomous flight systems and integration of ArduPilot, ROS/MAVLink and NVIDIA Jetson-based edge computing. Currently transitioning from part-time to full-time involvement. |
Relatives |
1 |
We have known and worked closely with each other for several years, combining complementary experience across electrical systems, software development, AI and autonomous drone systems. We began working together on AXALON to combine these capabilities into an integrated solar inspection platform. |
Axalon Systems pvt ltd |
axalonsystems.com |
Vadodara |
AXALON builds autonomous drone-and-dock systems for AI-powered thermal inspection of utility-scale solar PV parks. |
Utility-scale solar parks rely heavily on manual or outsourced drone inspections that are slow, infrequent and often provide raw thermal imagery without actionable analysis. As a result, hotspots, bypass-diode failures, PID degradation, string disconnections and other PV faults can remain undetected, causing generation losses and increasing O&M effort. Large solar parks also require repeatable inspection coverage that is difficult to achieve through conventional pilot-led operations. |
AXALON is an autonomous drone-and-dock platform designed for recurring thermal inspection of utility-scale solar parks. The drone launches from a dock, executes structured GPS-based thermal and RGB inspection missions, processes imagery using onboard AI, detects and classifies PV faults, and GPS-localizes defects for actionable maintenance. It then returns to the dock for charging. The system generates severity-ranked, GPS-tagged reports in PDF, Excel and GeoJSON formats, with outputs aligned with IEC 62446-3 requirements. |
AXALON combines autonomous drone operation, dock-to-dock charging, onboard edge AI, thermal defect classification, GPS-level fault localization and actionable reporting into one integrated system. The technology stack includes our own airframe integration, flight-control and mission-planning layer, AI models, radiometric thermal processing and inspection software. This end-to-end integration is designed to enable frequent, repeatable inspections without requiring a pilot or extensive manual image review, while maintaining an India-focused cost and deployment model. |
MVP |
Pilots |
Utility-scale solar park owners, solar developers, EPC companies, and solar O&M service providers. The initial focus is on large Indian solar parks where recurring thermal inspection can reduce inspection effort, identify faults earlier and improve asset performance. |
₹744 Crore/year |
₹514 Crore/year |
₹48 Crore/year |
AXALON follows a phased model. Initially, we offer Inspection-as-a-Service (DaaS) at approximately ₹2,500/MW for base inspection, with AI analytics as an additional ₹3,500–₹5,000/MW upsell. In the next phase, we will sell autonomous drone-and-dock platforms at approximately ₹15–20 lakh per unit, supported by recurring SaaS and analytics subscriptions. Hardware sales will also be complemented by training and deployment services for regional O&M contractors. |
Raptor Maps, Sitemark, Above Surveying, DJI Enterprise/DJI Dock, and conventional Indian solar O&M teams and drone-based inspection service providers. |
We plan to acquire customers through direct engagement with solar O&M operators and asset owners, pilot demonstrations, existing industry relationships, and partnerships with solar EPC and O&M companies. Successful DaaS pilots will be used to demonstrate measurable inspection and maintenance benefits and convert customers into recurring contracts. |
AXALON will initially enter the market through DaaS pilots with solar O&M operators and utility-scale asset owners. After validating the system and customer ROI, we will convert pilots into recurring annual inspection and AI analytics contracts. The next phase will expand through partnerships with solar EPCs and regional O&M contractors, followed by direct engagement with large solar developers. Over time, AXALON will offer both recurring inspection services and autonomous drone-and-dock platform sales with SaaS. |
Our long-term vision is to build an autonomous inspection and predictive-maintenance platform for utility-scale renewable energy infrastructure. We aim to make solar-park inspection continuous, repeatable and data-driven through autonomous drone-and-dock systems, onboard AI and asset-level inspection history. Starting with thermal PV inspection, AXALON plans to build a broader infrastructure intelligence platform that helps renewable-energy operators detect issues early, prioritize maintenance and improve asset availability and generation. |
Incorporated |
₹15L - Self Funded |
Yes |
We are applying to Avinya to accelerate AXALON from a validated technical prototype toward commercial deployment. We are looking for mentorship, industry connections, customer access and support in validating our autonomous drone-and-dock platform with solar asset owners and O&M operators. The programme can also help us strengthen our business model, regulatory roadmap and prepare for scaling across India's utility-scale solar market. |
Our primary goals are to complete and validate the autonomous drone-and-dock MVP, conduct 2–3 solar-site pilot deployments, validate AI-based thermal fault detection under real field conditions, and convert successful pilots into paying customers. We also want to establish our regulatory and operational roadmap, refine our DaaS and platform pricing, develop strategic partnerships with solar O&M/EPC companies, and prepare AXALON for scalable commercial deployment. |
Yes |
Yes |
Core engine |
AXALON uses an edge-AI computer vision pipeline for automated solar PV thermal fault detection. Radiometric thermal and RGB images are captured during structured drone missions and processed onboard an NVIDIA Jetson Orin Nano. A trained object-detection model identifies and classifies PV anomalies across 11 fault classes. The detected faults are combined with GPS/mission data to localize defects and assign severity levels (CRITICAL/HIGH/MEDIUM/LOW). An LLM-assisted classification layer is being developed to support contextual defect interpretation. The resulting structured data is used to generate actionable PDF, Excel and GeoJSON inspection reports. |
Our data advantage is the combination of radiometric thermal imagery, synchronized RGB imagery, GPS coordinates and solar-panel-level fault annotations collected specifically for PV inspection. As deployments increase, these field datasets will enable continuous retraining and improvement of our defect-detection models across different panel technologies, environmental conditions and fault types. The platform is also designed to retain the underlying per-pixel temperature information rather than only storing processed detections. |
The defensibility comes from the combination of domain-specific thermal datasets, proprietary fault annotations, our trained PV defect-detection models, and integration with the autonomous flight and inspection workflow. Rather than providing a standalone AI model, AXALON connects image acquisition, edge inference, fault classification, GPS localization, severity triage and reporting into a single system. Increasing field deployments will further strengthen the proprietary dataset and model performance. |
We currently evaluate the thermal defect-detection model using mAP@0.5 and mAP@0.5:0.95, along with precision, recall and class-wise performance. Our current PoC achieved 0.792 mAP@0.5 and 0.669 mAP@0.5:0.95. As the system moves into field pilots, we will additionally measure false-positive/false-negative rates, localization accuracy, inference latency, temperature measurement accuracy and performance across different environmental conditions. |
AXALON is designed around edge/on-premise processing so thermal and RGB inspection data can be processed without requiring cloud transmission. Customer inspection data remains under the customer's control. Access to inspection data and generated reports will be restricted to authorized users, with secure storage and controlled access implemented as the platform scales. The system is also being developed with applicable Indian drone regulations and IEC 62446-3-aligned inspection requirements in mind. |
Yes. Drone operations in India are subject to DGCA requirements, including applicable requirements for drone registration/UIN, Remote Pilot Certification, operating permissions and insurance. BVLOS operations introduce additional regulatory requirements and approvals. These requirements could affect deployment locations, operating procedures and the pace of commercial rollout. AXALON is therefore incorporating regulatory compliance into its deployment roadmap. |
At 10x deployment scale, the primary challenges would be managing larger and more diverse thermal datasets, maintaining model performance across different panel types and environmental conditions, and scaling model retraining and validation. Edge inference capacity and data-quality monitoring would also need to scale with the number of inspection missions. Our architecture mitigates cloud dependency by performing inference at the edge, while centralized systems can be used for model versioning, monitoring and continuous improvement. |
Yes. The AI/ML function is led in-house by Parakh Sharma, Co-Founder – AI & Flight Systems. Expertise includes computer vision, thermal image analysis, AI model training and evaluation, autonomous drone systems, ArduPilot, ROS/MAVLink and NVIDIA Jetson edge inference. The team is developing PV-specific defect detection and classification models and integrating them directly with the autonomous inspection workflow. |
We are currently using internally collected and annotated solar PV thermal/RGB inspection data for model development, evaluation and validation. The PoC dataset includes annotated PV fault imagery used to train and evaluate the defect-detection model. As the platform moves into field pilots, customer-generated inspection data will expand the dataset subject to customer agreements and applicable data-use permissions. Any third-party datasets used for development will be used only under their applicable licensing terms. |
We will continuously improve the models through a field-data feedback loop. Inspection data and model predictions from deployments will be reviewed, difficult and misclassified cases will be added to the training set, and models will be periodically retrained and validated against held-out datasets. Performance will be monitored by fault class and operating conditions to identify model drift. Customer/O&M feedback will also be incorporated to improve fault prioritization and reduce false positives. |
AXALON is initially focused on the Indian utility-scale solar market, where large solar parks require frequent and cost-effective inspection. Once the platform is validated and commercially established in India, we plan to expand to international solar markets with similar inspection and O&M requirements. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/6j58DYuecXxZ-Axalon-Systems_Pitchdeck_WOA_2026.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/6j58DYuecXxZ-Axalon-Systems_Pitchdeck_WOA_2026.pdf |
https://drive.google.com/file/d/1d6zV78BA2LVMoqFOSekIZLKKbKGm5Ilc/view?usp=sharing |
AXALON is an impact-focused clean-energy technology startup working to improve the efficiency and reliability of existing solar infrastructure. Undetected PV faults can reduce electricity generation and lead to avoidable energy losses. By enabling frequent, autonomous thermal inspections and early fault detection, AXALON helps solar operators identify and prioritize maintenance before faults result in larger generation losses. This can increase energy output from existing solar assets without requiring additional land or generation capacity, supporting more efficient utilization of renewable-energy infrastructure. |
NA |
IIT Bombay |
checked |
| Sep 16, 2026 @ 4:48 PM |
Soma Bhattacharjee |
Soma@mahimahi.in |
https://www.linkedin.com/in/somabhatta/ |
https://www.mahimahi.in/ |
9739752792 |
Soma is a CEO & Co-founder; heads product and marketing. Comes from advertising and product marketing background. Naji is COO & Co-founder, heads sales and operations. Comes from Operations and Digital Marketing background. |
We were colleagues and have been working together from 5 years |
All |
Our biggest strength as a team is that we understand this space. We have the domain knowledge of marketing and have been working with AI tools and martech stacks from a long time |
Mahi Mahi Tech Solutions Pvt. Ltd |
https://www.mahimahi.in/ |
Bangalore |
Mahi Mahi is a AI native marketing services company that has its own product at its core |
Marketing has been a cost center but nowadays it is increasingly becoming responsible for ROI and revenue. Which is why, the teams are becoming leaner and marketing budgets are not growing. Teams are asked to do more with less. We are solving the resource part of the problem while giving the marketers visibility of their ROI. |
mAIVen is a marketing operating system for lean marketing teams. It has different agents configured as a marketing team to make end to end marketing seamless on one platform. No broken workflows. No context lost. Only trackable marketing initiatives. And measurable ROI. mAIven can analyse, strategise, and build and execute brand-driven campaigns and content. |
Market is filled of individual specialist tools to support marketing individuals. No platform is built to service the entire marketing team. Marketing has to function like a system. mAIven is a system for that. |
MVP |
Pilots |
Mid Enterprise with $100-500M revenue |
The U.S. gross universe is ~199.7K companies in the $10M–$500M band. |
~40% of total institutional demand is for AI branding: content, campaigns, brand monitoring is a $3.29B market in 2025 → $6.45B by 2030. |
we achieve 5% penetration at $999/month, the ACV is $11,988 and the resulting Year 3 ARR is approximately ~$120M ARR. |
AISaaS |
WPP ONE, JustAI, Omega from Power Digital Marketing |
Founder led Sales |
B2B Sales |
Making a marketing platform for the world from India. Every marketing tool ever existed will be a feature in mAIven |
Pvt Ltd incorporated in 2023 |
Bootstrapped |
Yes |
GTM, Warm introductions |
understand PMF better |
virtual |
Yes |
Core engine |
Maiven combines a marketing analytics dashboard with an AI workspace for strategy, content and creative production. There are three connected parts to the architecture.
The dashboard brings together GA4, Search Console, competitive search data and first-party pixel events. Deterministic rules compute structured findings; Gemini on Vertex AI turns those findings into analysis and recommendations with evidence references. The model interprets measured data rather than calculating the underlying metrics.
The strategy pipeline uses Google's Agent Development Kit. It combines dashboard findings, brand and project context, retrieved documents and external research to produce a seven-section brief covering strategic foundations, positioning, topic authority, research, content ideas, distribution and quality assessment.
Content and creative workflows use LangGraph specialist agents for outputs such as blogs, emails, brochures, images, posters and presentations. They collect requirements, use brand context, generate outputs and support refinement. Retrieval-augmented generation, or RAG, supplies relevant passages from documents and websites at generation time.
The stack spans Python/FastAPI, React/TypeScript, Vertex AI, PostgreSQL, MongoDB, Firestore and Google Cloud Storage. The dashboard uses Render and Redis; the Maiven backend uses Google Cloud Run. Different models serve text, images and embeddings. |
Our data advantage is the customer-specific context we bring together across measurement and creation. The dashboard connects traffic, search demand, competitive visibility and journey signals, brand guidelines, product information, uploaded documents and the history of content and creative work.
This gives the system context about how a brand performs, what it stands for and what the team is trying to produce. The strategy pipeline can use measured findings alongside brand knowledge and research, while specialist agents use that context to create relevant outputs.
Customer data remains the customer's; third-party information retains its own rights and licensing terms. We do not claim an exclusive foundation-model training corpus. The advantage is our data model, retrieval and workflow integration. Linking approved outputs back to later performance is a further opportunity, rather than an automatic learning capability we claim today. |
The defensible work is the marketing system built around the models: the dashboard's rules and cross-source findings, the strategy pipeline's structured reasoning, and specialist workflows that carry brand and project context into usable content and designs.
We have encoded how a brief moves from business goals and evidence to positioning, topic selection, content planning and distribution. Content-to-creative handoffs carry requirements and copy forward, reducing repeated briefing. Brand knowledge, connected accounts, projects and saved assets make the workspace more useful as a customer uses it.
Evidence handling and quality checks are also part of that system. Dashboard citations reference measured findings; strategy and creative output need additional checks for source support, brand alignment and practical usefulness. These are engineering and domain assets we can improve independently of any one model, although changing providers still requires integration work and evaluation. |
We use Langfuse to trace Maiven's strategy pipeline and record execution latency, token usage and evaluation scores. Automated checks assess output completeness, schema validity, brand consistency and whether dashboard evidence reaches the relevant agents.
An editorial evaluation agent scores brand alignment, strategic relevance, evidence strength and execution feasibility. We combine these checks with regression tests and human review; AI-generated scores support review rather than prove factual accuracy. Our evaluation framework also includes mocked regression tests and live Vertex AI scenarios for content, creative generation, routing and retrieval.
For the dashboard, automated tests verify deterministic findings, while narration checks validate output structure and finding references. A valid reference helps trace a claim; it does not by itself prove that the wording is correct. Our next step is a representative benchmark with human ratings for factual grounding, brand fit, visual quality and usefulness, supported by consistent latency and cost tracking. |
Security has to cover analytics, uploaded knowledge, generated assets and the services connecting them. The dashboard uses database row-level security for customer isolation and verifies login tokens. Its pixel supports consent modes and minimizes identifiers, including a rotating daily visitor hash. Dashboard event deliveries use signed messages.
Maiven also needs access control across PostgreSQL, MongoDB, Firestore and Cloud Storage. Its workflows use authenticated user, brand and project context, including ownership checks for content handoffs and scoped retrieval. These controls have to be tested across each store; dashboard database isolation alone does not establish isolation for the full product.
Privacy work must cover uploaded documents, website extracts, embeddings, generated content, connector credentials and AI traces, including retention, deletion and third-party processing terms. We use customer context for retrieval and generation rather than training any model. The security review and remediation described for the dashboard are part of the wider assurance work; the planned independent penetration test should cover both products and their integration. |
Yes. Across Maiven, the main areas are data protection, tracking consent, intellectual property, AI transparency and platform dependency.
Analytics, pixel events and uploaded documents bring data-protection obligations, including GDPR and India's DPDP framework where applicable. We need clear processing roles, agreements, retention and deletion processes, and controls over subprocessors and cross-border transfers. Applicable requirements and commencement dates need to be assessed for each market and use case.
Content and image generation add copyright, trademark, likeness and advertising-claim risks. Website access and external research also carry source-specific rights and terms. Human review remains important before publication.
Generative and interactive AI can have transparency obligations under the EU AI Act, with requirements depending on our role and the output. Human oversight does not remove those obligations. Google APIs, model services and research providers also create operational and contractual dependencies.
SOC 2 is a separate commercial assurance requirement. We do not hold a SOC 2 report today. Preparing for it requires evidence of controls across both products, including access reviews, change management, incident response and tested recovery.
Regulatory references: European Commission AI transparency guidance | MeitY DPDP Rules and enforcement timeline |
The likely pressure points are ingestion volume, concurrent generation, model quotas, rendering and database capacity. A tenfold increase in customers does not produce a single predictable bottleneck; the mix of analytics traffic, briefs and creative jobs matters.
On the dashboard, ingestion has already needed narrower rebuilds to avoid reprocessing full history. Scheduling, pixel writes and event delivery need load testing as volumes grow. Cached narration reduces repeat model calls when the underlying findings are unchanged.
On Maiven, strategy briefs involve multiple dependent agents, research calls and retrieval. Creative work adds image generation, rendering, storage and export costs. More simultaneous jobs can amplify provider rate limits, timeouts, memory use and spend. In-process execution also needs a clear recovery path if an instance stops during a long job.
The priorities are measured concurrency limits, durable job execution where needed, retry and deduplication controls, quota planning, database connection management, and tested backups and recovery across stores. Token and credit tracking should be reconciled with actual provider costs. We should validate capacity under representative workloads before claiming readiness for 10x scale. |
Yes. We have a three-member in-house AI/ML team. Our AI work covers Maiven's strategy, content and creative workflows, together with the dashboard's intelligence layer. The focus is applied AI engineering: orchestration, retrieval, evaluation and reliable integration of foundation models with marketing data and workflows. |
We use several types of data across the product. Customer-authorized GA4 and Search Console data, licensed competitive search data and first-party pixel events support the dashboard. Brand profiles, goals, guidelines, product briefs, uploaded documents and reference assets support Maiven's strategy and generation workflows.
Maiven also retrieves website content and uses external research services. Relevant document and website passages can be stored and indexed for retrieval, including embeddings in Firestore. Generated briefs, content, designs and task history are additional application records. These categories have different ownership, licensing and retention requirements; public availability does not imply unrestricted reuse.
We use these inputs for retrieval and inference rather than training or fine-tuning our own foundation model. Customer and third-party rights continue to apply to the source material. Customer content should remain scoped to the authorized workspace, and model-provider retention and training policies must be assessed against the provider terms and the configuration actually in use. |
We will build on the evaluation framework already present. First, expand it into a representative set of dashboard questions, strategy briefs, content tasks and visual tasks, including incomplete data, conflicting sources and difficult brand requirements.
Second, combine automated checks with human ratings. Track factual support, brand alignment, source quality, strategy usefulness, visual quality, task completion, latency and cost. Reviewers' accept, edit and reject decisions can inform evaluation cases and product improvements; they do not automatically imply model training.
Third, improve upstream context: dashboard findings, document ingestion, retrieval relevance, brand information and the handoff into agent prompts. Traces help show whether the right evidence reached each stage. Then compare prompt, workflow and model changes against the same scenarios before release.
Over time, connect approved outputs with observed marketing performance to test which recommendations help. Treat outcomes carefully: changes in traffic or conversions can have several causes, so they are not automatic proof that an AI recommendation worked. |
Global market first, and then Indian market when it becomes self-serve |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Ryl6ORTgjmN7-mAIven-OS-Investor-Deck-2-compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Ryl6ORTgjmN7-mAIven-OS-Investor-Deck-2-compressed.pdf |
https://canva.link/70dikcai4138wmd |
An mission-driven start-up. We want Mahi Mahi/mAIven to become that one Indian start-up that becomes synonymous to marketing across the globe. All marketing tools are from the global market adopted in India. We want to reverse the script. Most product-driven brands in India services Indian market. We have decided to do the other way round. |
NA |
Came in my GEO search on accelerator programs |
checked |
| Sep 16, 2026 @ 8:44 AM |
Siva Kumar |
siva_k@hotmail.com |
https://www.linkedin.com/in/sivakumartm/ |
http://NA |
7348897719 |
Siva Kumar - Founder, Chief Product Architect & Principal backend developer, Vishnu - UX Architect, Shopify Plugin Developer, Manesh - Front End Developer, Nishant - AWS/Cloud & Devops, Ahammed Raqueeb - Marketing |
Most of the team were hired into Pivotrics over the past 2 years |
one |
We are domain insiders applying AI — not AI generalists discovering retail. Our founder has spent over 30 years building payment and promotion systems for large retailers — multi-lane chains and petroleum/convenience networks — mostly in advanced markets like the USA. These are systems that process enormous transaction volumes every day, where a slow checkout or a failed promotion directly costs the retailer money. Engineering at that level is complex and unforgiving, and it teaches you things no framework or AI coding tool can substitute for: what to build, what will break at scale, and what a merchant will actually trust.
But we are not just technologists. Decades of working alongside retailers has made us deeply business-focused — we think in the merchant's terms: margins, footfall, repeat customers, promotion budgets. And it has given us genuine empathy for the people we build for. We've worked directly with the retail teams who live with these systems daily, seen what frustrates them and what actually moves their business. That empathy is what turns technology into a meaningful product rather than a clever demo.
So our strength is the combination: deep retail domain knowledge, proven engineering at transaction scale, business focus, and real understanding of our end users — merchants. We adopt ML and LLMs practically, where they genuinely help the merchant, not because they're fashionable. We know this domain, we've built for it at the highest level, and we understand the people we're building for. The founder is also the author of the book called "The Pivot Model" - which helps to build high-performance product teams in the age of AI. |
perkqi |
na |
bangaloer |
PerkQi is an AI-native SaaS platform for merchants to author promotions in plain language and deliver them to consumers with deep personalization, across all sales channels. |
Merchants spend 20–25% of revenue on promotions, yet they lack the intelligent tools required to author, manage, and deliver promotions to customers across all channels — e-commerce, POS, agentic sales, and more. The tools available to them today fail in three ways. (1) They work post-mortem — analyzing past data, with no awareness of a customer's live intent while they are still shopping and still persuadable. (2) They are rigid — pre-fixed workflows and templates that leave no room for the merchant's own intuition and understanding of their customers' emotions, which is often their sharpest asset. (3) The tools that do offer real flexibility are enterprise solutions which are expensive, and dependent on developers to configure every promotion. So the merchant who best understands their customers is the one least able to act on that understanding — at the one moment it matters, while the customer is still deciding. |
PerkQi uses AI to remove these technology boundaries entirely.
Authoring, freed from templates. Merchants describe the promotion they want in plain language — however specific or layered: "anyone who's bought from us three times this year and spent over $400 on dresses — when they're browsing coats, give them 15% off and 500 bonus points, weekends only, first 200 customers." PerkQi turns it into live, running logic in minutes. No developers, no templates, no tickets. The merchant's intuition finally has somewhere to go.
Delivery, in the live moment. This is the piece that has been missing. PerkQi analyzes each customer's present activity — what they're browsing, what's in their cart, their history and loyalty standing — scores every eligible promotion against that individual shopper, and delivers the one they're most likely to respond to, right while they're still deciding. The same engine serves every channel: e-commerce, POS, and emerging agentic sales, so one promotion authored once reaches customers everywhere they shop.
Intelligence that protects the merchant. Every offer is checked against the merchant's margin limits before a customer ever sees it, and PerkQi ranks offers by the profit they earn — not just the conversion they win. Each cycle feeds back into the merchant's own model, so promotions get sharper with every campaign, on the merchant's data alone. |
Promotion and loyalty solutions have existed for two decades, but they are legacy systems — and very few modern players cover both e-commerce and traditional POS in one platform. That gap leaves the space open for democratization, especially for small and medium merchants who have already gone digital through platforms like Shopify, and it makes PerkQi a disruptive modernization path for established retailers still on legacy systems. Merchants already run personalized marketing themselves on social media; nobody has given them that same self-serve, AI-driven power over promotions. We are that player. Our defensibility rests on two moats. First, the technology bar: real-time, per-shopper promotion decisions across e-commerce and POS demand high-volume transaction engineering — a craft our founder spent 30+ years building for the world's largest retailers, and one that AI generalists and coding tools cannot shortcut. Second, the data: operating the full loop — from authoring to delivery to the customer's response to the sale — accumulates per-merchant data on which offers were shown and what happened, data no other tool in the stack ever sees. Each merchant's model compounds on their own history, making PerkQi harder to replicate and harder to leave the longer it runs. |
Users |
Pilots |
Our initial target market is small and medium-sized Shopify e-commerce stores globally. Shopify is the ideal beachhead: millions of merchants worldwide, a single integration that reaches all of them, self-serve onboarding with no sales team required, and merchants who already spend heavily on promotions but are limited to basic discount tools. Our app is already live on Shopify. From there, we expand along two lines: upward to larger and enterprise retailers who need the same intelligence at greater scale and deeper integration, and outward to physical channels — POS and in-store — plus emerging agentic sales channels, where the same engine serves every touchpoint. This progression mirrors how merchants themselves grow: we land where they start, and expand as they do. |
$15B+ global market, projected to double by 2031 |
~$3.5B ARR |
~$18M ARR |
Monthly fee ranging from USD.150 to USD.1500 |
Promotion Authoring (Talon.One, Voucherify, Eagle One, Capillary) + Delivery (Agentis + Nibble) |
Tie up with B2B Sales Channels |
We have ready integration with Shopify - Initial focus on Shopify stores, before moving to retail POS |
We want to become the Promotion Management System for a good percentage of Merchants/Retailers across the globe. |
Currently in Progress |
NA |
one |
Converting a working product into a growing business is a huge challenge — and that is exactly where we are. Our immediate goal in joining Avinya is to strengthen our go-to-market strategy, and we believe the association will open opportunities across multiple dimensions: structured mentorship from people who have taken products to market, access to the IIT alumni and investor network, and the Bengaluru retail-tech ecosystem around the programme. |
Our goals for the 16 weeks are concrete: (1) Paying merchants. Convert our early pilots into referenceable, paying customers on Shopify, and establish a repeatable self-serve acquisition motion — our target is a clear week-on-week growth curve we can present at Demo Day. (2) Go-to-market sharpened by mentorship. Pressure-test our pricing, positioning, and merchant onboarding with mentors who have sold SaaS to businesses — we know retail deeply; we want to compress our learning curve on selling to it at SMB scale. (3) Partnerships and channel. Use the ACSEL and IIT alumni network to open conversations with retail ecosystem players — POS providers, D2C enablers, and potential pilot merchants — that would take us months to reach alone. (4) Investment readiness. Leave the programme with the metrics, narrative, and network to raise a strong seed round, with Demo Day as the launch point. |
Yes — we are open to hybrid participation and will attend all key in-person sessions in Bengaluru, participating virtually as required. |
one |
one |
PerkQi's AI architecture has two layers:
Layer 1 — Promotion authoring (LLM-powered). Merchants author promotions in plain English. As they do, PerkQi feeds rich analytics from the merchant's own store — customer segments, purchase history, past promotion performance, reward effectiveness — into the LLM's context, so the drafted promotion is grounded in the merchant's actual data rather than generic templates. The LLM translates the merchant's intent into precise, executable promotion logic, and validates it before it ships.
Layer 2 — Personalized delivery (ML-powered). Once promotions are live, machine learning models learn continuously from how customers receive and use them — which offers were shown, engaged with, redeemed, or ignored. These ML scores drive real-time delivery: for each customer, the system scores every eligible promotion and serves the one they are most likely to respond to, personalized to that individual at that moment.
The two layers form a closed loop: the delivery layer's learnings feed back into the analytics that ground the authoring layer, so every promotion cycle makes the next one sharper. Each merchant's models are trained exclusively on their own data — never pooled across merchants. |
Data is largely that we accrue from transactions |
Plain-English authoring alone is not defensible — anyone can put an LLM in front of a form. Our defensibility is what we feed it. PerkQi is end-to-end: it authors, delivers, observes the customer's response, and processes the sale. That full chain yields data no single-point competitor ever sees — which offers each shopper was shown, engaged with, redeemed, or ignored — and that is the richness we inject into the LLM's context at authoring time. It compounds with every cycle, per merchant, on their own data, and it cannot be bought or scraped. |
The most meaningful measure of our AI is business outcome: whether promotions designed from the merchant's past data actually improve conversion and deliver a tangible lift in cart size and overall revenue. That is the target we hold both AI layers to — the LLM authoring layer is judged by whether its data-grounded promotion designs outperform what merchants ran before, measured against non-promoted baselines.
Beneath that sit operational measures: prediction accuracy of the personalization models (did the customer engage with and redeem the offer we scored highest for them), offer engagement and redemption rates, and standard reliability metrics — decision latency, uptime, and a logged reason for every decision. Merchants review and approve every promotion before it goes live, which gives us a continuous human check on authoring quality. |
Data minimization: PerkQi only touches the data needed to personalize offers—profiles, purchase history, loyalty tiers, and active browsing—sourced directly from the merchant with explicit authorization. We never store or handle payment credentials; checkouts stay entirely within the merchant's existing payment infrastructure.
Per-merchant isolation: Every merchant’s data is strictly ring-fenced. Models are trained only on that specific merchant's data—never pooled across stores or used to train shared systems. If a merchant leaves, their data is deleted.
Compliance and traceability: As a data processor, we support core GDPR requirements, including user access, export, and deletion requests. We are also aligning with India’s DPDP Act as we expand locally. Every personalization decision is logged for a clear audit trail.
Security: All data is encrypted in transit and at rest, backed by security protocols developed from years of building high-volume retail transaction systems. |
Consumer-protection rules are tightening fast, but instead of treating regulation like an afterthought or a looming headache, we’ve used it as a blueprint from day one.
Fair, discount-only pricing. Algorithmic pricing puts regulators on high alert, especially in regions like the EU where automated decision-making triggers strict disclosure rules. PerkQi solves this by design: our system only discounts down from a merchant’s posted price. No buyer is ever charged above list price. That single rule takes the biggest consumer harm off the table, giving us a clean, defensible story for any audit.
Zero fake urgency. Scrutiny on "dark patterns"—like artificial countdown timers or bogus low-stock warnings—is peaking. With heavy penalties in play (like the UK’s DMCC Act fining up to 10% of global turnover and ongoing FTC enforcement in the US), an AI generating shopper messaging carries real liability. We handle this at the data layer: every claim our system makes relies on live, verified data (actual stock levels, true expiration windows, real redemption numbers), with every decision logged and traceable.
Compliance built to travel. Handling customer data across borders means GDPR is table stakes for us today, and India’s DPDP Act is already mapped out as we expand locally. As dedicated AI frameworks roll out—such as the EU AI Act’s transparency mandates—our built-in merchant approval flows and audit trails keep us ahead of the curve.
The bottom line: stricter enforcement works in our favor. When regulators crack down on deceptive marketing hacks, they make the sloppy legacy tools untenable—leaving room for platforms that actually built for transparency from the start. |
Keeping LLM costs flat: Token burn naturally climbs as more merchants use the system, so we keep authoring sessions lean through aggressive context caching and precalculated analytics. More importantly, we never put an LLM in the live checkout or browsing path—shopper delivery is handled entirely by lightweight ML scoring.
Serving traffic spikes without lag: Handling a 10x surge in real-time scoring comes down to smart precomputation and caching. It’s the exact engineering playbook our team relied on for decades building high-volume payment infrastructure.
Scaling per-merchant models: Training separate ML models across thousands of stores is an orchestration and pipeline challenge, not an unproven research problem. It just requires solid operational discipline to automate.
Guaranteed correctness: Even under peak loads, hard guardrails like merchant budgets and discount caps are enforced deterministically, not probabilistically. Surging scale might put pressure on speed or infra costs, but it will never compromise system correctness. |
We do not have in house team. We used friends from outside the organization to design the basic framework. |
We operate on merchant's data. They will have to agree and allow us to use thier data. But we strictly do not use one merchant's data for another merchant's benefit. Tenant isolation is strict. |
Two aspect - our current ML scores need to be tested in large merchants for their accuracy. That is the immediate goal. As we learn more, we might need to fine tune the algorithms and math at one level, or even change the parameters at later stage. |
We are targeting the global market. Our strength comes from having been in the global market. |
|
|
|
|
|
|
| Sep 16, 2026 @ 8:21 AM |
Srinivas Pai, Krishna Kiran |
srinivas@goodhart.app |
https://www.linkedin.com/in/srinivas-pai |
http://goodhart.app |
8088037335 |
Srinivas Pai - CEO , Early at BYJU'S and Unacademy. 10 years across Amazon and PhonePe; led Ops at Lokal (YC S19)./ Krishna Kiran - CPO -8 years in ethical product design; exPrincipal Designer, Infosys & ZEUX Innovation. NIT-Calicut, IIT-Bombay |
Through mutual friends/ since 2023 |
All |
Complementary execution and approach. Srinivas understands logic and analytics, operations and efficiency, sales and customer experience. Krishna understands product and design, marketing and branding, ethos and impact. Also when Srinivas poses a visionary challenge, Krishna looks for practical approach to fulfill it. |
ViVidVed Technologies| Goodhart |
goodhart.app |
Mangaluru |
Goodhart measures how students think, not what they finish, giving schools and coaching centres in Karnataka a per-seat tool that helps teachers teach sharper and students learn faster. |
Ask a student in Karnataka how their test went and you'll get a number. Ask them what they actually understood, and they go quiet. That silence is the whole problem. Somewhere along the way the exam score stopped telling us anything real. A kid can score seventy percent by guessing well and cramming the night before, and nobody knows the difference, not the teacher, not the parents, not even the student. The number looks fine, so everyone moves on, until the board exam shows up and finds out what the score was hiding. Teachers know this better than anyone. A good teacher can tell when a student is guessing and when they truly get it, but you cannot hold forty students in your head at once, so teachers teach to the middle and hope, and the kids who are quietly lost stay lost. The result is a system optimising for a number that has come loose from the thing it was supposed to represent. Students cram for scores instead of learning, teachers teach blind, and the gap between looking prepared and being prepared only shows up when it is too late to close it. |
Goodhart measures how students think, not just what they finish. During every quiz we capture what a score throws away: how long a student took, whether they changed their answer, where they re-read. These signals sort every attempt into one of four states: mastery, effortful understanding, guessing, or confusion.
For students, that means an honest picture of where they stand, so they stop studying everything and start fixing what matters. For teachers, it means walking into class already knowing who is stuck and on what.
No other product does both. Student apps ignore teachers. Teacher tools ignore how students learn. Goodhart is built for both sides of the same classroom. |
Every other product measures what a student got right or wrong. Goodhart measures how they got there, reaction time, answer changes, re-reads, mapped into four states: mastery, effortful understanding, guessing, or confusion. No other platform in India makes that distinction. We solve both sides of the classroom at once. A teacher who cannot see the gap and a student who cannot name it are the same problem, and our behavioral model is psychologist-validated, not guessed. Our moat compounds with use: every quiz attempt sharpens cohort-relative benchmarks that a competitor cannot shortcut without years of real data, and once teachers build lesson planning around our dashboards, switching means losing a live picture of student understanding, not just scores. We also run on Gemini Flash over cheaper Chinese alternatives, since Indian institutions will not hand behavioral data to a vendor with sovereignty questions, which is why DPDP 2023 compliance is core to institutional trust, not a checkbox. |
two |
one, Pilots |
B2B - Institutions, coaching centers, (11th, 12th, Science, PCM-B, entrance prep)/ B2C - Parents, students |
8800 Cr |
2249 Cr |
223 Cr |
Subscription |
Fermi.ai, Prolearn, yolearn |
Offline strategy |
Founder led onboarding of South Kanara institutions. Post traction expand to education hubs of India |
Every quiz on Goodhart generates a learning event: hesitation, doubt, re-reads, reaction time. One student's data tells you about one student. Millions of these events tell you something India has never built: a foundation model of how students actually learn.
That is our vision. Not just a product for Karnataka classrooms, but a Student Behaviour Foundation Model trained on real learning journeys, one that sees mastery, effort, guessing, and confusion in a way no score ever could. Our goal is 1 crore students and 10 lakh teachers across India, aligned with Viksit Bharat 2047. A developed nation needs a generation that understands what it learns, not just one that passes. |
LLP |
NA |
one |
Mentorship, Investor Connects, Government focus, Strategic connects, Market access, Funding |
Market penetration, government aided |
Yes |
one |
one |
Frontend is a React, TypeScript and Vite PWA with an Android kiosk build; Supabase (Postgres, row-level security, edge functions) handles data and auth. Telemetry from ten learning event types feeds a pipeline producing eight cognitive signals. On top sit nine model families: IRT for ability and item calibration, Bayesian Knowledge Tracing for concept mastery, LSTM-based Deep Knowledge Tracing (PyTorch) for trajectory and decay, K-Means and DBSCAN for learning personas, Hidden Markov Models for latent cognitive states, XGBoost and LightGBM for risk and exam readiness, per-student behavioural embeddings, and Graph Neural Networks over a student-concept-question graph. Three LLM personas (Charles for student coaching, Albert for teacher pedagogy, Eric for institution and parent reporting) run on hosted inference APIs today, with self-hosted serving planned as volume grows. |
Our advantage is behavioural, not content. Question banks and syllabi are commodities. What no one else holds is the interaction substrate underneath them: hesitation intervals, answer-change sequences, hint-to-retry latency, revision timing and session decay curves, captured continuously across the two years of the Indian Class 11 and 12 science stream. Because teachers author questions inside the platform, we also own the mapping between item, concept and observed behaviour, which is what makes mastery estimates calibrated rather than generic. This corpus compounds: every cohort sharpens the priors for the next, and it cannot be bought or scraped, only accumulated through sustained classroom deployment |
Three layers. First, the data asset above can only be accumulated over time, and improves every term. Second, the architecture is multi-model and multi-modal rather than a prompt layer on a foundation model. A competitor can copy the interface in a quarter; replicating calibrated IRT parameters, per-concept BKT priors and a trained knowledge graph requires longitudinal behavioural data they do not have. Third, distribution. We sell into institutions, where switching cost is high once teachers have built question banks, administration runs on our reporting, and parents receive updates through us. The long-term position is the Student Behavioural Foundation Model: whoever accumulates the sequence corpus first owns infrastructure others build on. |
Per model family. IRT is checked on parameter stability across cohorts, item fit statistics and test information functions. BKT and DKT are evaluated on next-question prediction using AUC and RMSE, with calibration curves and Brier scores, since a confident wrong mastery call is worse than an uncertain one. Risk models are scored on precision at k, because teachers act on a ranked list of roughly ten students, plus recall on students who subsequently underperform. Clustering is assessed on silhouette score and on whether personas hold stable term to term. Everything is backtested against held-out cohorts and validated against the real outcome signal, actual test performance weeks later. We monitor drift on feature distributions and calibration, with alerting beyond threshold. |
Our architecture is built around the obligations in the DPDP Act 2023. Processing rests on consent collected at onboarding with notice under Sections 5 and 6, security safeguards and fiduciary duties under Section 8, and the child-specific protections under Section 9, with parental consent flows for students under 18 and institutional consent where the college is the fiduciary and Goodhart the processor. Rights of access, correction and erasure under Sections 11 and 12 are supported in-product, with a grievance route under Section 13.
Technically, data is minimised at capture. The webcam attentiveness signal is computed on-device and discarded, with no video or image stored, and no biometric or location data retained. Row-level security enforces strict isolation, so a teacher sees only their classes, a parent only their child, an institution only its own students. Data is encrypted in transit and at rest, hosted in-region, with role-based access and audit logging.
On certification, we are compliant with DPDP by design today, and we are formalising that into recognised third-party attestation on a defined sequence. SOC 2 Type II and ISO 27001 are next, since those are what institutional buyers and investor due diligence actually ask for at our stage. ISO 27701, the privacy extension to ISO 27001, and ISO 42001, the AI management standard for systems that make automated decisions about a person, follow as our AI review pipeline moves toward the near-autonomous phase. ISO 9001 sits further out, tied to our human-in-the-loop content pipeline, and becomes relevant once we're operating at institutional-tender scale rather than pilot scale. |
Our users are minors, so the DPDP Act 2023 applies in full, including verifiable parental consent under Section 9(1) and restrictions on behavioural monitoring under Section 9(3). The DPDP Rules 2025 exempt educational institutions where monitoring serves educational activities or child safety. Whether that exemption extends to edtech vendors operating direct to consumer is still an open question industry-wide, which matters more for our B2C line than our institutional one. We have also mapped secondary exposure: state-level regulation of coaching, NEP-driven assessment reform, and likely future AI-specific rules on automated decisions affecting minors. We structured our B2B model so the institution sits as data fiduciary and Goodhart as processor, placing us inside the compliance boundary by design, not by interpretation. We are not relying on the exemption alone. Consent flows and contracts are built conservatively now, ahead of any regulator forcing the point, so a stricter reading of the rules later does not require us to rebuild anything. |
Three things, in order. Telemetry ingestion first: live assessments already produce bursty high-frequency writes, and at ten times the concurrency the Postgres write path and current caching layer become the bottleneck, so this moves to streaming ingest with a time-series or columnar store behind it. Second, inference cost and latency. Nine model families served independently is affordable at pilot scale and not at tens of thousands of students, which is why we are evaluating consolidated GPU serving through Triton and quantised inference through TensorRT rather than scaling instances linearly. Third, retraining cadence. BKT and DKT are retrained on a schedule that assumes manual review; at scale this needs automated pipelines with drift-triggered retraining and canary evaluation. The LLM persona layer is the largest cost line and the strongest argument for self-hosting. |
AI and ML development is run in-house. The model stack, telemetry pipeline and evaluation harness are built internally rather than outsourced. Our AI/ML direction is guided by Manish S, who brings ten years across Amazon, Flipkart and Philips, and advises on model selection, architecture and the path to self-hosted inference. The rest of the founding team and our wider advisory group contribute across engineering scale, product and signal design, and pedagogical validity, the last of which matters because our models make claims about student cognition rather than engagement. Part of the pre-seed is earmarked for hiring dedicated ML engineers as data volume grows. |
All behavioural data is first-party and owned: interaction telemetry generated on our own platform under consent. Question content is drawn from publicly available sources, principally past board and entrance examination papers and openly published curriculum material aligned to the Karnataka PU and NCERT syllabi. We use no purchased or scraped third-party student datasets. Public benchmark datasets are used only for offline validation of knowledge-tracing implementations, never in production. Foundation model work will train exclusively on sequences we generate ourselves. |
The product is the training loop. Every session adds labelled sequences, and every subsequent test provides delayed ground truth against which prior mastery predictions and risk flags are scored. We retrain IRT parameters and BKT priors per cohort, with drift monitoring triggering retraining off-cycle. Teacher feedback is an explicit signal: Albert's question critiques are accepted or rejected, and every rejection is a labelled correction. Risk models are evaluated on whether flagged students actually improved after intervention, measuring usefulness rather than accuracy alone. Near term we are building a replay harness so new model versions can be scored against historical cohorts before release, with canary rollout by class. Longer term, accumulated sequence volume across the Karnataka expansion is what makes the Student Behavioural Foundation Model trainable. |
India first. Our immediate focus is Karnataka, beginning with Dakshina Kannada and Udupi, then Bengaluru and statewide. Building for this system specifically, with its high-stakes board and entrance examinations, large class sizes and coaching economics, is what makes the diagnostics accurate. The underlying layer is not India-specific: cognitive signals, knowledge tracing and behavioural embeddings transfer to any examination-driven education system, so geographic expansion is a natural extension once the model is trained at Indian scale. |
|
http://NA |
Goodhart exists because of a gap the exam system refuses to see. National learning scores fall from fifty-nine percent in Class 3 to 37.8 percent by Class 10 (UNSDG 4), and eighty-one percent of students cite exams as their main source of anxiety, with exam-linked suicides crossing two thousand in a single year (UNSDG 3). Early, honest insight into where a student stands is one of the few things that can catch a student before that spiral peaks at result day.
That gap is also unequally distributed. Urban households spend more than double what rural households spend on coaching, and Kota-style packages price out most of the country (UNSDG 10). Goodhart's diagnostic layer travels with the institution, not the family's wallet, and its flat per-seat cost also removes a quieter bias, families already spend thirteen percent less on girls' education than boys (UNSDG 5).
The other half of the mission is the teacher. Indian teachers spend only nineteen percent of school hours actually teaching (UNSDG 8), and Goodhart gives that time back for free, so they know exactly who needs help and on what. We built Goodhart to be the accountable link between EdTech, schools, and the capital that India's education outcomes actually need (UNSDG 17), not another dashboard measuring the wrong thing more precisely. |
NA |
NA |
checked |
| Sep 15, 2026 @ 2:45 PM |
Soma Bhattacharjee, Naji Rahman |
Soma@mahimahi.in |
https://www.linkedin.com/in/somabhatta/ |
https://www.mahimahi.in/ |
9739752792 |
Soma is a CEO & Co-founder; heads product and marketing. Comes from advertising and product marketing background. Naji is COO & Co-founder, heads sales and operations. Comes from Operations and Digital Marketing background. |
Were colleagues at Regalix (now Marketstar). Have been working together from 5 years. |
All |
We come from marketing, martech, and AI marketing backgrounds and understand the domain very well to be able to build and scale our business. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Sep 14, 2026 @ 7:26 PM |
AKASH CHANDRAKANT PAREKH |
akash@arglyeenigma.com |
https://www.linkedin.com/in/parekha/ |
https://www.arglyeenigma.com/ |
08780367120 |
Akash Parekh, Founder & CEO, with 15+ years of experience in finance and technology; formerly with Morgan Stanley (USA), holding a degree in Computer Engineering and a Master’s in Computer Science. |
I am solo-founder. |
1 |
Our biggest strength is the combination of deep financial expertise and advanced technology capabilities, enabling us to build AI solutions tailored to real-world financial, audit, and compliance challenges. |
ArgyleEnigma Tech Labs |
https://www.arglyeenigma.com/ |
Gandhinagar |
FinDocGPT is building finance-specific AI infrastructure to automate financial analysis, auditing, compliance, and enterprise workflows with greater accuracy and efficiency. |
FinDocGPT addresses the limitations of generic AI in finance, where accuracy, regulatory compliance, and secure handling of sensitive data are critical. It automates complex financial workflows, reducing manual effort, errors, and operational costs. |
FinDocGPT delivers finance-specific AI models and tools that automate financial analysis, auditing, compliance, and document processing, enabling enterprises to improve accuracy, efficiency, and decision-making. |
FinDocGPT combines deep financial expertise with proprietary finance-specific AI models, specialized audit and compliance capabilities, and secure enterprise-ready infrastructure designed for high-accuracy financial workflows. |
Revenue |
Revenue |
Banks, financial institutions, accounting and audit firms, fintech companies, and enterprises seeking AI-powered financial analysis, auditing, compliance, and automation. |
$190B |
$10B |
$100M |
FinDocGPT follows a B2B SaaS and enterprise licensing model, generating recurring revenue through subscriptions, usage-based pricing, specialized audit and compliance modules, and enterprise deployment and integration services. |
Main competitors: Bloomberg (BloombergGPT) AlphaSense FinChat Hebbia Daloopa Rogo Aiera |
FinDocGPT acquires customers through founder-led B2B sales, strategic partnerships with financial institutions and audit firms, targeted LinkedIn outreach, industry networking, and pilot deployments that convert into recurring SaaS subscriptions and enterprise licenses. |
FinDocGPT will adopt a focused B2B go-to-market strategy, targeting financial institutions, audit firms, and enterprises through founder-led sales, strategic partnerships, and pilot deployments. We will initially focus on high-value financial workflows, converting successful pilots into recurring SaaS subscriptions and enterprise licenses, followed by expansion into international markets. |
To build FinDocGPT into a globally trusted AI infrastructure platform for finance, transforming financial analysis, auditing, compliance, and enterprise decision-making through secure, accurate, and specialized AI. |
yes |
NA |
Yes |
We are applying to Avinya to gain access to strategic mentorship, investor networks, and growth opportunities that can accelerate FinDocGPT’s development. We aim to strengthen our finance-specific AI technology, expand our enterprise customer base, and scale our business into a globally competitive AI infrastructure company. |
During the programme, we aim to strengthen FinDocGPT’s finance-specific AI technology, secure strategic investor connections, acquire enterprise customers, and establish scalable B2B partnerships. Our goal is to validate our business model, accelerate revenue growth, and build a strong foundation for global expansion. |
yes |
Yes |
Core engine |
FinDocGPT uses a modular AI architecture combining finance-specific language models, document processing, retrieval-augmented generation (RAG), and financial workflow automation. Our roadmap includes enhancing a 3B-parameter finance-specific model and developing specialized 300M-parameter models for auditing and compliance. The architecture is designed for domain-specific accuracy, explainability, and enterprise integration. |
Our advantage lies in domain-specific financial knowledge, structured financial document processing, and specialized datasets developed for financial analysis, auditing, and compliance. We aim to build a continuously improving data and evaluation framework using appropriately licensed, public, and customer-authorized data. |
FinDocGPT’s defensibility comes from finance-specific model development, specialized audit and compliance capabilities, domain-focused evaluation frameworks, and deep financial expertise. Our long-term advantage will be strengthened through proprietary workflows, customer integrations, and continuous improvement from authorized domain-specific data. |
We evaluate model performance using financial question-answering accuracy, factual consistency, numerical calculation accuracy, document extraction precision and recall, hallucination rates, and task completion rates. We also assess latency, reliability, and performance on domain-specific test datasets, with human review for critical financial outputs. |
Our approach prioritizes data minimization, encryption in transit and at rest, access controls, and secure handling of financial documents. We aim to implement enterprise-grade privacy safeguards, appropriate data retention policies, and compliance with applicable Indian and international data protection requirements. Customer data will not be used for model training without appropriate authorization. |
Potential risks include evolving data protection regulations, financial-sector cybersecurity requirements, and regulatory obligations related to AI-generated financial outputs. We will address these through privacy-by-design, human oversight, appropriate disclosures, and ongoing regulatory monitoring, with requirements varying by jurisdiction and customer use case. |
At 10x scale, the primary challenges could include GPU capacity, inference latency, document-processing throughput, infrastructure costs, and maintaining accuracy across diverse financial workflows. We plan to address these through model optimization, scalable cloud infrastructure, caching, workload management, and continuous performance monitoring. |
Our founding team combines 15+ years of finance and technology experience, including experience at Morgan Stanley (USA), with expertise in computer engineering and computer science. Our technical focus includes finance-specific language models, AI-driven document intelligence, and financial workflow automation. |
Our data strategy includes publicly available financial documents, appropriately licensed datasets, and private financial datasets obtained with explicit customer consent and authorization. We also develop internal datasets and evaluation benchmarks. All data used for training and evaluation will comply with applicable licensing, privacy, confidentiality, and usage restrictions. |
We plan to continuously improve model performance through domain-specific benchmarking, human feedback, error analysis, retrieval optimization, and targeted fine-tuning. Performance will be monitored across accuracy, hallucination rates, latency, and financial task completion, with periodic model updates and regression testing to maintain reliability. |
Yes |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/f1mbjqywf6JQ-Latest-Intro-Fili-FinDocGpt.pptx.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/f1mbjqywf6JQ-Latest-Intro-Fili-FinDocGpt.pptx.pdf |
https://youtu.be/U6nar1or7tU |
Yes. FinDocGPT is mission-driven, with a vision to democratize access to advanced financial AI and make financial intelligence more accessible, accurate, and affordable. We aim to empower businesses, accounting firms, and financial professionals—particularly small and medium-sized enterprises—through AI-powered financial analysis, auditing, and compliance automation. Our goal is to reduce manual workloads, improve financial transparency, and enable more informed decision-making. |
NA |
Carolin Taenzer |
checked |
| Sep 14, 2026 @ 4:05 PM |
MANIK SEHGAL ,Surbhi Sehgal |
manik.sehgal@raasakarts.com |
https://www.linkedin.com/in/manik-sehgal/ |
https://raasakarts.com/ |
09811259603 |
Manik Sehgal – Founder & CEO: 15+ years of experience in advertising, brand strategy, consumer insights, business growth and marketing. Leads overall strategy, expansion, partnerships, brand building and franchise growth. Surbhi Sehgal – Co-Founder, Business Strategy & Operations: 10+ years of experience in e-commerce, business operations and development. Leads operations, vendor management, customer experience, franchise coordination, strategic partnerships and investor relations. |
We are co-founders and have been working together on Raasa Karts for over four years, since its inception in 2022. Our complementary backgrounds in brand strategy, marketing, e-commerce and operations brought us together to build and scale the business. |
2 |
Our biggest strength is our complementary skill set and strong execution on the ground. Manik brings expertise in strategy, branding, marketing and business growth, while Surbhi brings strength in operations, e-commerce, vendor management and partnerships. Together, we combine vision with execution and have spent over four years working closely with street food vendors, deeply understanding their challenges and building solutions around real market needs. |
Raasa Karts India Private Limited |
https://raasakarts.com/ |
Ghaziabad |
Building India's Largest Network of Organized Street Food Courts to Empower Street Food Vendors and Organize India's Street Food Ecosystem. |
India has nearly 2 million street food vendors, yet most remain unorganized and lack basic infrastructure, digital access and stable business locations. Vendors often operate without clean water, electricity, washrooms, weather protection or secure storage, resulting in unpredictable income and difficult working conditions. The challenge is greater for women due to safety and sanitation concerns. Customers love affordable street food but often face inconsistent hygiene, limited seating and unorganized environments. Raasa Karts addresses this gap by creating organized, technology-enabled Street Food Courts that provide vendors with dignified infrastructure and customers with a clean, safe and family-friendly street food experience. |
Raasa Karts is building a technology-enabled, organized Street Food Court ecosystem to address the challenges faced by India’s fragmented street food sector. Street food vendors often operate without secure locations, basic infrastructure, predictable footfall, branding or business support, while customers face concerns around hygiene, safety, seating and basic facilities.
Raasa Karts brings independent street food entrepreneurs together in professionally managed Food Courts and provides them with stable business spaces, electricity, RO water, washrooms, weather protection, seating, branding, marketing and operational support. Digital systems such as centralized billing, cashless payments, order management and sales tracking help bring greater transparency and efficiency to operations.
For customers, Raasa Karts offers multiple authentic street food options under one roof in a clean, hygienic, safe and family-friendly environment while retaining the affordability and authenticity of traditional street food. |
Our solution is unique because it combines organized physical infrastructure, technology, vendor enablement and customer demand in one ecosystem designed specifically for street food entrepreneurs. Unlike traditional food courts, delivery platforms or vending zones, Raasa Karts provides vendors with secure spaces, essential infrastructure, branding, marketing and operational support while preserving their independence and authenticity. Our defensibility comes from 4+ years of on-ground experience, a network of 5,500+ vendors, deep understanding of vendor economics, standardized operating SOPs, location and vendor-selection capabilities, technology-enabled operations and a scalable franchise model. This ecosystem becomes stronger and harder to replicate as the network expands. |
Product-Market Fit |
Revenue |
Our target customers are street food vendors and micro food entrepreneurs seeking secure, organized and professionally managed business spaces. On the consumer side, we target urban families, women, young professionals and students looking for authentic, affordable street food in a clean, hygienic and comfortable environment. Franchise partners are also an important customer segment as we scale our Food Court network across India. |
50 Billion Dollar |
20 Billion Dollar |
2 Billion Dollar |
Our revenue model has three key streams: monthly vendor rentals, franchise revenue share, and central billing revenue share. Vendors pay ₹15,000–₹20,000 per month for space in our Food Courts. In the franchise model, Raasa Karts earns 10% of the total vendor rental revenue collected by the franchisee. As centralized billing is implemented, Raasa will additionally earn 10% of vendor food sales, creating a scalable recurring revenue stream as the Food Court network expands. |
Our main competitors are traditional food courts, government street-vending zones/organized food streets, and food-delivery platforms such as Zomato and Swiggy. However, there are currently few direct organized players combining street food vendor infrastructure, technology, vendor support, customer experience and franchise-led expansion in a single model. |
We acquire customers through hyperlocal digital marketing, social media, local influencers, Google and Meta advertising, community outreach, events and word-of-mouth. Our Food Courts are promoted within targeted local catchments to drive nearby families, students and working professionals. On the vendor side, we leverage our existing 5,500+ street food vendor network, direct vendor outreach and referrals. As we expand, franchise partners will also drive local customer acquisition and vendor onboarding in their respective markets. |
Our go-to-market strategy is hyperlocal and franchise-led. We identify high-footfall locations, onboard a curated mix of street food vendors, and launch organized Food Courts supported by targeted local marketing. Customer acquisition is driven through Meta and Google advertising, social media, local influencers, community engagement, events and word-of-mouth within the surrounding catchment. For vendor acquisition, we leverage our 5,500+ vendor network, direct outreach and referrals. As we scale, franchise partners will fund and operate local Food Courts while Raasa provides the brand, vendor network, technology, SOPs, marketing support and operational framework. |
Our go-to-market strategy is hyperlocal and franchise-led. We identify high-footfall locations, onboard the right mix of street food vendors, and create organized Food Courts for the surrounding community. We drive customer footfall through hyperlocal digital advertising, social media, local influencers, events, community outreach and word-of-mouth. Vendor acquisition comes through our 5,500+ vendor network, direct outreach and referrals. For expansion, we follow an asset-light franchise model, where franchise partners invest in local Food Courts while Raasa provides the brand, vendor ecosystem, technology, SOPs, marketing support and operational expertise. |
Yes |
1.5 crore |
Yes |
We are applying to Avinya to access strategic funding and ecosystem support to accelerate our growth. Avinya’s focus on impact-driven ventures aligns with our mission to formalize India’s street food economy, and their support will help us strengthen operations, expand faster, and build a sustainable, scalable business. |
We want to leverage Avinya’s hybrid programme (virtual + incubation) to build and scale Raasa Karts in a structured and accelerated way. |
Yes |
Yes |
Supporting feature |
Our AI architecture is being developed around AI-enabled camera analytics and centralized billing systems across our Food Courts. Smart cameras can help monitor customer footfall, peak hours, queue patterns, vendor activity and operational compliance, while centralized billing captures vendor-wise transactions and sales performance. Combining these data sources will enable AI-driven analytics for sales forecasting, vendor performance, customer demand patterns, product/category trends and operational alerts. At the network level, this will create a centralized intelligence layer enabling Raasa Karts to monitor and optimize multiple Food Courts and support franchise partners and vendors with data-driven decisions. |
Our proprietary data advantage comes from 4+ years of working directly with street food vendors and customers. We have built a database of 5,500+ vendors and 50,000+ customer orders, providing insights into vendor categories, menus, pricing, customer preferences and demand patterns. As we scale our Food Courts, centralized billing and AI-enabled camera analytics will generate additional proprietary data on vendor-wise sales, customer footfall, peak hours, category performance and location-level demand. This growing dataset can strengthen vendor selection, location planning, demand forecasting, menu optimization and overall Food Court performance, creating a data advantage that becomes stronger as our network expands. |
Our AI defensibility comes from the proprietary, hyperlocal data generated across our Street Food Court network. By combining centralized billing data with AI-enabled camera analytics, we can build location- and vendor-level intelligence on sales, footfall, peak hours, customer patterns, vendor performance and category demand. This is strengthened by our existing network of 5,500+ street food vendors and 50,000+ customer orders. As more Food Courts join the network, the dataset becomes richer, enabling better demand forecasting, vendor selection, menu optimization and location planning. The combination of proprietary data, physical network and operating experience creates our defensibility. |
We evaluate AI model performance using accuracy, precision, recall and error rates, depending on the use case. For camera analytics, we measure footfall-count accuracy, queue/occupancy detection accuracy and consistency across different lighting and operating conditions. For billing and predictive analytics, we track sales-data accuracy, forecast error, demand-prediction accuracy and anomaly-detection rates. Model outputs are periodically compared with actual billing records and manual observations to validate reliability. As more Food Courts generate data, models will be continuously monitored and improved to maintain consistent performance across different locations and vendor categories. |
We follow a privacy-by-design approach, collecting only the data required for billing, operations, analytics and customer service. Transaction and operational data is stored with role-based access controls, authentication, encryption and regular backups. For AI-enabled camera analytics, our focus is on aggregated footfall, occupancy and operational patterns rather than personal identification or facial recognition. Access to sensitive business and customer data is restricted to authorized personnel. We aim to comply with applicable Indian data-protection requirements, including the Digital Personal Data Protection framework, and will continuously strengthen consent, retention, security and data-governance practices as our network scales. |
Yes. Key regulatory risks include FSSAI compliance, municipal/trade licences, fire and safety approvals, property/lease regulations, food hygiene standards, franchise compliance, payment/data regulations and privacy requirements for AI-enabled cameras. We mitigate these through appropriate licences and agreements, standardized hygiene and safety SOPs, regular compliance checks, clearly defined vendor/franchise responsibilities, and privacy-by-design technology. Our AI camera systems are intended for aggregated footfall and operational analytics, not facial recognition, reducing privacy risk. As we expand across cities, we will assess local regulatory requirements before launching each Food Court. |
Yes. Key regulatory risks include FSSAI compliance, municipal/trade licences, fire and safety approvals, property/lease regulations, food hygiene standards, franchise compliance, payment/data regulations and privacy requirements for AI-enabled cameras. We mitigate these through appropriate licences and agreements, standardized hygiene and safety SOPs, regular compliance checks, clearly defined vendor/franchise responsibilities, and privacy-by-design technology. Our AI camera systems are intended for aggregated footfall and operational analytics, not facial recognition, reducing privacy risk. As we expand across cities, we will assess local regulatory requirements before launching each Food Court.
What will break in your AI system if you scale 10
If our AI system scales 10×, the main challenges will be handling significantly larger volumes of camera feeds, billing transactions and multi-location data while maintaining accuracy and real-time performance. Differences in lighting, crowd density, Food Court layouts and vendor behaviour could also reduce camera-analytics accuracy. We will address this through cloud-based scalable infrastructure, standardized camera deployment, automated data pipelines, continuous model monitoring and location-specific model calibration. Centralized billing data will also provide a reliable reference point to validate AI outputs and identify anomalies as the network grows. |
No, we currently do not have a dedicated in-house AI/ML team. Our core team focuses on business, technology and operations, while specialized AI development will be supported through external technology partners and future AI/ML hires. As we scale, we plan to build in-house capabilities for computer vision, data analytics, demand forecasting and AI-driven vendor performance insights. |
We primarily use first-party data owned by Raasa Karts, generated through our operations. This includes data from 5,500+ onboarded vendors and 50,000+ customer orders, covering vendor categories, menus, pricing, order patterns and customer demand. Our Food Courts are also beginning to generate vendor-wise sales, billing, footfall and operational data. As centralized billing and AI-enabled camera analytics are implemented, these will create additional proprietary datasets on sales, footfall, peak hours and vendor performance. Where third-party technology platforms are used, data access is governed by their respective agreements and licensing terms. |
We plan to continuously improve model performance by feeding new first-party data from centralized billing, vendor sales and AI-enabled camera analytics into our systems. We will regularly compare AI outputs with actual billing records and manual observations to identify errors and retrain or recalibrate models. As our Food Court network expands, data from different locations, layouts, customer volumes and vendor categories will improve model robustness. We will continuously monitor accuracy, forecast error, anomaly detection and footfall-count accuracy, with human review for exceptions and periodic model updates. |
We are currently building primarily for India, starting with Delhi-NCR and expanding across Tier 1, Tier 2 and Tier 3 cities. India is our core market because of its nearly 2 million street food vendors and deeply established street-food culture. In the long term, once the model is proven at scale in India, we see potential to replicate it in other emerging markets with large informal street-food ecosystems. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/LjUljX78Fadk-Raasa-Karts_September-2026_final-deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/LjUljX78Fadk-Raasa-Karts_September-2026_final-deck.pdf |
https://drive.google.com/file/d/1YJSZ0kJpzQHYXatqTKi9X9817HzZ4dhs/view?usp=sharing |
Yes. Raasa Karts is a mission-driven, impact-focused startup working to create a Street Food Revolution in India. Our mission is to transform street food vendors into empowered, recognized and sustainable micro-entrepreneurs by providing secure business spaces, essential infrastructure, technology, branding, marketing and customer access.
Our impact focuses on improving livelihoods, creating dignified working conditions, supporting grassroots entrepreneurship and enabling greater participation of women entrepreneurs through safer and more organized workplaces. At the same time, we provide consumers with authentic and affordable street food in clean, hygienic and family-friendly environments.
Our long-term vision is to positively impact India’s nearly 2 million street food vendors while building a commercially sustainable and scalable business. |
Yes. Raasa Karts is a women co-founded startup, with Surbhi Sehgal playing a core leadership role across business strategy, operations, vendor management, partnerships and expansion. We also work primarily with street food vendors and micro-entrepreneurs, a traditionally underserved and largely informal segment with limited access to organized infrastructure, technology and business support. Our model particularly aims to create safer and more dignified entrepreneurship opportunities for women in the street-food ecosystem. |
LinkedIn |
checked |
| Sep 12, 2026 @ 10:08 PM |
Bhasker |
omega.bhasker@gmail.com |
https://www.linkedin.com/in/bhasker-kumar-100 |
https://borg-cube.netlify.app/ |
7903437237 |
Founder-led AI researcher and deep-tech systems architect with 9+ years across AI/ML, defence-oriented technical systems, computer vision, sensor fusion, edge deployment and Space Situational Awareness-related work. Mechanical-engineering background combined with first-principles AI architecture research. Founder and principal architect of Project Cube, a compact sovereign foundation-model programme combining specialist intelligence, tensor-network research, structured memory, scientific/engineering constraints and verification-before-authority |
we are college friends and bhasker, i am the geeky introvert one and Rahul is good communicator vary easily he connect with people |
2 |
Our biggest strength is our ability to turn difficult AI research ideas into working prototypes and continuously improve them through experimentation. We combine deep technical interest in AI architecture, tensor networks, reasoning systems, and efficient model development with a strong hands-on approach to building and testing.
We are not approaching the cube as another application layer on top of existing AI models. We are developing the underlying architecture ourselves, including tensor-network-based compression, adaptive cognitive routing, symbolic reasoning, verification, and energy-efficient execution.
Our strength is also persistence: we work iteratively, study failures, redesign the architecture, and validate ideas through measurable experiments rather than relying only on theory. This combination of research mindset, engineering execution, and long-term commitment gives us the ability to tackle technically ambitious problems and turn them into deployable technology. |
stableomega technolgy |
https://borg-cube.netlify.app/ |
Bengaluru (Bangalore) Urban |
We are building Flowless++, a new AI architecture for highly efficient frontier intelligence—targeting frontier-level capability and AGI with dramatically fewer parameters through tensorized computation, cognitive routing, symbolic reasoning, verification, and knowledge-dense learning. |
Current frontier AI systems increasingly depend on enormous parameter counts and compute budgets to achieve stronger reasoning and general-purpose capabilities. This creates high training and inference costs, large memory requirements, and limits deployment on affordable and resource-constrained hardware. The deeper problem is that scaling parameters alone does not guarantee efficient reasoning, reliable knowledge use, or efficient computation. We are solving this architectural inefficiency by developing an AI system that can achieve substantially higher capability per parameter and per unit of compute. |
THE CUBE is a new foundation-model architecture designed around a structured cognitive computation layer rather than relying only on conventional dense transformer scaling. Its architecture combines tensor-network-based computation and compression, adaptive cognitive routing, structured reasoning states, specialized reasoning modules, symbolic reasoning graphs, law-based knowledge, contradiction detection, exact verification, self-checking, and failure-driven learning.
The objective is to increase the knowledge density and reasoning capability of every parameter while reducing unnecessary computation. Our long-term target is a frontier AI model around 20B parameters that can compete with much larger models on capability and efficiency, with a path toward AGI-class general intelligence below 100B parameters. We are validating this direction incrementally through measurable reasoning and efficiency benchmarks rather than treating parameter count alone as the measure of intelligence. |
Our defensibility comes from the architecture itself and the research stack built around it. THE CUBE is not simply an agent orchestrator or an application layer around existing foundation models. We are developing a unified computational architecture combining tensor-network representations, cognitive routing, structured state, symbolic reasoning, law governance, verification, contradiction handling, and self-improvement. This creates a tightly integrated system where computation can be allocated according to problem complexity, reasoning can be verified rather than only generated, and failures can be used to improve routing, modules, and learned knowledge. Our research direction also includes autonomous discovery and self-improving compute architecture, creating a long-term path for the system to improve its own reasoning and computational efficiency. The key moat is therefore not a single model checkpoint, but the accumulated architecture, training methodology, reasoning infrastructure, evaluation framework, and empirical knowledge required to reproduce and extend the system. |
MVP |
Pilots |
Enterprises, AI companies, research institutions, governments, and developers needing powerful, deterministic , efficient, and deployable foundation models. |
$500B+ global AI and foundation-model market |
$100B+ market for enterprise AI, efficient inference, and foundation-model infrastructure |
$1B+ initial opportunity through enterprise deployments, model licensing, APIs, and strategic partnerships. |
API and inference services, enterprise licensing, customized models, and foundation-model licensing. Design and research partnership |
OpenAI, Google DeepMind, Anthropic, Meta, DeepSeek, and leading open-source foundation-model providers. |
Through research demonstrations, benchmark results, pilot deployments, strategic partnerships, and direct enterprise outreach. |
Start with research and enterprise pilots, demonstrate superior capability-per-parameter and inference efficiency, then scale through API, licensing, and strategic partnerships. |
Build a highly efficient frontier AI platform capable of AGI-class intelligence with dramatically lower parameter and compute requirements, enabling powerful AI from data centers to edge devices. fully autonomous research lab AI model collaboration/control with silicon to adapt and evolve with nature like efficiency so all labs and factory runs fully automatically and frontier AI researcher to solve problems like fusion and climate design chips |
incorporated |
NA |
Yes |
We are applying to Avinya to accelerate the development and validation of project cube, our efficient foundation-model architecture. We want access to experienced AI mentors, industry connections, IIT ecosystem expertise, compute and pilot opportunities, and a strong startup network that can help us convert our research prototype into a scalable AI product. |
Our goal is to strengthen the project cube architecture, validate its efficiency and reasoning advantages through rigorous benchmarks, improve our prototype toward a production-ready foundation model, and establish our first research or industry pilots. We also want to develop our go-to-market strategy and prepare for the next stage of scale and investment. |
Yes. We are open to hybrid and virtual participation |
Yes |
Core engine |
The cube is a modular foundation-model architecture combining Tensor-Train/low-rank tensor computation, Adaptive Tensor Attention, adaptive cognitive routing, specialized reasoning modules, structured memory, symbolic reasoning, and verifier-driven execution. Computation is activated conditionally according to task complexity and necessity rather than keeping the entire model active. |
Our current advantage is primarily architectural rather than proprietary data. We use public benchmark and research datasets while building proprietary training pipelines, reasoning traces, architectural telemetry, verification data, failure cases, and model-generated evaluation artifacts. |
Our defensibility comes from the integrated architecture: tensor-network computation, adaptive ranks, metabolic/resource-aware routing, modular reasoning, structured memory, symbolic verification, and auditable transition contracts. The system is designed to improve capability-per-parameter rather than relying primarily on parameter scaling. |
We use exact-match/task accuracy, validation loss/perplexity, reasoning-family accuracy, verifier acceptance, contradiction/closure checks, calibration, inference latency, active parameters, FLOPs, memory/VRAM usage, and structural metrics such as module overlap and ecosystem size. Our evaluation framework also requires reproducible dataset and hardware metadata. also compare them with other enterprise model like gsm8k accuracy result our 122mn model performs better then 30bn model granite 4.1 model |
The current system is primarily research-stage and does not require sensitive personal data for its core validation. For deployment, we plan data minimization, access controls, encryption, audit logging, dataset provenance, secure isolation, human oversight, on prem data adaption with local finetuning as our model adapt the new data in real time like free finetuning as service as very less computation is required and compliance with applicable Indian and international data-protection requirements. |
Yes. As the system scales, risks may include AI safety requirements, data licensing and copyright, privacy/data protection, model accountability, sector-specific regulation, and export or compute restrictions. We plan to address these through documented provenance, auditing, human oversight, controlled deployment, and compliance-by-design. |
The main risks are compute and memory bandwidth, tensor-contraction efficiency, routing complexity, verification latency, module-management overhead, and maintaining stable adaptive-rank behavior. Our architecture explicitly addresses these through bounded active modules, adaptive ranks, fused tensor kernels, hierarchical memory, and audit/fallback mechanisms. |
Yes. we have developed the architecture and training stack in-house, with expertise across deep learning, tensor networks, low-rank computation, model training, reasoning systems, symbolic verification, efficient inference, and AI systems engineering. |
We currently use public research and benchmark datasets, including WikiText-2, WikiText-103, and reasoning benchmarks such as GSM8K. These are used for research and evaluation under their respective dataset licenses/terms; our proprietary assets are the training infrastructure, evaluation methodology, architectural telemetry, and generated research artifacts |
We use benchmark-driven iteration, failure analysis, ablation studies, adaptive routing/rank optimization, structured verification, hard-case replay, improved data curation, and scaling experiments. Each architecture change is evaluated through reproducible experiments before promotion. |
Both. We are building the core technology for global markets while prioritizing India for research partnerships, enterprise pilots, affordable AI infrastructure, and deployment on resource-constrained hardware. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/V5iov0ZF7fGd-Project_Cube_small-deck-1.pdf, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/PW25HRxVzlzw-Project_Cube_Technical_Deck_v2-1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/V5iov0ZF7fGd-Project_Cube_small-deck-1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/PW25HRxVzlzw-Project_Cube_Technical_Deck_v2-1.pdf |
https://drive.google.com/file/d/13Jd5YsV62qg-l0v4YhukEavb465och3m/view?usp=drive_link |
Yes. Our mission is to make advanced AI significantly more efficient, affordable, and accessible as from country like india can not have big gpu cluster so we need more efficient architecture not fully transformer dependent architecture . Flowless the cube aims to reduce the compute, memory, and infrastructure required for capable AI systems, enabling frontier-level intelligence to become more practical for startups, enterprises, researchers, and resource-constrained environments. We ultimately want to develop to a new generation of efficient and broadly deployable AI. we will achieve AGI in less then 100bn parameter model will beat the frontier model with just 20bn model that can adapt and evolve in real time |
NA |
Self-referred |
checked |
| Sep 10, 2026 @ 5:21 PM |
Anup Ganiger |
support@hackerhub8.in |
https://www.linkedin.com/company/hacker-hub-8/ |
http://hackerhub8.in |
7411310321 |
CEO - Founder |
We met through our shared interest in technology and cybersecurity and began working together through HackerHub8. Over the years, we have built strong working relationships through community initiatives, projects, events, and cybersecurity activities, developing complementary skills and a shared vision for the company. |
2 |
Our biggest strength is our diverse and complementary expertise combined with strong execution. We bring together skills across cybersecurity, technology, product development, business, community building, and operations. Our ability to collaborate, adapt quickly, and turn ideas into practical solutions gives us a strong advantage as we scale |
HackerHub8 |
hackerhub8.in |
Vijayapura |
HackerHub8 is a cybersecurity product and services company helping organizations identify, manage, and reduce digital security risks. |
Organizations, especially startups and growing businesses, face increasing cyber threats but often lack affordable, accessible, and practical cybersecurity capabilities. Security gaps, vulnerabilities, and limited in-house expertise can leave businesses exposed to attacks, data breaches, and operational disruption. |
HackerHub8 combines cybersecurity products with professional security services to help organizations proactively identify and address security risks. Our offerings include vulnerability assessment, penetration testing, security audits, security consulting, awareness, and technology-driven security solutions, helping businesses continuously improve their security posture. |
Our strength lies in combining practical cybersecurity expertise, technology, and a strong security ecosystem. We focus on making enterprise-grade security capabilities more accessible and actionable for organizations while continuously developing our products around real-world security challenges. Our technical talent, cybersecurity community, industry network, and execution experience create a strong foundation for long-term differentiation |
Idea |
Users |
Startups, SMEs, enterprises, technology companies, financial services organizations, educational institutions, and organizations seeking cybersecurity products and professional security services. |
$200B+ — Global cybersecurity market |
$20B+ — Addressable cybersecurity services and solutions across our priority markets |
$100M+ — Initial realistically targetable opportunity through focused B2B customer acquisition and partnerships |
B2B revenue model through cybersecurity products, subscription-based solutions, professional security services, security assessments, penetration testing, consulting, and customized enterprise engagements |
Tenable, Qualys, Rapid7, CrowdStrike, Sophos, and regional cybersecurity service |
We acquire customers through direct B2B outreach, referrals, partnerships, cybersecurity events and conferences, our professional community, digital marketing, content, and strategic enterprise relationships. |
Our strategy combines direct enterprise sales with channel and technology partnerships. We initially target startups, SMEs, and technology-driven enterprises through security assessments and consulting engagements, then expand into recurring product and managed security offerings. We plan to scale through strategic partnerships, referrals, industry events, and expansion into international markets |
To build HackerHub8 into a globally recognized cybersecurity technology company providing scalable products and services that help organizations proactively identify, prevent, and respond to cyber threats. Our long-term vision is to create an integrated cybersecurity ecosystem combining innovative technology, security intelligence, and expert services for businesses worldwide. |
yes LLP |
NA |
Yes |
We are applying to Avinya to accelerate HackerHub8’s growth as a cybersecurity product and services company. We are looking for strategic mentorship, industry connections, customer opportunities, and support in strengthening our product-market fit and business model. We also see the programme as an opportunity to connect with enterprises, investors, technology partners, and international markets. |
Our key goals are to validate and strengthen our product-market fit, acquire enterprise customers, build strategic technology and channel partnerships, refine our go-to-market strategy, and prepare HackerHub8 for scalable growth. We also want to strengthen our product roadmap, business positioning, and readiness for future fundraising and international expansion |
yes |
Yes |
Automation layer |
Our architecture combines cybersecurity data collection and processing with AI/ML models for threat analysis, vulnerability prioritization, anomaly detection, and security insights. The system is designed with modular data ingestion, preprocessing, model inference, risk scoring, and reporting layers, with human oversight for high-impact security decisions. |
Our advantage comes from security-focused datasets generated through authorized assessments, vulnerability findings, threat intelligence, security telemetry, and real-world cybersecurity engagements. We aim to continuously build proprietary, high-quality datasets while ensuring appropriate authorization, privacy, and data protection. |
Our defensibility will come from domain-specific cybersecurity data, proprietary detection and risk-scoring workflows, continuous learning from real-world security scenarios, and integration of AI with our cybersecurity expertise and products. This creates a specialized security layer rather than a generic AI application |
We evaluate models using precision, recall, F1-score, false-positive and false-negative rates, detection rate, accuracy, latency, and consistency across different datasets. Security-critical outputs are also validated through human review and controlled testing before deployment. |
We follow security-by-design principles, including data minimization, encryption, access controls, secure data storage, logging, authorization, and controlled data retention. Customer data is handled only with appropriate authorization, and we aim to align our practices with applicable privacy and cybersecurity regulations |
Potential risks include evolving data-protection, cybersecurity, AI governance, and sector-specific regulations across different markets. We plan to continuously monitor regulatory requirements and design our products and data practices to support compliance as we expand internationally. |
The primary scaling challenges would be data-processing volume, model inference costs, infrastructure capacity, monitoring, and maintaining model accuracy as data diversity increases. We plan to address these through scalable cloud infrastructure, optimized inference, automated monitoring, efficient data pipelines, and modular architecture. |
We currently have a technical cybersecurity and software team, with AI/ML capabilities being developed internally and supplemented through specialized expertise where required. |
We use a combination of authorized security assessment data, vulnerability and threat intelligence datasets, publicly available security research, and internally generated data. Data sources are evaluated for licensing, authorization, privacy, and permitted usage before being incorporated into our systems. |
We will continuously improve performance through validated new security data, feedback from real-world deployments, error analysis, periodic model evaluation, adversarial testing, retraining, and monitoring for model drift. Human security expertise will remain an important part of the validation loop. |
only for INDIA |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/iHMYbITzEfQI-HH8-brochure-2026.pdf, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/8eMdXsMoKtri-HH8-Company-DOC.pdf, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/axgAQrWnx7hm-pitch.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/iHMYbITzEfQI-HH8-brochure-2026.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/8eMdXsMoKtri-HH8-Company-DOC.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/axgAQrWnx7hm-pitch.pdf |
http://na |
NA |
NA |
na |
checked |
| Sep 10, 2026 @ 2:40 PM |
Rajani Vemulakonda , Shivkumar Gurram , Abhishek Das |
gs@sourcingxpress.com |
https://www.linkedin.com/in/shivkumargurram |
https://www.sourcingxpress.com/ |
8050020304 |
Shivkumar Gurram – Founder with 15+ years of experience in Talent Sourcing, Technical Recruitment and Talent Acquisition, with experience across organizations including Amazon, IBM, Twilio and Capgemini. Recruitment practitioner, trainer and mentor with deep expertise in sourcing technology talent and understanding recruiter pain points. At SourcingXPress, he leads the vision, product strategy and domain innovation, building technology solutions that help recruiters discover and engage talent beyond traditional job portals. Abhishek Das – Co-Founder & CTO Technology entrepreneur and product engineer with expertise in building and scaling technology products, AI systems and modern software architectures. He has experience applying emerging technologies such as Generative AI, Knowledge Graphs and Graph Databases to solve real-world problems. He is also a Neo4j Ninja and community leader in the Graph/AI ecosystem. At SourcingXPress, Abhishek leads the technology, product architecture and AI innovation, translating the recruitment domain problems into scalable technology solutions. |
Shivkumar and Abhishek first connected when Shivkumar was consulting with a client to hire a Full Stack Engineer, and Abhishek was a candidate in the process. Our conversations soon moved beyond the hiring opportunity to the larger challenges in recruitment, viewed from both the recruiter’s and candidate’s perspectives. We realized that our experiences complemented each other and that we could solve these problems together. This shared insight led us to join forces and build SourcingXPress. |
All |
Our biggest strength is the combination of deep recruitment domain expertise and strong technology expertise. Shivkumar brings years of hands-on experience in sourcing and recruitment, while Abhishek brings expertise in technology, AI and product engineering. Together, we understand the problem from both the recruiter and candidate perspectives, enabling us to build practical, technology-driven solutions for real recruitment challenges. |
GET SOURCINGXPRESS PVT LTD |
sourcingxpress.com |
Bangalore |
Amazon for Hiring - Pay when you buy, not when you browse unlike teh other job portals which force you to pay to browse. |
Recruiters and companies spend significant time and money searching through profiles, job portals and databases, often paying for access, views or subscriptions without knowing whether they will find relevant talent. SourcingXPress is solving this by making talent sourcing outcome-driven. Our AI-powered platform helps companies discover, match and shortlist relevant tech talent faster, while enabling them to pay for qualified shortlists, not for profile views or searches. We are building a more efficient and outcome-focused way to connect companies with the right talent. |
SourcingXPress is an AI-powered talent sourcing and matching platform that helps recruiters identify the right candidates faster. It analyzes applicants against job requirements, provides an exact Match% and detailed skill-level analysis, and gives recruiters visibility into full candidate profiles before they decide to unlock contact details.
Our outcome-driven model allows recruiters to post jobs for free, receive AI-matched candidates, and pay only to unlock relevant candidates, not for searches or profile views. This makes recruitment faster, more transparent and cost-efficient.
In short: Find the right talent. See the fit. Pay only when you want to connect. |
Our defensibility comes from the combination of proprietary matching and fitment signals, a growing candidate and recruiter network, and continuous product co-creation with our recruiter community. Since launch, we have onboarded 1000+ recruiters/companies, created 2000+ jobs and built a 500K+ tech-candidate pool without advertising. In short: Better signals + transparent matching + pay for outcomes + a growing recruitment network. |
Revenue |
Users, Revenue, Pilots, Signups, Testimonials |
Any Startup, Recruiting Agencies and Enterpries companies hiring of Software Engineering / Technology roes |
₹10,000 Cr in India / $800+ Bn in USA |
Tech hiring market, approximately 3.5 lakh jobs in India / 6 lakh jobs in USA |
5% of India SAM = ~17,500 jobs / 2% of USA SAM = ~13,000 jobs |
SourcingXPress follows a freemium, outcome-based SaaS model. Recruiters can post unlimited jobs and view unlimited candidate profiles/resumes for free, including AI match scores. Revenue is generated when recruiters unlock candidate contact details, with additional subscription plans offering larger monthly unlock volumes and multiple recruiter seats. Our plans range from ₹100 per unlock on Pay-as-You-Go to ₹2,000–₹40,000 per month for Seed, Startup, Scaleup and Enterprise plans, with discounts on annual subscriptions. Core principle: Pay for access to the talent you want to engage, not for profile views. |
Naukri, Instahyre, CutShort in India |
Naukri, Instahyre and CutShort in india |
We are building SourcingXPress through a community-led, digital-first and partnership-driven GTM strategy. Our primary customers are corporate recruiters, recruitment agencies, freelancers and startups. Our growth strategy focuses on: Recruiter community & word of mouth, leveraging our existing community for adoption, feedback and product co-creation. Digital marketing & content, educating recruiters on modern sourcing and AI-powered hiring. Offline events & recruitment communities, building trust and acquiring high-intent users. Product-led growth, free job posting and candidate discovery lowers the entry barrier, while recruiters pay only when they unlock relevant talent. Strategic partnerships with IIT, Already partnered with IIT Bambay and IIT Ropar, leveraging technology and ecosystem partners to accelerate product adoption and scale. We have already onboarded 1000+ companies/recruiters and created 2000+ jobs, providing an initial base for scaling this model. |
Our vision is to build SourcingXPress into an omni-channel hiring platform that transforms how companies discover, evaluate and connect with talent.
We aim to move beyond traditional job portals by combining AI, ML and LLM capabilities, intelligent talent matching, recruiter learning, referrals, gig opportunities and a curated talent ecosystem into one platform. Our long-term goal is to make hiring faster, more intelligent, transparent and outcome-driven for companies and recruiters globally. |
Privated Limited |
650000 |
Yes |
For Incubation and funding support |
Mentorship and Funding |
Yes |
Yes |
Core engine |
SourcingXPress uses AI-driven candidate–job matching to analyze job requirements against candidate profiles and generate an Overall Match % and detailed Skill Match Analysis. This enables recruiters to quickly identify the most relevant candidates instead of manually screening large volumes of applications.
The architecture is designed as a scalable, distributed technology platform, with AI/ML/LLM capabilities forming the foundation for increasingly intelligent talent matching and hiring workflows. The platform also uses Neo4j as its database layer, with Azure infrastructure supporting scalability.
In simple terms:
Job → AI understanding → Candidate matching → Match Score + Skill Analysis → Recruiter shortlist → Contact unlock
The architecture will progressively incorporate deeper AI, ML and LLM capabilities as SourcingXPress evolves toward its long-term omni-channel hiring platform. |
Yes. SourcingXPress is building a proprietary recruitment data layer through its platform activity, job requirements, candidate profiles, applications, recruiter interactions and matching/fitment signals. This data is continuously enriched through real hiring activity and recruiter feedback.
We already have a growing network of 500K+ tech candidates, 2000+ jobs and 1000+ recruiters/companies, built without paid advertising, along with an active recruiter community contributing product feedback and insights.
This creates a data flywheel: more jobs and candidates → more matching interactions → better understanding of talent–job fit → better recommendations → more adoption. |
Our defensibility comes from combining AI-powered job–candidate matching with recruitment-specific fitment signals generated from real hiring activity. SourcingXPress evaluates candidates against job requirements and provides an Overall Match % and detailed Skill Match Analysis, helping recruiters understand why a candidate fits rather than relying only on keyword matching.
Our growing proprietary data layer creates a strong foundation for continuously improving these matching signals. Today, SourcingXPress has 500K+ tech candidates, 2,000+ jobs and 1,000+ companies/recruiters, creating a growing data and feedback loop across jobs, candidates, applications and recruiter interactions.
Our moat is not just the AI model, it is the recruitment-specific data, fitment signals and feedback loop built around real hiring activity. |
We evaluate our AI matching system primarily on candidate–job relevance and recruiter outcomes, rather than relying only on generic AI benchmarks.
Key metrics include:
Match Accuracy – how accurately the AI identifies candidates relevant to the job requirements.
Skill Match Accuracy – alignment between required and demonstrated skills.
Shortlist/Unlock Rate – percentage of AI-recommended candidates that recruiters choose to shortlist or unlock.
Recruiter Feedback – qualitative feedback on match relevance and fitment explanations.
Hiring Conversion – candidates progressing from application → shortlist → contact → interview/hire.
False Positive / False Negative Rate – monitoring irrelevant recommendations and missed relevant candidates.
Consistency & Reliability – stability of match scores and recommendations across similar jobs and candidate profiles.
We continuously use real recruiter interactions and hiring outcomes as feedback signals to improve the matching system. |
SourcingXPress is designed with privacy, security and controlled data access as core principles. Candidate information and contact details are protected, with contact details unlocked only when a recruiter chooses to engage with a relevant candidate.
Our technology stack includes Okta for authentication and Azure infrastructure, with plans to strengthen platform security and compliance through ISO and GDPR certification as we scale.
We follow a privacy-by-design approach, limiting unnecessary exposure of candidate information and applying controlled access to sensitive data.
Note: The current pitch deck supports the above controls and planned certifications; it does not provide detailed information on specific encryption standards, data-retention policies, consent mechanisms, or current ISO/GDPR certification status. |
Yes. As a recruitment technology platform handling candidate and employment-related data, SourcingXPress may be impacted by data privacy, consent, data protection, AI governance and employment-related regulations across the markets we operate in.
We recognize that regulations around AI-driven hiring and candidate data are evolving, and changes could affect how candidate information is collected, processed, matched and shared. We are therefore taking a privacy- and security-first approach, with controlled access to candidate contact information and plans to strengthen compliance through ISO and GDPR certification as we scale.
Our approach is to continuously monitor applicable regulations and build compliance requirements into the product architecture as we expand into new markets.
Primary risks: Data privacy & consent, AI regulation, Employment discrimination/bias, Cross-border data compliance. |
The primary challenge at 10x scale would be maintaining matching accuracy, response time and consistency as the volume and diversity of jobs and candidate profiles increase significantly.
We expect the key pressure points to be:
Data processing: Handling significantly larger volumes of candidate and job data.
Matching latency: Maintaining fast AI matching as concurrent searches and applications increase.
Model consistency: Ensuring reliable Match% and Skill Match Analysis across diverse roles and profiles.
Infrastructure scalability: Scaling compute, database and AI workloads without increasing costs disproportionately.
Data quality: Preventing duplicate, outdated or incomplete profiles from degrading match quality.
Our architecture is already designed around distributed and scalable systems, with Azure infrastructure and Neo4j as part of the technology stack, giving us a foundation to address these challenges as we scale.
Our biggest risk at 10x is therefore not whether the system can store 10x the data, but whether we can maintain the same quality and speed of AI-driven matching at that scale. |
Yes. We have an in-house product and technology team led by our Co-Founder & CTO, Abhishek Das, with expertise in full-stack engineering, distributed and scalable systems, AI/ML and product architecture. The team is responsible for building and evolving SourcingXPress’s AI-powered matching and recruitment technology in-house.
The team combines recruitment-domain expertise with engineering and AI capabilities, enabling us to build, test and continuously improve our AI systems based on real-world hiring data and recruiter feedback. |
We currently use first-party data generated within the SourcingXPress platform, including job descriptions, candidate profiles, applications, job–candidate matching signals and recruiter interactions. Our platform currently has 500K+ tech candidates, 2,000+ jobs and 1,000+ companies/recruiters, creating a growing proprietary recruitment dataset.
The core platform data is owned/generated through our platform, subject to applicable candidate consent, privacy and data-protection requirements. We also use technology infrastructure and databases such as Azure and Neo4j to process and manage the data.
We do not rely on purchasing large third-party candidate datasets as the core of our AI system; our long-term advantage is the proprietary data generated through actual recruitment activity on SourcingXPress. |
We will improve the AI through a continuous feedback loop from real recruitment activity. Recruiter interactions, candidate applications, shortlists, unlocks and hiring outcomes will provide signals to evaluate and refine job–candidate matching.
Our approach will focus on:
Real-world feedback: Learn from recruiter actions and outcomes.
Match-quality evaluation: Continuously measure Match% and skill-level relevance.
Human-in-the-loop validation: Use recruiter feedback to identify incorrect or missed matches.
Data enrichment: Improve the quality and freshness of our growing candidate and job dataset.
AI/ML/LLM iteration: Continuously improve matching and recommendation capabilities as the platform scales.
With 500K+ candidates, 2,000+ jobs and 1,000+ companies/recruiters, the increasing volume of real hiring interactions gives us a strong feedback loop for model improvement.
Our goal is to make every hiring interaction a learning signal that improves the next match. |
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/rDfq989XSIMf-Pitch-deck-Dec-SISF.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/rDfq989XSIMf-Pitch-deck-Dec-SISF.pdf |
https://www.youtube.com/watch?v=HuHC6_u-5n4 |
Yes. SourcingXPress aims to make hiring more accessible, transparent and outcome-driven for both companies and job seekers.
We are addressing the inefficiency of traditional recruitment by helping companies identify relevant talent faster while giving candidates better visibility and opportunities to be discovered based on their skills and fit, not just keywords, pedigree or access to expensive hiring platforms.
The platform is already creating measurable impact: 500K+ tech candidates, 2,000+ jobs and 1,000+ companies/recruiters are part of the ecosystem. Our broader mission is to build a more merit- and skills-driven talent ecosystem, where technology reduces hiring barriers and connects the right people to the right opportunities.
In short: We want to make talent discovery more equitable, efficient and accessible, not just another recruitment marketplace. |
NA |
I got an email |
checked |
| Sep 10, 2026 @ 12:39 PM |
Bryan Rodrigues |
bryan@singleaxis.ai |
https://www.linkedin.com/in/bryanrodrigues123 |
http://www.singleaxis.ai |
7975563057 |
Bryan - Software Engineer, Been working with ai implementation for the past 2 years, Chrysilla - Software Engineer working at netflix, graducated in cs and stats from uiuc |
We are siblings and have been working on side projects over a long time |
one |
We are both technical and working with ai systems for a long time we perfectly understand what the bottlenecks to wide spread adoption would be, hence we are we are building the infrastructure that will help us integrate with it |
SingleAxis |
www.singleaxis.ai |
Bangalore |
Building the assurance layer for ai deployments in critical workflows like healthcare |
Deploy ai needs constant monitoring evaluation and oversight. Take for example an ambient scripe, most hospitals use! These days they are becoming more and more capable and integrate and connect to different systems, ensuring these systems are implemented correctly, their outputs are accurate and they are reliable is an ongoing challenge that we intend to solve, the variation in systems even differs based on healthcare departments - a cliniocal ai assistant deployed in a cardiology workflow might not perform well in a oncology workflow. |
We are building a platform to 1. understand workflows for the healthcare deployment use case in and out 2. building the needed guardrails in the workflow 3. route the edge cases to humans on the platform to interact with and review. |
Aggregating large number of specialists with the mission of safeguard ai deployments globally is a non trivial task. |
two |
Pilots |
We are currently targeting hospitals in the us deploying ai systems and large ai labs looking to train these systems. |
50 billion |
20 billion |
5 billion |
we charge a retainer and a ongoing monitoring fee per deployment |
the data labelling companies might pivot to provide human in to loop service. consultancy companies like mckenzie deloitte will enter the space |
we are attending converences in usa where the top directors of all hospitals attend to build relations, we will be publishing research papers and get featured in magazines |
1. confernces 2. research papers 3. magazines |
assurance and monitoring layer for global ai deployments and colllect enough data measure control and underwrite these ai deployments |
private limited |
NA |
one |
To help increase my traction and get mentorship and guidance |
We want to get our first pilot done and onboard our first customer |
yes |
one |
one |
We are building the monitoring architecture for ai deployments |
We are building the datasets needed to understand an end to end workflow and where do they make mistakes and build the necessary guardrails at those points |
mapping out each and every work flow and building the infrastructure to monitor them is extremely defensible. replacing a vendor would be nearly impossible and we tending towards compliance necessitating this aspect |
we created our internal benchmarks for the same |
this is a challenge and we haven't yet figured it out fully, currently we are aligned to what is necessary from a hipaa perspective |
would only benefit us |
scaling humans in the loop would be the challenge |
Both the founders have enough experience and depth needed to engineer and manage an end to end ai deployment. |
we are working with some public dataset for clinical ai and transcription, and we are licensed to use them |
only way to improving it is collecting data and fine tuning it. |
USA, Europe |
|
|
|
|
|
|
| Sep 9, 2026 @ 3:42 PM |
Raghavendra V |
Raghu@towner.taxi |
https://www.linkedin.com/in/raghavendra-yadav-towner/ |
https://towner.taxi |
9739758870 |
Raghu Yadav is the Founder and CEO of Towner, leading the company’s vision to build a driver-first digital operating platform for India’s independent commercial drivers. He is responsible for product strategy, driver and union partnerships, business development, fundraising, market expansion, and building Towner’s long-term technology and financial-infrastructure roadmap. Through direct engagement with commercial drivers, he identified a key gap: drivers have multiple income sources and daily operating expenses, but lack a simple system to manage their complete business. Raghu is building Towner to help drivers digitise trips, kilometres, earnings, expenses, payments, customer records, and invoices—enabling them to operate with greater visibility, independence, and professionalism. As Founder & CEO, he is focused on scaling Towner’s recurring SaaS model, strengthening union-led distribution, developing the AI-enabled Driver Business OS, and expanding the platform across India’s commercial-driver ecosystem. |
YC Cofounder match |
1 |
Our biggest strength is our deep, practical understanding of commercial drivers and our ability to convert that understanding into a simple, useful product.
We are not building Towner from a distance. Our team works closely with drivers, driver unions, and mobility stakeholders to understand the complete reality of a driver’s working day—offline trips, direct customers, multiple ride platforms, cash payments, fuel costs, expenses, and daily income uncertainty.
This has helped us build a driver-first platform that addresses real operational problems, not just a theoretical technology use case.
Our second strength is execution. We have taken Towner from an idea to a live product with 54,000+ registered drivers, 238,000+ trips recorded, six union partnerships, and 100,000+ app installs.
The combination of driver empathy, field-level distribution, product execution, and a clear vision for AI-enabled driver business intelligence is our strongest advantage. |
Towner Solutions Pvt Ltd |
https://towner.taxi |
Bengaluru |
Towner is an AI-enabled Driver Business OS that helps commercial drivers manage trips, earnings, expenses, payments, and invoices in one place. |
Towner solves the lack of a single digital system for commercial drivers to manage their complete business—trips, income, expenses, payments, and profitability. |
Towner is an AI-enabled Driver Business OS for independent commercial drivers. It brings trips, kilometres, digital-meter activity, income, expenses, customer payments and invoices into one simple mobile platform. Unlike ride-hailing apps that see only rides on their own network, Towner gives drivers a complete view of their daily business, helping them track profitability, manage records and make better operating decisions. Our affordable flat-fee model lets drivers retain 100% of their fares. |
Towner is unique because it is built around the driver’s complete business, not only rides completed on one platform. It captures offline trips, direct customers, travel-agency and corporate bookings, cash and digital payments, kilometres, income, expenses and invoices in one workflow. Our affordable flat-fee model aligns with drivers: they retain 100% of fares, while Towner succeeds only when drivers use the product consistently. Our driver-first product insight, union-led distribution partnerships, daily-use workflow and privacy-safe business-data intelligence create a defensible foundation for an AI-enabled Driver Business OS. |
Revenue |
Users, Revenue, Pilots, Signups, Testimonials |
Our primary target customers are independent commercial taxi drivers in India, especially drivers operating in urban and airport mobility, local taxi services, corporate travel, travel-agency bookings, and intercity transport. Our secondary customers are driver unions, taxi associations, fleet operators, travel agencies, and mobility partners that want to digitise driver operations, improve visibility, and support their driver communities through a scalable SaaS platform. |
₹3,000–3,300 crore annual SaaS opportunity in India. |
₹300–450 crore annual SaaS opportunity. |
₹6–9 crore annual recurring-revenue opportunity over the next 24–36 months. |
Towner operates a commission-free SaaS model for commercial drivers. Our primary revenue comes from an affordable ₹20 per active day subscription fee, |
Ride-hailing platforms: Ola, Uber, Rapido, Namma Yatri and other ride-booking platforms Fleet-management platforms: LocoNav, Fleetx, Onelap and similar fleet-operating tools Generic business tools: Excel, paper notebooks, WhatsApp, accounting apps and manual invoice records Digital meter, trip-log and invoicing apps used by individual commercial drivers Towner is differentiated because it is focused on the driver’s complete business, including offline rides, direct customers, travel-agency bookings, corporate trips, cash payments and multi-platform work—not only rides within one marketplace. |
Driver-union and taxi-association partnerships Field onboarding and driver-community engagement Driver referrals and word-of-mouth adoption Google Play Store and organic app discovery Fleet, travel-agency and corporate mobility partnerships Local mobility hubs, including airport, railway-station and taxi-stand ecosystems Digital community engagement through WhatsApp, LinkedIn and driver-focused content |
Deepen Bengaluru: Improve activation, paid conversion, retention and daily product usage among existing drivers. Scale through unions: Partner with driver unions, taxi associations and fleet communities to onboard drivers in clusters. Create daily product value: Drive regular use through trip recording, income tracking, expenses, payments, invoices and digital-meter workflows. Convert usage into subscriptions: Use the affordable ₹20 active-day model to create recurring SaaS revenue. Expand city by city: Launch in Ahmedabad and other selected high-density commercial-driver markets through local partnerships. Build AI-enabled business insights: Introduce simple, actionable driver business intelligence to improve retention and long-term value. |
Towner’s long-term vision is to become the trusted digital and financial operating system for India’s commercial drivers and gig workforce.
We aim to help every driver manage their complete business—from trips, kilometres, income, expenses, customers, payments and invoices to actionable business insights and, over time, access to relevant financial and protection products through regulated partners.
Towner will not be limited to ride booking. We aim to build the infrastructure that helps independent drivers operate with the visibility, discipline and confidence of successful small-business owners. |
Towner Solutions Private Limited is a registered private limited company in India, headquartered in Bengaluru, Karnataka. |
60 Lakhs from angel investors |
Yes |
Towner has built a live driver-first SaaS product with early adoption, and we are now ready to build the next AI-enabled layer of our Driver Business OS. We are applying to Avinya because ACSEL’s focus on AI innovation, expert mentorship, Karnataka’s startup ecosystem and investor readiness strongly aligns with our next phase. We want support in converting driver trip, kilometre, earnings, expense and payment data into privacy-safe, actionable business intelligence for commercial drivers. Avinya can help us strengthen our AI product strategy, data architecture, responsible data practices, product-market fit, GTM execution and fundraising readiness. We also value the opportunity to learn from mentors, collaborate with fellow founders and present Towner to investors at Demo Day. |
Build and validate an AI-enabled Driver Business Assistant prototype. Strengthen our privacy-safe data architecture and driver analytics foundation. Improve driver activation, recurring subscription conversion and retention. Run a structured pilot with drivers and union partners to measure product value. Create a clear KPI dashboard, financial model and investor-ready data room. Refine our city-expansion and union-led go-to-market strategy. Prepare Towner for a successful angel / pre-seed fundraising round and Demo Day. |
YES |
No |
Supporting feature |
Training and support |
user business behaviour |
Knowladge and exicution |
already traction speaking |
DPDP |
No |
NA |
1 person |
NA |
Start with a limited pilot group of drivers and union partners.
Compare AI-generated summaries and recommendations against source records and user corrections.
Measure driver usage, output accuracy, satisfaction, error rate and repeat engagement.
Improve business rules, data validation, prompts, model selection and multilingual performance.
Expand gradually only after the product meets predefined reliability and safety standards.
Maintain human oversight and user correction options for important business records. |
YES, “We are actively building and validating an AI-enabled Driver Business Assistant.” |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/B8CUTws114k1-Towner___Investor_Pitch__Aug_2026_F3.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/B8CUTws114k1-Towner___Investor_Pitch__Aug_2026_F3.pdf |
https://youtu.be/EOGQCCstYk8?si=KIXTeSSKdBG2gOMm |
es. Towner is a mission-driven startup focused on the digital and economic empowerment of independent commercial drivers.
Commercial drivers are essential to India’s mobility and gig economy, yet many operate without clear business records, earnings visibility, expense tracking, invoice tools, or access to organised digital systems.
Towner helps drivers manage trips, kilometres, income, expenses, payments and invoices in one place. This enables greater transparency, better daily decision-making and stronger business independence.
Our commission-free model is central to our impact: drivers retain 100% of their fare income, while Towner charges an affordable daily software fee. We succeed only when drivers find continuing value in the platform.
Through driver-first technology, union partnerships and future AI-enabled business insights, Towner aims to become the digital and financial infrastructure layer that helps commercial drivers run more organised, resilient and sustainable businesses. |
Yes. I am a first-generation entrepreneur building a technology startup focused on empowering independent commercial drivers. My journey is rooted in understanding the realities of India’s working and mobility communities, which are often underrepresented in mainstream technology and startup ecosystems. |
Email |
checked |
| Sep 9, 2026 @ 2:18 PM |
Prasad Desai |
prasad@adis.co.in |
https://www.linkedin.com/in/prasad-desai-a532734b?utm_source=share_via&utm_content=profile&utm_medium=member_ios |
https://adis.co.in/ |
8792365170 |
I am Prasad Desai, Co-Founder of ADIS Technologies Pvt. Ltd. I lead the technology, product development, and overall execution of our digital solutions. My background is in technology and product development, and I have been actively working on building digital solutions for the livestock sector. At ADIS, I focus on developing mobile applications, cloud-based platforms, animal biometric identification, AI-based solutions, IoT integration, and digital traceability systems. My co-founder, Sujit Hukkerikar, focuses on business strategy, partnerships, and business development. Together, we combine technology and business expertise to build practical and scalable solutions for the livestock ecosystem. Our goal is to create a trusted digital infrastructure for animals, covering digital identity, health records, traceability, insurance, valuation, and other livestock-related services. |
Sujit and I came together through our shared interest in building technology-driven solutions for real-world problems. We identified that the livestock sector has a significant gap in digital infrastructure, particularly in areas such as animal identification, health records, traceability, and insurance.
We have been working together for several years and have built ADIS Technologies by combining our respective strengths in technology, product development, business strategy, and partnerships. Over the years, we have worked closely on developing and deploying solutions for cattle, sheep, goats, and the broader livestock ecosystem. |
All |
Our biggest strength as a team is that we complement each other well. I focus primarily on technology, product development, and execution, while Sujit focuses on business strategy, partnerships, and market development.
We also have a strong understanding of the real challenges faced by farmers and other stakeholders in the livestock ecosystem. Instead of just developing technology, we focus on building practical solutions that can work in real field conditions.
Another strength is our ability to build and execute. We have developed working digital platforms and mobile applications and continue to improve them based on real user feedback and market requirements. Our combination of technology expertise, domain understanding, and business execution helps us build scalable solutions for the livestock sector. |
ADIS Technologies Pvt. Ltd. |
https://adis.co.in/ |
Belagavi, Karnataka, India |
ADIS Technologies is building a digital identity and technology infrastructure for livestock, enabling animal identification, health management, traceability, insurance, and other livestock-related services. |
The livestock sector still lacks reliable digital infrastructure for identifying and managing individual animals. Most animals do not have a trusted and permanent digital identity, making it difficult to verify ownership, maintain health and breeding records, track animal history, and provide services such as insurance and financing. Farmers, veterinarians, insurance companies, financial institutions, and government agencies often work with fragmented or manual records. This leads to inefficiencies, lack of trust, insurance fraud, delayed claim processing, and limited access to data-driven livestock services. We are solving this problem by creating a digital ecosystem that connects the animal, farmer, service providers, and institutions through a trusted digital identity. |
ADIS Technologies provides a digital platform for livestock management and identification.
Our platform enables the registration of individual animals using biometric and visual identification technology. Once registered, the animal can have a digital profile containing important information such as ownership, health history, vaccination records, breeding records, milk records, insurance information, and other lifecycle data.
We provide mobile and cloud-based applications for farmers, field staff, veterinarians, and organizations. Our technology can also integrate AI, animal biometrics, IoT devices, and digital traceability systems.Our long-term vision is to create an Aadhaar-like digital identity infrastructure for animals that can support multiple services including livestock management, insurance, financing, traceability, valuation, and government programs. |
What makes ADIS unique is our focus on building a complete digital infrastructure around the individual animal rather than offering only a livestock management application. Our platform combines animal biometric identification, digital identity, mobile applications, cloud infrastructure, health and lifecycle records, traceability, insurance integration, and IoT capabilities. The animal biometric identification and continuously growing animal data create a strong technology and data advantage over time. As more animals are registered and their lifecycle information is recorded, the platform becomes more valuable for farmers, insurers, financial institutions, governments, and other stakeholders. We have also built our solutions based on real field requirements and have experience working with livestock ecosystems across different animal categories and markets. |
Revenue |
Revenue, Pilots |
Our target customers include livestock farmers, dairy farms, sheep and goat farms, livestock cooperatives, veterinary service providers, insurance companies, banks and financial institutions, government departments, and organizations working in the livestock sector. We primarily work through B2B, B2G, and institutional partnerships, while our mobile applications also provide direct value to individual farmers and animal owners. Our focus is on organizations that require reliable animal identification, digital records, traceability, livestock management, insurance verification, and data-driven decision-making. |
$343 Million – Indian livestock insurance market covering 512M+ livestock |
$152 Million – India and emerging livestock markets, targeting farmers, dairy farms, government programs, veterinary networks, insurance companies, and dairy organizations |
$76.6 Million+ – Initial focus on India through government programs, institutional partnerships, livestock organizations, and commercial farms |
Our revenue model is based on a combination of technology licensing, SaaS subscriptions, implementation projects, transaction-based services, and enterprise partnerships. We generate revenue by providing our digital livestock platform to organizations such as insurance companies, government agencies, livestock enterprises, and ecosystem partners. Revenue streams include animal registration and identification services, platform licensing, software subscriptions, implementation and customization services, data and analytics services, and transaction-based fees for services delivered through the platform. As the platform grows, we also see opportunities in insurance, financing, animal valuation, traceability, and other digital services built around the animal's digital identity. |
Our competitors include livestock management software companies, animal identification and tagging providers, livestock traceability platforms, and emerging agritech companies working in animal health and farm management. However, most existing solutions focus on a single part of the ecosystem, such as farm management, physical tagging, insurance, or animal health records. ADIS differentiates itself by building an integrated platform around a trusted digital identity for each animal, combining biometric identification, lifecycle records, traceability, insurance, IoT, and digital services within one ecosystem. |
We acquire customers through direct business development, institutional partnerships, pilot projects, government and industry engagements, referrals, and participation in livestock and technology ecosystems. Our primary approach is to work with organizations that have access to large livestock populations, such as insurance companies, cooperatives, dairy organizations, government departments, livestock enterprises, and veterinary networks. Once deployed within an organization or ecosystem, the platform can scale through farmer onboarding and field-level adoption. |
Our go-to-market strategy is based on strategic partnerships, technology integration, and large-scale pilots. We focus on B2B and B2G partnerships with insurance companies, banks, government departments, livestock organizations, and other ecosystem stakeholders. We enable seamless adoption through APIs and SDKs that integrate with existing systems with minimal disruption. We start with pilot projects to validate our technology and demonstrate value, and then scale successful deployments across larger animal populations. Our experience with large-scale livestock initiatives, including projects in Saudi Arabia connected with UN FAO, provides a strong foundation for expanding ADIS across India and international markets. |
Our long-term vision is to build a trusted digital infrastructure for the global livestock ecosystem.
We envision every animal having a unique and verifiable digital identity, similar to a digital identity system, which can connect its ownership, health, breeding, productivity, insurance, traceability, valuation, and lifecycle information.
Through this infrastructure, ADIS aims to enable a connected ecosystem where farmers, veterinarians, insurers, financial institutions, governments, and service providers can securely access and use verified animal information.
Our goal is to become a leading global platform for animal digital identity and livestock intelligence, starting with livestock and expanding our technology across different animal ecosystems and international markets. |
Private Limited Company – Incorporated |
₹2.29 Cr+ — External funding secured through grants, challenge programs, and innovation support. |
Yes |
We are applying to Avinya because we believe ADIS is at a stage where the right mentorship, industry connections, and strategic support can help us scale faster. We have already built and deployed our technology for livestock identification and management and are working towards expanding across larger livestock populations and international markets. We want to strengthen our business model, go-to-market strategy, partnerships, and investment readiness. Avinya's incubation ecosystem can help us connect with the right mentors, industry partners, investors, and institutions while giving us the support needed to scale ADIS into a larger livestock technology platform. |
During the programme, we want to strengthen our product and business strategy, expand our B2B and B2G partnerships, and develop a clear roadmap for scaling ADIS across India and international markets. We want to establish strategic partnerships with insurance companies, banks, government organizations, livestock organizations, and other ecosystem stakeholders. We also want to improve our go-to-market strategy, validate our business model, prepare for future fundraising, and identify opportunities for large-scale deployments. Our ultimate goal is to use the programme to accelerate ADIS from successful pilots and deployments into a scalable digital infrastructure platform for the livestock ecosystem. |
Yes |
Yes |
Core engine |
Our AI architecture combines computer vision, animal biometric identification, machine learning models, and cloud-based data processing.
For animal registration, our computer vision models analyze images captured through our mobile applications to identify and match individual animals. We use object detection and biometric feature extraction to generate animal-specific representations and compare them against existing records.
The AI layer is integrated with our mobile and cloud platforms, allowing animal identification and matching to work as part of the overall digital identity and livestock management workflow. We are also developing AI capabilities for livestock health insights, monitoring, and decision support using animal, health, and IoT-generated data. |
Our key data advantage comes from the structured animal-level datasets generated through real-world livestock registration and management.
For each registered animal, we can build a longitudinal record containing images, biometric information, health records, breeding history, milk records, ownership information, insurance information, and other lifecycle events.
As the number of registered animals increases, this continuously growing dataset can be used to improve identification accuracy, develop livestock intelligence models, and create better decision-support solutions. Our advantage is not only the volume of data but also the linkage of visual and biometric information with real-world animal lifecycle data. |
Our defensibility comes from the combination of proprietary animal data, biometric identification technology, real-world field validation, and integration with the animal's complete digital lifecycle.
Generic computer vision models can be replicated, but building a reliable animal biometric system requires domain-specific datasets, continuous field validation, model improvement, and integration with real livestock workflows.
As more animals are registered and their lifecycle data accumulates, our models and datasets become increasingly difficult to replicate, creating a data and domain-specific technology advantage. |
We evaluate our AI models using standard computer vision and biometric identification metrics, including accuracy, precision, recall, F1-score, false acceptance rate, false rejection rate, and matching similarity scores.
For animal identification, we evaluate both detection performance and biometric matching performance using labeled datasets and real-world field images. We continuously monitor model performance across different breeds, lighting conditions, camera angles, image quality, and field environments.
We also use human verification and field validation as an additional layer for cases where the model confidence is below the required threshold. |
We follow a privacy-by-design approach for our platform. Access to animal and user data is controlled through authentication, authorization, and role-based access mechanisms.
Data is transmitted securely between applications and our backend infrastructure, and access to sensitive information is restricted based on user roles and organizational requirements.
We also follow applicable Indian data protection and cybersecurity requirements and design our platform to support the specific privacy and data governance requirements of government, enterprise, and international deployments. |
Yes. Our product operates across livestock, animal health, insurance, data, and government ecosystems, so regulatory requirements may vary by country and use case.
Key considerations include data protection and privacy, government data policies, animal identification standards, veterinary regulations, insurance regulations, and cross-border data requirements.
We address these risks by working with relevant stakeholders and adapting our technology and data practices to the regulatory requirements of each market. |
At 10x scale, the main challenges would be computational costs, image processing capacity, data storage, model inference latency, and maintaining consistent identification accuracy across a much larger and more diverse dataset.
Our architecture is designed to address this through cloud-based infrastructure, scalable APIs, optimized model inference, data pipelines, and continuous monitoring. We also see opportunities to perform more processing at the edge or on mobile devices where appropriate to reduce cloud dependency and latency. |
Yes. We have in-house technical expertise covering AI/ML, computer vision, mobile application development, backend systems, and data engineering.
Our team has experience working with animal image datasets, object detection, biometric feature extraction, image matching, model deployment, and integrating AI models into mobile and cloud-based applications.
We combine AI/ML expertise with livestock domain knowledge, which helps us develop models specifically for real-world animal identification and livestock applications. |
We primarily work with datasets generated through our own livestock registration and field deployments, including animal images and associated animal-level records.
These datasets are collected through our applications and projects with relevant users and partner organizations, subject to applicable permissions and data-use requirements. We also use publicly available datasets and standard benchmark datasets where appropriate for model development and validation.
Our long-term objective is to build a large, high-quality, domain-specific animal dataset through our own deployments and partnerships. |
We continuously improve our models by collecting additional field data, identifying difficult and low-confidence cases, retraining models, and validating performance across different breeds, environments, and image conditions.
We use feedback from field users and human verification to identify incorrect or uncertain predictions. This creates a continuous feedback loop between deployment, data collection, validation, and model improvement.
As our animal database grows, we expect the increasing diversity and volume of real-world data to further improve model robustness and accuracy. |
We are building for both India and global markets.
India is our primary market for developing and scaling the technology because of its large livestock population and the significant need for digital animal identity and traceability.
At the same time, our platform is designed for international deployment. We have already worked on livestock-related use cases outside India, including projects in Saudi Arabia, and we see significant opportunities across emerging livestock markets globally. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/IWM2x6cJz6GC-ADIS-Pitch-Deck.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/IWM2x6cJz6GC-ADIS-Pitch-Deck.pptx |
https://drive.google.com/drive/folders/1IVbpcpoB6dOfdMieOnGbm9e1l6tYoHc5?usp=sharing |
Yes. ADIS Technologies is building technology to address real challenges faced by livestock farmers and the broader livestock ecosystem.
Our mission is to create trusted digital identity and infrastructure for animals, enabling better access to animal health records, traceability, insurance, valuation, financing, and other livestock services.
By bringing animal-level data and digital identity into the livestock ecosystem, we aim to reduce losses, improve transparency, support farmers, strengthen livestock insurance, and help government and other institutions deliver services more effectively.
Our long-term goal is to create a scalable digital infrastructure that can improve the productivity, financial security, and welfare of livestock and the communities dependent on them. |
NA |
IIIT Bangalore |
checked |
| Sep 9, 2026 @ 2:08 PM |
Arvind Sethi |
arvind.sethi@dualsafe.in |
https://www.linkedin.com/in/arvind-sethi-9a50444/ |
http://www.dualsafe.in |
9900617790 |
Co-Founder & CEO |
Married with two boys - both working now |
1 |
Sales, marketing, product design, critical thinking, finance, and Operations |
DualSafe Private Limited |
http://www.DualSafe.in |
Bangalore |
DualSafe’s patented Dual HeadGear integrates two helmets into one, improving adoption, storage, and retention. Featuring Airy ventilation, MIPS protection, and IoT alerts, it enhances rider safety, focus, and comfort—supporting CSR goals of injury reduction and safer mobility. |
Carrying/Keeping even one helmet on a two-wheeler is inconvenient, let alone two |
The PATENTED Dual HeadGear integrates two helmets as ONE, ensuring safety for both the rider and their companion, equipped with advanced IoT features such as alarms for improper locking and retention mechanisms, intrahelmet communication for uninterrupted road focus, and connectivity to mobile for music, GPS instructions, etc., via Bluetooth, RF, or NFC protocols. |
Patented in India, the UK and the USA after PCT clearances on 10 claims for design and Utility |
MVP |
Pilots |
Two-wheeler OEMs, Taxi aggregators, Rental organizations and College university students where we may get bulk orders / long-term strategic engagements for business |
$6.9 Billion |
$4.4 Billion |
$727.38 Million |
B2B; organic flow to B2B2C, B2G and once threshold of 2.00 Lakh annual units number achieved switch to B2C/D2C |
Studda, Vega and Steelbird |
NDAs in place with 2-W OEMs; outside India, strategic engagements with 2-W OEMs and rental organizations |
Start with B2B engagements with 2-W OEMs, Taxi aggregators, Rental organizations and Delivery services organization, Organic flow to B2B2C; Then B2G and eventually when threhold of 1.5 lakh units (domestic and global sales) are sold consistently for 2-3 quarters switch to B2C and D2C also. Royalty engagements with existing helmet manufacturers in EMEA and Americas regions |
Have DualSafe brand as market disruptor, considering these are only safety gadgets offering Comfort, Convenience and tech enablement without compromising safety |
Private Limied |
INR 3000000 |
Yes |
Industry engagements and mentoring apart from funding of INR 1 Crore |
Mold completion and soft launch after due certification |
Virtual |
Yes |
Supporting feature |
AR / VR/ ML and AI as embedded PCB would have an LLM defined to capture lots of information using sensors, cameras, etc to assess the cause of accident/damage and inform before the same happens |
Association with insurance companies; organizations like Rapido, Flipkart, etc have shared use cases for monitoring/reporting, etc |
NA |
Not as of now; furture possibilities |
Confined within private data centers |
BIS (ISI certification) |
NA |
Not as of now |
Private data center only |
Not planned yet |
Global Markets |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/nJfNVKU4yjND-DualSafe-DECK.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/nJfNVKU4yjND-DualSafe-DECK.pdf |
https://drive.google.com/file/d/1QMEL8R0gVJ7Fit6Wt0pslw4mLeeuVRnF/view?usp=drive_link |
Both; Driving Helmet Compliance Through Adoption-Led Innovation
• Assured Safety for Rider & Pillion: One patented helmet splits into two ISI-compliant helmets, removing barriers of carrying, storage, and availability.
• Measurable Safety & Economic Impact: Improves helmet usage, reduces non-compliance, lowers ownership cost, and supports domestic manufacturing and jobs.
• Policy-Aligned & Scalable: Directly supports Motor Vehicles Act provisions, BIS standards, and the two-helmet mandate with a Make-in-India solution. |
AN |
info@acsel.in |
checked |
| Sep 9, 2026 @ 1:51 PM |
Lata Vishwanath |
latav@altiussuryatech.in |
|
http://www.altiussuryatech.in |
7022409311 |
Managing Director |
Managing Partner is an alumnus of IIT Madras |
two |
AI Application and the founder is completing a DBA program on emerging technology from Golden Gate University and we are committed to applying to rural India, farmers uplift using technology. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Sep 9, 2026 @ 1:39 PM |
Prasad Desai & Sujit Hukkerikar |
prasad@adis.co.in |
https://www.linkedin.com/in/prasad-desai-a532734b?utm_source=share_via&utm_content=profile&utm_medium=member_ios |
https://adis.co.in/ |
8792365170 |
I am Prasad Desai, Co-Founder of ADIS Technologies Pvt. Ltd. I lead the technology, product development, and overall execution of our digital solutions. My background is in technology and product development, and I have been actively working on building digital solutions for the livestock sector. At ADIS, I focus on developing mobile applications, cloud-based platforms, animal biometric identification, AI-based solutions, IoT integration, and digital traceability systems. My co-founder, Sujit Hukkerikar, focuses on business strategy, partnerships, and business development. Together, we combine technology and business expertise to build practical and scalable solutions for the livestock ecosystem. Our goal is to create a trusted digital infrastructure for animals, covering digital identity, health records, traceability, insurance, valuation, and other livestock-related services. |
Sujit and I came together through our shared interest in building technology-driven solutions for real-world problems. We identified that the livestock sector has a significant gap in digital infrastructure, particularly in areas such as animal identification, health records, traceability, and insurance.
We have been working together for several years and have built ADIS Technologies by combining our respective strengths in technology, product development, business strategy, and partnerships. Over the years, we have worked closely on developing and deploying solutions for cattle, sheep, goats, and the broader livestock ecosystem.
My background is in technology and product development, and I have been actively working on building digital solutions for the livestock sector. At ADIS, I focus on developing mobile applications, cloud-based platforms, animal biometric identification, AI-based solutions, IoT integration, and digital traceability systems.
My co-founder, Sujit Hukkerikar, focuses on business strategy, partnerships, and business development. Together, we combine technology and business expertise to build practical and scalable solutions for the livestock ecosystem.
Our goal is to create a trusted digital infrastructure for animals, covering digital identity, health records, traceability, insurance, valuation, and other livestock-related services.
The founders have worked together for several years, combining their complementary strengths in technology, product development, business strategy, and ecosystem partnerships. Their long-term collaboration has helped ADIS Technologies develop and deploy digital solutions for cattle, sheep, goats, and other animal sectors. |
two |
Our biggest strength as a team is that we complement each other well. I focus primarily on technology, product development, and execution, while Sujit focuses on business strategy, partnerships, and market development.
We also have a strong understanding of the real challenges faced by farmers and other stakeholders in the livestock ecosystem. Instead of just developing technology, we focus on building practical solutions that can work in real field conditions.
Another strength is our ability to build and execute. We have developed working digital platforms and mobile applications and continue to improve them based on real user feedback and market requirements. Our combination of technology expertise, domain understanding, and business execution helps us build scalable solutions for the livestock sector. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Sep 9, 2026 @ 1:35 PM |
Sneha Kiran Manded and Uroojakhtarkhan Inamdar |
snehamanded@gmail.com |
https://www.linkedin.com/in/sneha-manded-427378271/ |
https://www.ayeenihealthtech.com/ |
7411730876 |
Co Founder |
We met in college and have been working together for two years |
All |
Our biggest strength is our seamless communication and high level of trust, which we built back in college. Because we’ve worked closely for two years, we understand each other’s working styles perfectly, allowing us to divide tasks efficiently and solve complex problems without friction. |
Ayeeni Healthtech |
https://www.ayeenihealthtech.com/ |
Belagavi |
Ayeeni HealthTech is transforming healthcare through intelligent, technology-driven solutions that improve accessibility and efficiency. |
We are solving the problem of fragmented, inaccessible, and inefficient healthcare by using technology to connect people with the right healthcare resources, information, and services in a simpler and more affordable way. |
Ayeeni HealthTech is building an AI-powered digital healthcare platform that connects patients with relevant healthcare services, simplifies access to medical information, and enables healthcare providers to deliver more efficient, personalized, and accessible care. |
Our solution combines AI-driven personalization, an integrated healthcare ecosystem, and a user-centric approach to make healthcare more accessible and efficient. Our growing domain expertise, proprietary workflows, and data-driven insights create a foundation for a scalable and defensible platform. |
MVP |
Pilots |
Hospitals and doctors |
$20B |
$5B |
$50M |
B2B SaaS subscription model with tiered pricing for individual doctors, clinics, hospitals, and enterprise hospital chains, complemented by AI workflow modules, API/AI licensing, and implementation services. The deck projects revenue mix of 65% SaaS subscriptions, 20% AI workflow modules, 10% API licensing, and 5% services. |
Practo, HealthPlix, Eka Care, Apollo 24/7, and traditional Hospital Management Systems (HMS). MEDORA differentiates through AI-native workflow automation, connected hospital operations, and end-to-end care orchestration. |
We plan to acquire customers through direct B2B sales and pilot deployments with multi-specialty clinics, small and mid-sized hospitals, diagnostic centers, hospital chains, and medical colleges, converting successful pilots into long-term paying customers. |
We will start with focused pilot deployments in clinics and hospitals, validate workflow efficiency and adoption, and convert successful pilots into paying customers. We will then expand through enterprise deployments across hospitals and healthcare providers, supported by a tiered SaaS pricing model. |
Our vision is to build the AI-native operating system for healthcare—connecting patients, doctors, hospitals, laboratories, and pharmacies into one intelligent ecosystem that understands, automates, and orchestrates the complete circle of care. |
Done |
100000 |
Yes |
We are applying to Avinya because MEDORA is at a stage where the right mentorship, healthcare industry connections, and AI ecosystem can significantly accelerate our journey from product development to real-world deployment. Avinya’s strong AI focus, IIT alumni network, domain mentorship, and opportunities for pilot deployments align closely with our goal of building and validating an AI-native healthcare operating system. We also see the programme as an opportunity to learn from experienced founders and build meaningful partnerships that can help us scale responsibly. |
During the programme, we want to validate MEDORA through real-world healthcare pilots, strengthen our AI and product architecture, and refine our go-to-market and revenue strategy. We aim to work closely with healthcare and AI mentors, establish partnerships with hospitals and clinics, measure measurable improvements in workflow efficiency, and convert successful pilots into paying customers. By Demo Day, our goal is to have a stronger, validated product, clear product-market fit indicators, and a scalable roadmap for expanding MEDORA across India |
hybrid |
Yes |
Core engine |
MEDORA uses a four-layer AI architecture: Intelligent Document Processing using OCR and NLP; Contextual Medical Memory that builds a longitudinal patient intelligence graph; Intelligent Search & Reasoning using semantic and temporal search with RAG/document-grounded responses; and Care Continuity Intelligence for post-discharge monitoring and follow-ups. These layers are integrated with speech, text, medical reports, prescriptions, and workflow automation across the healthcare ecosystem. |
We are currently building our proprietary data advantage through structured healthcare workflows and longitudinal patient intelligence generated within MEDORA. As deployments grow, this can create domain-specific, workflow-oriented datasets for improving AI performance. We do not currently claim ownership of a large proprietary medical dataset. |
Our defensibility comes from combining multimodal clinical intelligence, longitudinal patient context, healthcare workflow orchestration, and domain-specific AI workflows rather than offering a standalone chatbot or documentation tool. MEDORA's contextual medical memory and integration across patients, doctors, hospitals, labs, and pharmacies create increasing workflow and data intelligence advantages as adoption grows |
We evaluate performance at both the model and workflow levels, including extraction accuracy, documentation accuracy, intent/triage accuracy, retrieval relevance, response grounding, latency, and human/clinician acceptance. For clinical workflows, outputs remain subject to appropriate human oversight rather than relying solely on automated model decisions. |
MEDORA is being designed with healthcare data privacy and security as core requirements. Our roadmap includes secure cloud infrastructure, controlled access, data protection, auditability, and compliance-oriented architecture, with security and scalability being strengthened through pilot and production deployments. |
Yes. As MEDORA operates in healthcare and processes clinical information, key risks include patient-data privacy, cybersecurity, health-data governance, interoperability requirements, and regulations governing AI-assisted clinical workflows. We plan to mitigate these through human oversight, secure architecture, appropriate consent and access controls, and compliance-by-design as the platform moves toward production. |
At 10x scale, the primary challenges would be inference cost and latency, multimodal processing throughput, retrieval and patient-memory scalability, data synchronization across healthcare systems, and maintaining reliability across diverse clinical workflows. Our architecture is therefore being designed around scalable cloud infrastructure, modular AI services, and optimized retrieval and workflow orchestration. |
Yes. Our founding team includes ML and product expertise, supported by technical and medical advisors. The founding team is focused on AI product development, healthcare workflow design, and platform engineering, with technical mentorship in AI/engineering and clinical guidance for real-world healthcare workflows. |
We are currently working primarily with healthcare workflow and clinical information required to validate MEDORA's AI capabilities, including documents, prescriptions, reports, voice/text inputs, and structured workflow data. We do not currently claim ownership of a large proprietary medical dataset; data sources and licensing/consent requirements will be handled appropriately as pilot deployments scale. |
We will continuously improve MEDORA through feedback from doctors and healthcare providers, evaluation of real-world workflow outcomes, error analysis, retrieval and prompt optimization, model evaluation, and iterative refinement of our AI workflows. Pilot deployments will provide the feedback loop needed to improve accuracy, usability, reliability, and efficiency before broader commercial deployment. |
India |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/htY2eAaS5XC1-Ayeeni_Healthtech.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/htY2eAaS5XC1-Ayeeni_Healthtech.pptx |
https://drive.google.com/drive/folders/1A2sH_0ynJtw2WBpKJXqFhYdx8Eq7SpWy?usp=drive_link |
Yes. Ayeeni HealthTech is mission-driven to make healthcare more connected, accessible, and efficient. Through MEDORA, we aim to reduce administrative burden on clinicians, eliminate fragmented healthcare workflows, improve continuity of care, and create a more seamless experience for patients. Our long-term goal is to use AI responsibly to improve healthcare delivery and outcomes at scale, particularly in underserved and resource-constrained settings. |
NA |
Cipherion Pvt Ltd |
checked |
| Sep 9, 2026 @ 11:37 AM |
Hemant Bihare |
hemantbihare@gmail.com |
|
|
07000933402 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Sep 1, 2026 @ 6:30 PM |
Dg |
gkbc |
http://hfksjd |
http://vsndn |
84648 |
Tsjdn |
Fejdn |
two |
Vsndbd |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| May 2, 2026 @ 9:46 PM |
Venkitesh Ramadas |
venkitesh@gmail.com |
https://www.linkedin.com/in/venkitesh/ |
https://nextoar.com/ |
9900214771 |
Founder and CEO. 25 years of experience in Product management, Supply chain management consulting, Manufacturing and supply chain operations. |
Founder with a strong technical team backing up. CTO and I were ex-colleagues at i2 Technologies. |
All |
1. Rare blend of business depth and technical excellence
Nextoar brings together decades of real-world experience in semiconductor, electronics, manufacturing, and supply chain operations with deep expertise in AI, distributed systems, and modern software architecture.
2. Built by operators, not just technologists
The team includes leaders who have worked inside global companies like Cisco, Infineon, Cypress, i2, Blue Yonder and Grab. This ensures solutions are grounded in real operational challenges, not just theoretical AI.
3. Strong AI and systems engineering core
From LLMs and computer vision to MLOps, distributed systems, and agentic AI, the team has hands-on capability to design and deploy production-grade AI systems, not just prototypes.
4. Proven ability to build and scale platforms
Experience spans SaaS architecture, microservices, event-driven systems, and large-scale enterprise platforms across supply chain, ALM, and business process automation.
5. Deep semiconductor, electronics and industrial domain expertise
A strong understanding of fab operations, electronics manufacturing, and supply chain complexities enables highly contextual and high-impact AI solutions.
6. Research-backed and future-facing
The team brings academic rigour from institutions like IIT Madras, IIT Bombay, and global exposure, along with contributions to standards, patents, and advanced AI methods.
7. End-to-end capability from idea to production
Nextoar combines strategy, domain understanding, AI modeling, and engineering execution to move quickly from concept to real deployment and measurable impact. |
Nextoar Technology Solutions Pvt Ltd |
https://nextoar.com/ |
Bangalore |
Nextoar builds your Everyday AI for industrial operations. AI platforms that combine deep industrial expertise with advanced physical AI intelligence to solve real manufacturing and supply chain problems at scale. |
Nextoar solves the growing complexity in industrial operations where data is fragmented, frontline knowledge is underutilized, and problem solving is slow and reactive. Enterprises spend heavily on transformation, yet most fail because they cannot connect cross-functional data or scale frontline intelligence effectively. We turn this gap into an advantage by enabling faster, context-aware decisions, proactive problem solving, and continuous operational improvement across quality, capacity, and reliability. |
Nextoar provides an AI platform purpose-built for industrial operations that brings together data, domain knowledge, and frontline expertise into a single, context-aware intelligence layer.
It combines structured factory data, engineering documents, and real-time operational inputs with domain-specialized AI to enable faster root cause analysis, predictive insights, and autonomous decision support across quality, capacity, and reliability.
The platform is designed to work inside enterprise workflows, enabling teams to collaborate with AI, continuously learn from operations, and move from reactive problem solving to proactive and self-improving systems. |
Physical AI for real-world operations Nextoar goes beyond digital AI by directly interfacing with machines, processes, and shopfloor workflows, enabling AI to observe, reason, and act in real physical environments. From visual inspection and process control to predictive maintenance and self-healing systems, the platform closes the loop between insight and action in industrial operations. Deep domain + AI, not generic AI Nextoar is built specifically for semiconductor and industrial operations, combining domain knowledge with AI. This creates high switching cost because the value comes from contextual understanding, not just models. |
Users |
Pilots |
Hi-tech industries and related supplier ecosystems - Electronics, Semiconductor, Display and Hard-drive manufacturing |
$134B |
$22B |
$1B |
Usage based - Token consumption |
Glean, UiPath, Celonis |
Direct and partner-led |
Mid-tier ($200M to $2B annual revenue) in hi-tech industrial space |
Nextoar aims to build the intelligence layer for industrial operations, where AI continuously learns from data, people, and machines to drive autonomous, self improving factories. The goal is to transform every frontline worker into a high-impact problem solver and enable organizations to move from reactive operations to predictive and eventually self-healing systems. |
Yes, incorporated in Bangalore, India |
Pre-seed |
Yes |
Explore investment opportunity and guidance |
Zero to one scaling guidance and investment |
Hybrid |
Yes |
Core engine |
Agentic AI leveraging MCP, LLMs, VLMs, knowledge graphs, semantic search, machine vision. |
Industry specialized models and context layer. |
1. Domain depth creates real differentiation
Nextoar is built for supply chain and industrial operations, not generic enterprise AI. The combination of process knowledge, failure modes, and operational context makes it difficult for horizontal AI players to match.
2. Compounding knowledge moat
Every interaction captures frontline problem solving, decisions, and outcomes, building a proprietary, context-rich knowledge layer that improves over time and is hard to replicate.
3. Deep workflow and system integration
The platform sits inside day-to-day tools and processes, becoming part of how teams operate. This creates strong stickiness and high switching costs.
4. Closed-loop Physical AI capability
Nextoar connects insight to action by interfacing with machines, inspection systems, and operations, enabling real-world impact rather than just recommendations.
5. On-prem, secure, and explainable by design
Aligned with enterprise requirements, this removes a major barrier that limits many AI competitors. |
We combine standard AI metrics with a domain-specialized evaluation framework tailored for industrial operations.
1. Core AI performance metrics
Accuracy, precision and recall for defect detection and classification, latency, and consistency across repeated queries.
2. System-level reliability metrics
Uptime, response stability, drift detection, and performance under real production workloads.
3. Domain-specialized evaluation (key differentiator)
We evaluate outputs in real operational context using four core dimensions:
Relevance
Coherence
Persistence
Prevalence
4. Outcome-driven validation
Ultimately, we measure impact through improvements in yield, OEE, MTTR, and reduction in quality incidence etc |
Nextoar is designed for data-sovereign deployments. We operate fully on-premise or within the customer’s private cloud, with no SaaS model, ensuring that sensitive industrial and production data never leaves the customer environment.
We align with enterprise security and compliance standards through strict access controls, role-based permissions, encryption, and auditability. Our architecture is built to meet the needs of highly regulated industries like semiconductors, where IP protection, traceability, and system reliability are critical. |
Regulatory risk for Nextoar is relatively low because we operate in industrial and enterprise environments, not consumer-facing AI. However, there are a few areas we actively account for:
- Data sovereignty and cross-border regulations
Increasing focus on where data is stored and processed. Our on-prem and private cloud deployment model directly mitigates this by keeping all data within the customer’s controlled environment.
- Industry compliance requirements
Sectors like semiconductors and electronics require strict controls around IP protection, traceability, and auditability. Our platform is designed to align with these expectations from day one.
- Evolving AI governance and explainability norms
Regulations around AI transparency and accountability are emerging globally. We address this through explainable, traceable AI outputs and human-in-the-loop workflows. |
1. Context and knowledge scaling
As data sources, documents, and interactions grow, maintaining high-quality retrieval and context grounding becomes harder. Without strong curation and indexing, relevance can degrade.
2. Real-time performance under load
Latency and responsiveness can be impacted when handling large volumes of concurrent queries, streaming data, and edge interactions.
3. Continuous learning and feedback loops
At scale, ensuring that feedback improves the system without introducing noise or drift becomes challenging. Model governance and evaluation discipline become critical.
4. Workflow integration depth
Deeper integration with enterprise systems (MES, quality, equipment) increases dependency management and reliability requirements. |
Yes, we have a strong in-house AI and engineering team with both deep technical capability and real industrial context.
What defines our expertise
LLMs and applied AI systems
Fine-tuning, RAG architectures, agentic workflows, and domain-adapted reasoning systems for industrial use cases.
Computer vision for manufacturing
Visual inspection, defect detection, false call reduction, and edge AI deployment for shopfloor environments.
Data and platform engineering
Distributed systems, MLOps, real-time data pipelines, and scalable on-prem or private cloud deployments.
Domain-specialized AI modeling
Building AI that understands semiconductor and electronics manufacturing processes, not just generic data patterns.
Evaluation and reliability engineering
Custom evaluation frameworks measuring relevance, coherence, persistence, and prevalence to ensure real-world performance.
End-to-end execution
From model design to production deployment and integration with enterprise systems like MES, quality, and equipment data. |
Mostly proprietary customer data, some public data available as free source is also used. |
How we continuously improve model performance
1. Closed-loop learning from real operations
Every interaction, decision, and outcome on the shopfloor feeds back into the system, helping models learn from real-world problem solving, not synthetic benchmarks.
2. Human-in-the-loop refinement
Engineers validate, correct, and guide AI outputs, ensuring improvements are grounded in domain expertise and reducing drift over time.
3. Domain-specific evaluation framework
We continuously measure performance using relevance, coherence, persistence, and prevalence, ensuring the system improves in real operational context, not just accuracy.
4. Continuous data and knowledge enrichment
We expand and refine the knowledge base by integrating new data sources, documents, and cross-functional insights across plants and use cases.
5. Iterative model and system upgrades
Regular updates to models, retrieval systems, and workflows based on performance signals, with careful versioning and rollback mechanisms. |
Both, we are building in India for both India and global markets. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/UpTCEj7n7rZB-Nextoar-ACSEL-IITAlum-v1.0.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/UpTCEj7n7rZB-Nextoar-ACSEL-IITAlum-v1.0.pdf |
https://youtu.be/4P87bjgg63I?si=f99oKTOGfZGlUPHN |
Yes, Nextoar is deeply mission-driven, with a clear focus on real industrial impact.
Our mission is to transform how work gets done in industrial environments by turning every engineer and technician into a high-impact problem solver. We aim to shift organizations from reactive operations to continuous, data-driven, and AI-enabled improvement.
We are focused on tangible outcomes, not just technology. That means improving yield, reducing downtime, increasing capacity, and driving measurable gains in efficiency, cost, and sustainability. By helping teams make better decisions faster, we directly contribute to stronger, more resilient industrial ecosystems.
At a broader level, we believe AI should not replace the frontline workforce but amplify their expertise, making advanced problem solving accessible to everyone and unlocking innovation at scale. |
NA |
Sriram - CEO of ELCIA (Electronics City Industries Association) |
checked |
| Apr 30, 2026 @ 11:57 PM |
Sandesh Sanjeev Phalke |
s.phalke2029@gmail.com |
https://www.linkedin.com/in/sandesh-phalke-ph-d-63658b137/ |
https://www.heyattrangi.com/ |
9552324069 |
We are a multidisciplinary team building Aatrangi, an AI-driven mental health infrastructure platform. Core Team: Sandesh Phalke – Founder & Managing Director Leads vision, strategy, and institutional partnerships. Focused on building structured mental health systems. Medasani Bharath Reddy – Co-founder & CEO Drives execution, operations, and business development. Responsible for scaling institutional adoption. Charan Dusary – Co-founder Computer Science student at IIIT Dharwad. Leads full-stack development, AI integration, and system architecture. Extended Team: Kalyan Jakkoju – Marketing & Growth Lakshmi Prasad – Product Strategy Harshith Daraboina – Technology Support Rashmi Khare – Design & Experience The team combines technical execution, product thinking, and market understanding to build a scalable AI-powered platform. |
We met during our B.Tech at IIIT Dharwad and connected through our shared interest in building technology-driven solutions.
We are currently in our 6th semester and have been working together since August 2025, collaborating on problem exploration, research, and developing the Aatrangi platform. |
1 |
Our biggest strength is our ability to combine research-driven thinking with fast execution.
We have validated our approach by developing our AI chatbot system and having our research accepted at the IIT Mandi MBCC conference. In parallel, we have built our product prototype and secured incubation support at IIIT Dharwad Research Park (IDRP).
This balance of research, technical development, and structured execution allows us to move quickly while building a credible and scalable solution. |
AATRANGI PRIVATE LIMITED |
https://www.heyattrangi.com/ |
Hubli - Dharwad , Karnataka , India |
We are building an AI-driven mental health infrastructure that provides continuous, structured support by combining an AI companion with therapist-led care. |
Mental health support today is fragmented, expensive, and limited to therapy sessions, leaving users without continuous care between sessions and institutions without scalable systems to support large populations. This leads to low engagement, poor outcomes, and lack of visibility for both individuals and organizations. |
We are building an AI-driven mental health platform that enables continuous and structured care by combining an AI companion with therapist-led interventions.
The platform provides a 24/7 AI companion for emotional support, mood tracking, and early risk detection. Based on user interactions and structured inputs, the system generates insights that assist therapists in delivering more effective and personalized sessions.
Therapists can assign wellness tasks, track user engagement, and monitor progress through a centralized dashboard, while users continue to receive support between sessions. This creates a continuous care loop instead of isolated therapy interactions.
The system is designed to scale across institutions, enabling colleges and organizations to provide structured mental health support to large populations efficiently and affordably. |
Our solution is built as a continuous care system rather than a session-based or chatbot-only product. We combine a 24/7 AI companion with therapist-led care, creating a structured loop where user interactions, mood data, and engagement are continuously tracked and fed back into treatment. The core defensibility comes from our data layer and workflow design. As users interact with the platform, we generate structured behavioral data that improves personalization, therapist efficiency, and long-term outcomes. This creates a compounding advantage that is difficult to replicate. We are also focused on neurodivergent and long-term care segments such as ADHD, autism, and learning disorders, which require ongoing support rather than one-time sessions. Additionally, our platform is designed for institutional deployment, enabling scalable and affordable mental health infrastructure for large populations. |
MVP |
Users, Pilots |
Primary customers are educational institutions and organizations that need scalable mental health support for large populations. Secondary customers are individuals across India, including students, working professionals, and neurodivergent users such as ADHD, autism, and learning disorder groups. Tertiary stakeholders include therapists who deliver sessions and use the platform for structured care. |
India has over 300 million students and working professionals who can benefit from structured mental health support. At an average pricing of ₹99 per user per month, the addressable market exceeds ₹3 lakh crore annually. |
We are initially focusing on educational institutions and organizations. India has over 40,000 colleges and universities, representing a serviceable market of approximately ₹30,000+ crore annually at institutional pricing. |
In the initial phase, we are targeting 10–15 institutions in the Hubli–Dharwad region, covering ~10,000 users. This represents an initial obtainable revenue of ~₹1.18 crore annually, with planned regional expansion. |
We operate on a three-stream revenue model: Institutional co-pilot: ₹99 per user per month (B2B2C model) B2C subscriptions: ₹199–₹699 per month Therapist marketplace: 18% commission per paid session |
Existing solutions fall into three categories: Therapy marketplaces (Amaha, YourDost) AI/chat-based mental health apps (Wysa, Sukoon) General wellness platforms (Headspace-type models) These are either session-based or chatbot-only systems, lacking structured continuous care. |
Direct outreach to educational institutions and organizations Pilot-based onboarding through institutional partnerships Digital acquisition for B2C users via targeted campaigns Therapist onboarding through professional networks |
Phase 1: Hubli–Dharwad institutional pilots (10–15 institutions) Phase 2: Expansion across Karnataka institutions Phase 3: Pan-India rollout with institutional and B2C channels Parallel focus on therapist onboarding to support scale. |
To build a scalable, AI-driven mental health infrastructure that enables continuous care for millions of users, reduces therapy overload, and becomes the standard system for institutional mental wellness across India. |
Registered Private Limited Company in India (Aatrangi Private Limited, incorporated January 2026) |
We have received ₹2 lakh in grant support under the Yuva Udyami program at IIIT Dharwad Research Park (IDRP). The startup is otherwise bootstrapped with no external equity funding. |
Yes |
We are applying to Avinya to strengthen our AI capabilities and build a more robust, scalable product. As an AI-focused cohort, Avinya provides the right ecosystem for technical mentorship, product refinement, and access to networks that can help us deploy our solution at scale. We are particularly looking to accelerate our AI systems, improve model performance, and structure our product for large-scale institutional adoption. |
Strengthen our AI companion system and improve model accuracy Complete full product integration and deploy institutional pilots Onboard initial users and validate engagement metrics Refine our product architecture for scalability Keep it execution-focused. |
YES |
Yes |
Core engine |
We use a hybrid architecture combining large language models with structured data pipelines.
The system includes an LLM-based conversational layer for user interaction, integrated with backend services that capture mood inputs, behavioral signals, and task engagement data.
We use prompt-engineered models with context retention to generate responses, and structured data pipelines to convert interactions into usable insights for therapists.
The backend is built using standard web technologies with API-based model integration, enabling scalable deployment and continuous improvement. |
Our advantage comes from structured behavioral data generated through continuous user interaction.
Unlike typical chatbot systems, we capture mood trends, engagement patterns, and task completion data over time, which helps improve personalization and care recommendations.
As usage grows, this longitudinal data creates a compounding advantage that is difficult to replicate. |
Our defensibility lies in the integration of AI with structured care workflows.
We are not building a generic chatbot, but a system where AI continuously interacts with users, tracks behavior, and feeds structured insights back into therapist workflows.
This creates a feedback loop between AI, user behavior, and human intervention, making the system more effective over time and harder to replicate. |
We evaluate performance using both technical and user-level metrics.
These include response relevance, user engagement rate, session continuity, and task completion rates.
We also use qualitative feedback from users and therapists to improve response quality and system effectiveness. |
We follow a privacy-first approach with secure data handling practices.
User data is stored securely, with access restricted based on roles. Sensitive information is handled carefully, and we aim to align with Indian data protection regulations.
We ensure that personal data is not exposed and is used only to improve user experience and care delivery. |
Yes, mental health and health-tech platforms involve regulatory considerations around data privacy and ethical AI use.
We mitigate this by avoiding clinical diagnosis claims, ensuring transparency in AI usage, and positioning the system as a support tool rather than a replacement for medical professionals. |
At 10x scale, challenges would include model latency, infrastructure load, and maintaining response quality across large user volumes.
We are designing the system with scalable architecture and modular components to handle increased usage efficiently. |
Yes, our team includes technical members responsible for AI integration, backend systems, and product development.
We are currently focused on building strong implementation capability and plan to expand dedicated AI/ML expertise as we scale. |
We currently use publicly available datasets and general-purpose language models for initial development.
Additionally, we generate structured interaction data from user activity on our platform, which will become a key dataset as we scale. |
We improve performance through continuous iteration using user feedback, interaction data, and prompt refinement.
As more data is generated, we will enhance personalization and response quality through better context handling and structured insights. |
We are initially focused on India, particularly Tier-2 and Tier-3 regions, where access to mental health support is limited.
Our long-term vision includes scaling to global markets with similar accessibility challenges. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/QJuUx7gakurV-Version_6.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/QJuUx7gakurV-Version_6.pdf |
https://drive.google.com/file/d/1jA4SfGHwt0b2ScnS8B3OlQ3F7_YfiHuW/view?usp=sharing |
We are building a mission-driven platform focused on improving access to mental health support in India.
Mental health care today is often expensive, fragmented, and inaccessible, especially in Tier-2 and Tier-3 regions. Our goal is to create a scalable, technology-driven system that enables continuous and affordable support for students, working professionals, and neurodivergent individuals.
By combining AI with structured care workflows, we aim to reduce the gap between demand and availability of mental health services and enable institutions to support large populations effectively. |
NA |
NA |
checked |
| Apr 30, 2026 @ 11:45 PM |
Sreerupa Chowdhury |
sreerupa@legalcare.app |
https://www.linkedin.com/in/sreerupachowdhury/ |
https://legalcare.io/ |
7338258612 |
CEO and 10+ years as an aSreerupa Chowdhury — Founder & CEO, legal-tech operator building an AI-native execution platform for legal and compliance workflows Vignesh Chandrasekaran — Growth & Product, ex PayPal and Amazon, scaling B2B products and growth systems Uddeshya Barnwal — AI Lead, IIT Kharagpur, building applied AI systems for real-world execution platformsdvocate ,Alumna of NUJS, Ex business head of BW Business world |
We came together through a mix of founder networks and operator circles, aligning on a shared vision of building an AI-native legal execution platform, and have been working closely together since Mid 2025. |
All |
Our strength is combining deep legal domain insight with product, growth, and AI execution — allowing us to move from idea to real-world deployment quickly and build a category-defining system, not just a tool. |
Legalcare |
https://legalcare.io/ |
Bengaluru |
LegalCare is an AI-native platform that executes legal and compliance workflows end-to-end with experts in the loop. |
Legal and compliance work is fragmented across multiple providers with no single owner of execution, leading to delays, missed obligations, and unpredictable outcomes—especially for startups and SMEs without in-house legal teams. |
LegalCare provides a unified execution layer where AI structures and runs legal workflows—from intake to completion—while experts validate critical decisions. Businesses can manage compliance, contracts, disputes, and IP on a single platform with full visibility and predictable execution |
LegalCare owns the execution layer by combining AI orchestration with expert validation, delivering outcomes rather than advice. Its structured workflow system, integrated expert network, and data from repeated legal execution create a compounding advantage that is difficult to replicate. |
Revenue |
Revenue |
Startups and small-to-medium businesses (SMEs) with recurring legal and compliance needs but no in-house legal teams. Secondary customers include enterprises for legal wellness and litigation management, and law firms for workflow enablement. |
$36B |
$7.2B |
$288M |
Subscription + usage-based model. Businesses pay a monthly subscription to access the platform and consume credits based on legal workflow execution. |
Global competitors : Harvey, Casetext, Spellbook, Ironclad, Clio, MyCase, Law firms, freelance lawyers,LegalZoom, Rocket Lawyer. Indian competitors : Zolvit, Indiafilings, Lawyered, Legalkart, Legistify. |
Product-led onboarding through high-frequency workflows (compliance, contracts) • Ecosystem partnerships (incubators, founder networks, CA/compliance partners) • Outbound (email, LinkedIn, calls) for enterprise • Legal wellness programs (LRS + awareness sessions) as entry for enterprises |
Start with startups and SMEs through high-frequency legal workflows, expand usage across multiple workflows to increase revenue per customer, and scale through partnerships and product-led growth. In parallel, enter enterprises via legal wellness programs and expand into litigation and compliance systems. |
To become the execution layer for legal globally — transforming legal from fragmented services into a structured, AI-native infrastructure where workflows are automated and outcomes are predictable. |
Incorporated (India) |
NA |
Yes |
We are building a new category at the intersection of legal and AI, and Avinya’s focus on early-stage, high-conviction startups makes it a strong fit. We’re looking to leverage the network, mentorship, and strategic support to accelerate product-market fit and scale our go-to-market efforts. |
Strengthen product-market fit across SME and enterprise segments • Scale customer acquisition and validate repeatable GTM • Refine pricing and monetization model • Build strategic partnerships for distribution • Prepare for institutional fundraising |
Yes, we are open to both hybrid and virtual participation. |
Yes |
Core engine |
LegalCare is built on a multi-layer, agentic architecture. The Intelligence Layer handles intent classification, legal reasoning, and document understanding. The Workflow Layer orchestrates execution through structured task flows, drafting, and routing. A Validation Layer brings experts into the loop at critical decision points. The Execution Layer completes actions such as filings, notices, and compliance tasks. This architecture ensures end-to-end legal execution with AI handling repeatable workflows and humans handling judgment. |
Our advantage comes from structured workflow data generated through real legal executions—intake patterns, decision paths, document variations, and outcome data. Over time, this creates a proprietary dataset of legal workflows and execution intelligence that improves automation accuracy and efficiency. |
LegalCare’s defensibility lies in combining AI orchestration with expert validation inside structured workflows. Unlike standalone AI tools or service providers, we own the full execution layer. The compounding data from repeated workflows, integrated expert network, and system-level orchestration creates a high barrier to replication. |
• Accuracy of intent classification
• Completion rate of workflows without rework
• Expert override frequency
• Turnaround time per workflow
• User satisfaction / repeat usage
• Error rate in document outputs |
We follow a privacy-by-design approach with end-to-end encryption, role-based access controls, and audit logs. Once an expert engages, interactions are protected under attorney–client privilege. Data is securely stored, access is restricted based on roles, and all actions are traceable. |
Yes, legal practice regulations and jurisdiction-specific compliance requirements may impact how services are delivered. We mitigate this by positioning LegalCare as an execution platform with expert validation, ensuring licensed professionals handle regulated decisions |
Primary challenges would include workflow variability across jurisdictions, maintaining expert availability, and ensuring consistent output quality. These are addressed through standardized workflows, scalable expert networks, and continuous model improvement. |
Yes. The AI function is led by an IIT Kharagpur alumnus with global experience across India, Japan, and the US in applied AI systems, supported by engineers focused on building scalable, production-grade AI workflows. |
We use a combination of publicly available legal data (case law, statutes), licensed data where required, and internally generated workflow data from platform usage. Proprietary data is built through execution-level interactions and structured outputs. |
Through feedback loops from expert validation, workflow outcome tracking, and user interactions. Each executed workflow improves the system’s understanding of legal patterns, enabling better automation, reduced errors, and faster execution over time. |
Both. Initial focus is India and the US, with a design architecture that supports expansion into global markets with similar legal workflow structures. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/DylGWWfUZj9x-LegalCare-o1-1-compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/DylGWWfUZj9x-LegalCare-o1-1-compressed.pdf |
https://drive.google.com/file/d/14GBeV3jA--f7F8eoBKkVmnUcWVRCkFCX/view?usp=sharing |
Yes. LegalCare is built to make legal access more structured, predictable, and accessible for businesses. Today, legal support is fragmented, reactive, and often out of reach for startups and SMEs. By turning legal into an execution-driven system, we reduce friction, improve compliance, and enable businesses to operate with greater trust and confidence. Over time, this also contributes to stronger governance and reduced legal disputes at a systemic level. |
Yes — the company is founded by a woman founder with 99% shareholding and a legal background, contributing to representation in both the legal and technology startup ecosystem. Additionally, one of the directors represents a minority shareholder group, further strengthening diversity in leadership. |
Amit Mishra |
checked |
| Apr 30, 2026 @ 11:16 PM |
Ajay Nair and Achyut Nayak |
nairajay@hotmail.com |
http://linkedin.com/in/ajay-nair-6a8369 |
https://swar-ai.com/ |
8762803277 |
1. Ajay Nair (Founder): An expert in GenAI and NLP with a career defined by building high-scale architectures. He previously co-founded Grapheme Labs, where he built NLP products that improved productivity by up to 1200%. His background includes serving as an Architect for UIDAI (Aadhaar), where he managed massive distributed systems, and Director of Engineering at Asianet News. He is an IIT Delhi alumnus. 2. Achyut Nayak (Founder): A veteran Product Engineer with over 25 years of experience in building and scaling complex systems. He is currently a Lead Product Engineer and Shareholder at Circles.Life, a global disruptor in the telecommunications space. His deep expertise in telco-grade engineering is pivotal for Swar AI’s mission to provide premium voice AI at telephony prices. 3. Souradeep Bharti (Business Development): An IIT Madras and IIM Lucknow alumnus with a background in management consulting at EY and KPMG. He also brings data-driven process transformation experience in high-volume voice operations from his time in Collections at Home Credit India and at a US-based Growth-as-a-Service (BPO) firm. His experience enables him to architect the commercial playbooks required to scale Swar AI across global enterprise environments. |
We met one month ago through the IIT Alumni Center Bangalore (IITACB) network and the Antler Network. While the current team is new, Ajay and Achyut have a prior professional relationship, having worked together previously as founding and distinguished engineers at Fratics. Over the past month, we have aligned our technical roadmap with a structured Go-To-Market strategy and are currently preparing for initial pilot deployments. |
All |
Our primary strength is the intersection of high-scale engineering experience, specific domain expertise in voice operations, and strong institutional network access.
1) Technical Foundation: With Ajay’s background in government-scale NLP systems and Achyut’s 25+ years in telco-grade product engineering, we have the experience required to build a reliable, high-volume voice AI architecture.
2) Operational Insight: Our commercial strategy is informed by Souradeep’s experience managing tele-collections and BPO operations, ensuring the product addresses the specific efficiency and compliance needs of enterprise clients.
3) Institutional Network: As alumni of IIT Delhi, IIT Madras and IIM Lucknow, we are in a position to leverage deep institutional networks to identify high-intent professionals and accelerate GTM operations. |
Swar AI |
https://swar-ai.com/ |
Bengaluru, Karnataka, India |
Swar AI develops autonomous, multilingual voice agents to automate enterprise telecalling for sales and collections at a lower cost than traditional BPOs. |
Enterprise telecalling operations face high agent attrition, inconsistent call quality, and high operational expenses. Existing automated solutions often suffer from high latency or unnatural voice quality, making them ineffective for complex conversations like debt collections or sales. |
We build AI-powered (both open and proprietary models) vertically integrated voice agents capable of conducting natural, low-latency conversations in multiple languages. The platform is designed on a telephony-grade architecture to ensure reliability and scalability for enterprise-level interaction volumes. |
1) Telephony-Grade Pedigree: Our founding team includes an architect from UIDAI (Aadhaar) and a lead product engineer from Circles.Life with over 25 years of experience in scaling global telecommunications systems. 2) Operational Domain Expertise: The product strategy is informed by firsthand experience managing tele-collections at Home Credit India, growth-as-a-service at a US-based BPO and Futwork. 3) Full-Stack Optimization: By being vertically integrated, we can optimize for both cost and voice fidelity in ways that third-party "wrapper" solutions cannot, creating a sustainable cost-to-value advantage. |
Users |
Pilots |
a) BFSI Sector: Focused on debt collection agencies and lending institutions requiring high-volume tele-collections. b) E-commerce & EdTech: Companies managing outbound sales, lead qualification, and abandoned cart recovery. c) Outsourced BPOs: Service providers looking to automate Tier-1 voice interactions to reduce overhead and agent attrition. |
Estimated at $17.12 billion by 2026 |
Projected to exceed $800 million in 2026, growing at a CAGR of ~25.6% |
Our initial target is the Indian enterprise voice automation segment, a subset of the domestic GenAI market focusing on high-stakes sales and collections interactions where we have a direct operational advantage. |
Target capture of the $100M+ niche within high-intent BFSI and sales-driven tele-operations. |
a) Direct (India): Futwork (hybrid human-tech), SquadStack (sales/collections focused), Haptik (BFSI/support), and Vozzo AI. b) Infrastructure/APIs: Vapi and Retell AI, which provide the underlying voice-agent framework for developers. c) Global Enterprise: PolyAI and Cognigy, specializing in high-fidelity, branded voice experiences. |
a) Institutional Networking: Leveraging the IIT and IIM Alumni groups for warm introductions to enterprise decision-makers. b) Direct Outbound: Targeted outreach to COOs and Heads of Departments at mid-to-large-scale firms. c) Pilot Programs: Offering low-friction, 15-day proof-of-concept (PoC) deployments to demonstrate cost-per-minute savings over traditional BPOs. |
Phase 1 (Validation): Secure three anchor pilot clients in the tele-collections and EdTech sectors to refine the vertical-specific NLP models by leveraging our institutional networks. Phase 2 (Expansion): Scale through direct sales and partnerships with existing BPO providers looking to modernize their tech stack. Phase 3 (Dominance): Establish Swar AI as the vertically-integrated AI alternative to hybrid models, emphasizing telco-grade architecture to provide lower latency and sub-cent per minute pricing. |
To become the global infrastructure layer for autonomous voice interactions, replacing human-intensive telecalling with a vertically integrated AI stack that delivers telephony-grade reliability and enterprise-level cost efficiency. |
Incorporated in 5th July 2024 as Swar AI Technologies Pvt. Ltd. |
NA |
Yes |
We are applying to Avinya to bridge the gap between our high-scale technical architecture and large-scale enterprise adoption. While our team has a proven track record in scaling global systems at firms like UIDAI and Circles.Life, we recognize that navigating the BFSI and BPO procurement landscape requires the institutional validation and strategic mentorship that Avinya provides. We aim to leverage the program’s network to refine our value proposition and accelerate our entry into the enterprise market. |
Our primary objectives during the programme are: a) Pilot Validation: Secure and successfully execute our first three anchor pilots in the tele-collections and EdTech sectors to validate our vertical-specific NLP models. b) Architecture Refinement: Stress-test our vertically integrated telephony stack for extreme scalability under enterprise-level call volumes. c) Commercial Scalability: Finalize our enterprise GTM playbooks and prepare the operational framework required for a formal seed funding round. |
Yes |
Yes |
Core engine |
Swar AI uses a vertically integrated telephony-AI stack that employs a hybrid model approach, utilizing both proprietary and open-source AI models. The architecture integrates Speech-to-Text (STT), LLM reasoning, and Text-to-Speech (TTS) directly into the telephony layer to minimize latency and maintain high-fidelity voice quality. |
The platform is informed by domain-specific operational data from high-volume tele-collections and BPO workflows. This background provides specific linguistic patterns, negotiation logic, and compliance requirements unique to the debt recovery and sales verticals. |
Defensibility is based on the combination of telco-grade engineering and vertical-specific logic. The founders have experience architecting government-scale systems (UIDAI) and global digital telco platforms (Circles.Life), allowing for a more stable and cost-efficient telephony integration than standard API-based wrappers. |
- Technical: Latency (ms), Word Error Rate (WER), and Mean Opinion Score (MOS).
- Functional: Goal Completion Rate (e.g., payment commitments) and Sentiment Analysis accuracy.
- System: Concurrent call capacity and system uptime. |
The system utilizes end-to-end encryption for voice data and PII masking in conversation logs. The team applies security standards developed during their work on national-scale systems (UIDAI) to ensure compliance with regional data sovereignty and financial regulations. |
Primary risks include changing TRAI regulations regarding automated outbound calling and evolving data privacy frameworks in India and international markets. The architecture includes a modular compliance engine to adapt to these shifts. |
At 10x scale, the primary challenges are concurrency management in the telephony-to-inference bridge and real-time state management across distributed nodes. The system is built on microservices to mitigate these bottlenecks. |
Yes, the technical leadership consists of:
- Ajay Nair: Specialist in GenAI and NLP with experience in high-scale architectures at UIDAI.
- Achyut Nayak: Over 25 years of product engineering experience, specifically in scaling global telecommunications systems. |
We have option to use both open-source and pre-trained proprietary models. Open source models are trained using public datasets. We don't own any datasets. |
Performance is improved through Reinforcement Learning from Human Feedback (RLHF) and active learning loops. We use anonymized call logs to identify drop-off points and perform continuous fine-tuning on vertical-specific data. |
Both. While the initial focus is on the Indian BFSI and EdTech sectors, the telephony-grade architecture is designed for global scalability and portability. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/fgf2tL60YqfU-Swar-AI-Pitch-Deck-1.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/fgf2tL60YqfU-Swar-AI-Pitch-Deck-1.pptx |
https://drive.google.com/drive/folders/1JlY6m3puuAe3fb5F8mCeCCo7raXb7g0_?usp=sharing |
Yes. Swar AI is mission-driven to democratize high-fidelity voice technology for enterprises that are currently limited by the high costs and operational inefficiencies of traditional BPOs. By providing "premium quality at telephony prices," we enable smaller financial institutions and growth-stage companies to access sophisticated, autonomous customer engagement tools that were previously only available to the largest corporations. Our impact focus is on operational equity—reducing the "tech-debt" for companies in emerging markets by providing world-class AI at a disruptive price point. |
First-generation founder team |
Vivek Jha |
checked |
| Apr 30, 2026 @ 11:07 PM |
Kushal CI, Vaibhav Krishna |
kushalingole@IITBOMBAY.ORG |
https://www.linkedin.com/in/kushalcingole |
https://oabco.in/ |
09769140510 |
Kushal C. Ingole – IIT Bombay & IIM Ranchi alumnus, with 15+ years of experience in business management and strategy across multiple industries, functions, and domains, including family offices, social sector, corporates, startups, government roles, and MNCs. Dr. Vaibhav Krishna – IIT Bombay alumnus, with Masters & Ph.D. from Yale University, and 10+ years of experience in AI/ML, building and deploying advanced technology solutions across research, enterprise applications, and startups |
We have met at IIT Bombay and have worked together for 4+ years. |
two |
Together, Kushal Ingole and Vaibhav Krishna form a balanced founding team—where complementary skills, shared vision, and mutual trust drive exceptional results. |
Oabco Innovations |
Na |
Bangalore |
AGentic AI for Manufacturing |
According to recent operator surveys, autonomous agents will execute roughly 60% of transactional procurement activity within 24 months. That single statistic reframes every job description, every vendor contract, and every reporting line on a procurement org chart. |
According to recent operator surveys, autonomous agents will execute roughly 60% of transactional procurement activity within 24 months. That single statistic reframes every job description, every vendor contract, and every reporting line on a procurement org chart. |
Agentic AI + Humans |
Users |
Pilots |
Manufacturers |
$60 Bn |
$15Bn |
$3Bn |
Subscription |
Procure.ai |
Partnerships, B2B Marketing |
Pilots, Industry Connects |
To help 100 manufacturers use AI for procurement in the next 3 years. |
Yes |
NA |
one |
To generate IP & long-term value and learn from Industry leaders |
Problem solution fit |
Yes |
one |
one |
Core ML Models
Spend classification
Auto-tag line items (GL codes, categories)
Models: XGBoost / BERT fine-tuned on invoices
Price benchmarking
Detect overpricing vs historical/vendor benchmarks
Regression + clustering
Supplier risk scoring
Delivery delays, defect rates, compliance risks
Gradient boosting + anomaly detection
Demand-linked procurement
Input from forecasting → triggers procurement
Time-series models (TFT / LightGBM)
APIs (Critical Integrations)
ERP Systems
Finance Systems
Vendor Ecosystem
Internal APIs |
The advantage is a compounding dataset built from real procurement workflows—RFQs, vendor quotes, POs, invoices, and outcomes—normalized into a supplier–price–performance graph. Every transaction our agents execute improves vendor benchmarking, price intelligence, and decision accuracy, creating a flywheel that’s hard to replicate. |
Our defensibility comes from a closed-loop, agentic system that learns from real procurement decisions—combining proprietary transaction data, multi-agent workflows, and deep ERP integrations to continuously improve outcomes. Competitors can replicate models, but not the data, workflows, and execution layer we’ve embedded |
We evaluate performance across four layers—prediction accuracy, decision quality, execution reliability, and business outcomes—with offline benchmarks, online A/B tests, and strict guardrails for automation. Models only graduate to autonomy after consistently meeting ROI and risk thresholds |
We implement enterprise-grade security with strict tenant isolation, encryption in transit and at rest, least-privilege access, and full auditability. Sensitive data is never used to train shared models; we use private or zero-retention LLM setups, enforce policy guardrails, and align with global standards |
Customer data is never used to train shared models; we operate with private or zero-retention LLM configurations, enforce policy guardrails, and ensure all AI outputs remain within defined compliance boundaries. Our platform is aligned with global security standards, with support for data residency and enterprise-grade governance requirements |
At 10× scale, the pressure points are data pipelines, agent orchestration, LLM cost/latency, and decision reliability. We mitigate this with streaming architectures, hierarchical agent orchestration, model routing and caching, and strict policy/approval gates to keep automation safe as volume grows |
Yes, Vaibhav brings hands-on experience in building production ML systems, LLM-based agents, and data platforms. |
We are using customer-owned procurement data (ERP, POs, invoices, vendor data) accessed via secure integrations, combined with publicly available reference datasets and synthetic data for model bootstrapping. All customer data remains owned by the customer and is processed under strict data agreements; over time, we build proprietary, anonymized intelligence layers (pricing benchmarks, supplier performance) derived from transaction outcomes. |
We improve performance through a closed-loop system that learns from every procurement decision—combining continuous data ingestion, human feedback, online evaluation, and automated retraining with strict guardrails. Models only graduate to higher autonomy after consistently improving business outcomes |
Both |
|
|
It's a mission-driven approach as it solves one of the key problems for product availability. The finished product is dependent on its raw materials; if the sourcing is not sorted, it would lead to production delays, thus impacting the lives of people due to the unavailability of products. Some key sectors like fertilizers, agrochemicals, pharmaceuticals, oil, and gas are very important for the growth of the country. |
Yes |
Cdr. Pradeep Prasad |
checked |
| Apr 30, 2026 @ 11:04 PM |
Gajendra Chidambar Deshpande, Pradosh Pramod Hegde |
contact@thetadynamics.in |
https://www.linkedin.com/in/gajendradeshpande/ |
https://thetadynamics.in |
9008450948 |
Gajendra holds an M.Tech in Computer Science and Engineering and has 18 years of experience. As Co-founder and Managing Director, he primarily oversees the technological aspects of the organization. Pradosh has 20 years of experience. As Co-founder, he primarily manages the business operations of the organization. |
Gajendra and Pradosh first met 25 years ago during their Diploma studies and have maintained a strong professional relationship since then. For the past 3 years, they have been working together closely, combining Gajendra’s technology leadership with Pradosh’s business operations expertise. |
2 |
Our biggest strength as a team is our long-standing trust, complementary skills, and shared vision. Gajendra brings strong technology leadership, product development experience, and innovation capabilities, while Pradosh brings business operations, execution, and coordination expertise. Since we have known each other for 25 years and worked together closely for the past 3 years, we have strong mutual understanding, clear role clarity, and the ability to execute ideas effectively. |
Theta Dynamics Private Limited |
https://thetadynamics.in |
Belagavi |
Theta Dynamics develops AI-powered Wi-Fi sensing solutions for privacy-preserving monitoring, smart surveillance, and intelligent infrastructure applications. |
We are solving the problem of privacy-invasive and expensive surveillance systems by developing an AI-powered Wi-Fi sensing solution that can detect human presence, movement, and activity patterns without using cameras, while enabling affordable smart monitoring for homes, institutions, industries, and public infrastructure. |
Our solution is an AI-powered Wi-Fi sensing platform that uses existing Wi-Fi signals to detect human presence, movement, and activity patterns without using cameras. It analyzes signal variations through AI/ML models and provides real-time alerts, dashboards, and monitoring insights. The solution is designed for privacy-preserving smart surveillance, elderly care, institutional monitoring, industrial safety, and infrastructure security, offering a cost-effective alternative to camera-based systems. |
Our solution is unique because it enables camera-free, privacy-preserving sensing using Wi-Fi signals, making it suitable for places where CCTV is intrusive or ineffective. It is defensible through our focus on AI models trained for Indian indoor environments, low-cost hardware integration, real-time activity detection, and potential IP around Wi-Fi signal processing, anomaly detection, and deployment-specific learning. |
MVP |
Pilots |
Our target customers are institutions, industries, smart infrastructure operators, elderly-care facilities, security agencies, and public/private organizations that need privacy-preserving monitoring and camera-free surveillance. |
INR 20,000 CR |
INR 3,000 CR |
INR 50 CR |
Theta Dynamics follows a hardware-enabled SaaS revenue model. We generate one-time revenue through device sales, installation, and pilot deployments, and recurring revenue through monthly or annual subscriptions for AI-based dashboards, alerts, analytics, reports, AMC, and enterprise integrations. |
Our main competitors include global Wi-Fi sensing companies such as Origin Wireless / Origin AI, nami.ai, Cognitive Systems, and Synaptics, along with indirect competitors such as CCTV providers, smart alarm systems, PIR motion sensors, access control systems, and home security companies. Theta Dynamics differentiates itself by focusing on India-specific, affordable, privacy-preserving Wi-Fi sensing solutions for institutions, industries, elderly care, smart buildings, and infrastructure monitoring, where camera-based surveillance may be intrusive, costly, or unsuitable. |
Theta Dynamics will acquire customers through direct B2B sales, pilot deployments, institutional partnerships, channel partners, system integrators, and security service providers. We will first target institutions, industries, elderly-care facilities, smart buildings, and warehouses through MVP demonstrations and paid pilots. Successful pilots will be converted into long-term hardware + SaaS subscriptions, supported by referrals, case studies, and partner-led expansion. |
Our go-to-market strategy is to begin with B2B pilot deployments in institutions, industries, elderly-care facilities, smart buildings, and warehouses. We will validate the MVP through focused use cases, convert successful pilots into annual hardware + SaaS subscriptions, and scale through direct sales, channel partners, system integrators, IoT solution providers, security agencies, referrals, and case-study-driven marketing. |
Theta Dynamics’ long-term vision is to become a leading Indian deep-tech company in AI-powered Wi-Fi sensing and privacy-preserving smart surveillance. We aim to build affordable, indigenous, camera-free sensing solutions for homes, institutions, industries, elderly care, smart buildings, and critical infrastructure, while creating strong IP and scalable products for India and global markets. |
Incorporated |
NA |
Yes |
We are applying to Avinya to receive mentorship, market validation, and ecosystem support for scaling our AI-powered Wi-Fi sensing MVP into a deployable product. The program will help us refine our business model, validate pilot use cases, connect with mentors, customers, and partners, and accelerate our journey from MVP to commercialization. |
During the programme, we aim to refine our Wi-Fi sensing MVP, validate key use cases with mentors and potential customers, identify pilot opportunities, improve our business and revenue model, and build a clear go-to-market strategy. We also want to strengthen our product positioning, understand investor and customer expectations, and prepare Theta Dynamics for pilot deployments and early commercialization. |
Yes |
Yes |
Core engine |
Our AI architecture consists of a Wi-Fi signal sensing pipeline where raw Wi-Fi signal variations are collected, cleaned, processed, and converted into features for AI/ML models. The AI core analyzes these patterns to detect human presence, movement, activity, occupancy, and anomalies. The inference layer generates real-time alerts and insights, which are delivered through dashboards, notifications, reports, and API integrations. |
Our proprietary data advantage will come from collecting and building India-specific Wi-Fi sensing datasets from real deployment environments such as institutions, industries, smart buildings, warehouses, and elderly-care facilities. These datasets will capture variations in Indian building layouts, wall materials, room sizes, furniture placement, crowd density, device types, and real-world movement patterns.
Over time, this deployment data will help us improve model accuracy, reduce false alerts, and create environment-specific AI models that are difficult for generic global solutions to replicate. This data advantage will also support future IP creation in Wi-Fi signal processing, activity detection, anomaly detection, and privacy-preserving smart surveillance. |
Our AI component is defensible through our focus on India-specific Wi-Fi sensing data, environment-adaptive AI models, and deployment-specific calibration. Wi-Fi signal behavior changes based on building layout, wall material, room size, furniture placement, crowd density, and device configuration. By collecting real deployment data and training models for these variations, we can reduce false alerts and improve accuracy in local conditions. Over time, this creates a proprietary dataset and model advantage that is difficult for generic global competitors to replicate. |
We evaluate model performance and reliability using both AI metrics and real-world deployment metrics.
Key metrics include accuracy, precision, recall, F1-score, false positive rate, false negative rate, detection latency, and uptime. For Wi-Fi sensing, we also measure reliability across different environments such as room size, wall material, furniture placement, number of people, device position, and movement type.
We will validate the model through controlled lab testing, pilot deployments, continuous monitoring, and comparison of predicted events with ground-truth observations. Our goal is to achieve high detection accuracy, low false alarms, fast response time, and stable performance across different Indian indoor environments. |
We handle data privacy, compliance, and security through a privacy-by-design approach. Our Wi-Fi sensing solution does not use cameras or record images, videos, or audio, making it less intrusive than traditional surveillance systems.
We collect only the minimum required Wi-Fi signal data needed for sensing and analytics. Data is encrypted during transmission and storage, with role-based access control for dashboards and reports. We also plan to use secure APIs, audit logs, user authentication, and customer-controlled data retention policies.
For compliance, we will align with applicable Indian data protection requirements, including the Digital Personal Data Protection Act, 2023, and follow industry best practices for secure deployment, consent, access control, and responsible data handling. |
Yes, there are manageable regulatory risks related to data privacy, wireless device compliance, and deployment permissions.
We will address these through privacy-by-design, camera-free and audio-free sensing, minimum data collection, encryption, role-based access, consent-based deployment, and compliance with applicable Indian regulations such as the Digital Personal Data Protection Act, 2023 and wireless device approval requirements. |
At 10x scale, our AI system may face challenges in maintaining consistent accuracy across different building layouts, wall materials, Wi-Fi devices, crowd densities, and movement patterns. The data pipeline, real-time inference engine, alert system, and dashboard infrastructure may also need optimization to handle higher device volume and traffic. We will address this through scalable cloud architecture, edge processing, automated calibration, continuous model monitoring, retraining pipelines, and stronger MLOps practices. |
Yes.
Theta Dynamics has in-house AI/ML expertise led by Gajendra Deshpande, who has 18 years of experience in computer science, AI/ML, cybersecurity, research, and technology training. The team has experience in machine learning model development, signal/data analysis, Python-based AI implementation, cybersecurity applications, and building practical technology solutions. For the Wi-Fi sensing product, the team focuses on AI model design, data preprocessing, model evaluation, deployment, and continuous improvement. |
Currently, we are using internally collected experimental Wi-Fi signal data for MVP development and validation. This data is generated through our own controlled testing, so it is owned by Theta Dynamics.
As we move toward pilots, we plan to collect deployment-specific datasets from institutions, industries, smart buildings, and elderly-care environments with proper consent and customer permissions. We may also refer to publicly available research datasets for benchmarking, wherever licensing allows, but our primary dataset will be owned first-party deployment data. |
We will continuously improve model performance by collecting consent-based deployment data from pilots, comparing predictions with ground truth, tracking metrics such as accuracy, precision, recall, F1-score, false positive rate, false negative rate, and latency, and retraining models periodically. We will also use environment-specific calibration, edge-case analysis, user feedback, and MLOps practices to improve reliability across different Indian indoor environments. |
We are initially building for the Indian market, focusing on institutions, industries, smart buildings, elderly-care facilities, warehouses, and infrastructure monitoring use cases. India provides diverse real-world environments to validate and improve our Wi-Fi sensing solution. In the long term, we aim to scale the product to global markets where privacy-preserving, camera-free monitoring is needed. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/S2BHeH5BVUV8-Pitch-Deck-2-Business-Model.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/S2BHeH5BVUV8-Pitch-Deck-2-Business-Model.pdf |
http://NA |
Yes, Theta Dynamics is a mission-driven and impact-focused startup.
We are building AI-powered Wi-Fi sensing technology to make monitoring privacy-preserving, affordable, and accessible. Our mission is to reduce dependence on camera-based surveillance by offering a camera-free and audio-free sensing solution for institutions, industries, elderly care, smart buildings, and public infrastructure.
Our impact focus is on improving safety, security, elderly-care monitoring, infrastructure protection, and responsible AI adoption, especially in Indian environments where cost, privacy, and scalability are important. |
NA |
NA |
checked |
| Apr 30, 2026 @ 10:55 PM |
Dr Deepak KT , Mr Uroojakhtarkhan , Dr Prakash Pawar |
uruinam@gmail.com |
https://www.linkedin.com/in/uroojakhtarkhan/ |
http://NA |
9632342375 |
Mr. Deepak K. T. is an Assistant Professor at the Indian Institute of Information Technology Dharwad, specializing in AI, speech processing, and embedded systems, with a PhD from Indian Institute of Technology Guwahati and prior industry experience at NXP Semiconductors and STMicroelectronics. Mr. Dr. Prakash Pawar is an Assistant Professor at the Indian Institute of Information Technology Dharwad, with an academic background from National Institute of Technology Karnataka, involved in teaching, research, and academic development activities. Mr. Urooj Akhtar Khan Inamdar is a DeepTech founder and researcher at Humanoid.AI, currently pursuing her engineering at S G Balekundri Institute of Technology, with a strong focus on AI, intelligent systems, and innovation-driven projects. |
Mr Deepak and Dr Prakash have been working on Humanoid from past 3 Years . Mr Uroojakhtar Joined them on 2024 Nov .
Since then we have been working on this Idea togeather.
Mr Deepak and Dr Prakash are colleagues in IIIT Dharward , Mr Urooj worked as project associate with them |
All |
The biggest strength of this team is its powerful combination of academic expertise and cutting-edge AI innovation.
Deepak K T brings deep research experience, industry exposure, and funded project leadership in AI and speech technologies.
Dr. Prakash Pawar contributes strong academic foundation, teaching experience, and research support.
Urooj Akhtar Khan Inamdar adds fresh innovation, startup mindset, and hands-on development in AI systems. |
HUmanoid.AI |
NA |
Dharward |
A voice-driven humanoid AI Robotic system that autonomously assists elderly individuals with navigation, care, and real-time interaction. |
As the elderly population grows, there aren’t enough caregivers to provide continuous, personalized attention resulting in safety issues, loneliness, and delayed medical response. We solve this with intelligent humanoids that offer real-time, voice-driven assistance and monitoring.” |
Our solution is a smart humanoid robot built specifically for geriatric care, combining AI-powered assistance, autonomous mobility, and human-like interaction to support elderly individuals in homes, hospitals, and care centers. It is designed to improve safety, independence, and emotional well-being of senior citizens.
Key Features:
Voice-Driven Interaction: Understands multilingual voice commands and responds naturally.
Autonomous Navigation: Moves safely indoors with real-time obstacle avoidance.
Personalized Assistance: Provides medication reminders, daily task support, and emergency alerts.
User Identification & Tracking: Recognizes and follows authorized users when needed.
Companionship Support: Engages in conversations to reduce loneliness and improve mental wellness.
AI Intelligence: Powered by speech recognition, text-to-speech, and Large Language Models for smart decision-making. |
Our solution is unique because it combines AI-powered humanoid robotics, autonomous navigation, and elderly care assistance in one platform. Unlike regular robots or chatbots, it is specially designed for geriatric care with voice interaction, safety monitoring, reminders, and companionship. Defensibility: Specialized for Elderly Care rather than general robotics. Voice + Mobility Integration for real-world assistance. Multilingual & Localized AI for Indian users. Hardware + Software Complexity creates high entry barriers. Data & Partnerships Advantage with hospitals and care centers. This makes it a scalable and hard-to-replicate solution for the growing senior care market. |
MVP |
Pilots |
Our target customers are elderly individuals, families, hospitals, and senior care institutions looking for smarter and more efficient geriatric support solutions. Primary Customers: Senior Citizens needing daily assistance, safety, and companionship. Families of Elderly Individuals seeking reliable remote care support. Hospitals & Rehabilitation Centers for patient assistance and monitoring. Old Age Homes & Assisted Living Centers to improve care efficiency. Secondary Customers: Healthcare Providers integrating AI-driven elderly care solutions. Government & NGOs focused on senior citizen welfare programs. We primarily focus on the growing geriatric care market where demand for affordable, continuous, and intelligent care is rapidly increasing. |
₹2.8 lakh crore |
₹1.26 lakh crore |
₹30 crore |
1. Hardware Sales (One-time) I.e Sale of humanoid robot 2 . B2B / Institutional Plans Hospitals, rehab centers, elderly homes |
1. Humanoid & Social Robots SoftBank Robotics (Pepper robot) Hanson Robotics PAL Robotics 2. Elderly Care / Assistive Robots Intuition Robotics (ElliQ companion robot) Blue Frog Robotics PARO Robots 3. Indirect Competitors (Substitutes) Amazon Google Human caregivers / nursing services |
We acquire customers through a mix of B2B partnerships and direct outreach collaborating with hospitals, elderly care centers, and clinics, while also reaching families through digital marketing, demos, and referrals. |
We follow a B2B2C go-to-market strategy starting with pilot deployments in hospitals and elderly care centers to build trust, then expanding to families through partnerships, demos, and referrals |
Our long-term vision is to build the leading AI-powered humanoid care ecosystem for elderly assistance, transforming how senior citizens live with dignity, safety, and independence. We aim to scale from homes and hospitals in India to global healthcare markets, making intelligent robotic care accessible and affordable.
Future Vision:
Become the trusted companion robot for geriatric care worldwide.
Create connected healthcare ecosystems with hospitals and caregivers.
Advance robotics with emotional intelligence and personalized care.
Reduce caregiver burden while improving quality of life for seniors.
Position India as a global leader in healthcare robotics innovation. |
NA |
9.06 Lakhs |
Yes |
I am applying to Avinya because it represents a rare opportunity to work alongside ambitious founders, innovators, and mentors who are building solutions with real-world impact. My vision is to develop an AI-powered humanoid robotics solution focused on geriatric care, combining voice interaction, autonomous navigation, and intelligent assistance for the elderly. Avinya’s ecosystem of mentorship, strategic guidance, and entrepreneurial exposure can help me transform this vision into a scalable and impactful venture. |
During the programme, I want to achieve three key outcomes: first, validate and refine my product-market fit through expert feedback and customer discovery; second, strengthen my go-to-market, fundraising, and startup execution capabilities; and third, build a strong network of mentors, investors, and fellow founders who can accelerate the growth of the venture. I also aim to sharpen my leadership skills and convert my current prototype into a market-ready solution with long-term potential. |
Yes we prefer Hybrid |
Yes |
Core engine |
Our AI architecture is built as a multimodal intelligent robotics stack that combines perception, reasoning, decision-making, and action execution for geriatric care. It integrates voice, vision, and sensor data to enable natural interaction and autonomous assistance.
Core Architecture Layers:
Perception Layer: Uses Automatic Speech Recognition (ASR), cameras, microphones, and proximity sensors to capture voice commands, identify users, and understand surroundings.
Language Intelligence Layer: Large Language Models (LLMs) process speech/text inputs, understand intent, manage conversations, and generate context-aware responses in multiple languages.
Decision Engine: Combines user intent, environmental data, and safety rules to decide tasks such as navigation, reminders, emergency alerts, or follow-up questions.
Navigation & Control Layer: Uses SLAM, obstacle avoidance, and path-planning algorithms for safe autonomous movement in indoor environments.
Personalization Layer: Learns routines, preferences, medication schedules, and interaction history for customized elderly care support.
Cloud + Edge Deployment: Real-time critical tasks run on-device for low latency, while advanced model updates and analytics run on the cloud.
Key Strength:
The architecture uniquely combines AI conversation + robotics mobility + healthcare assistance into one unified system, enabling safe, intelligent, and human-like geriatric care. |
Our proprietary data advantage comes from building a specialized geriatric robotics dataset that is difficult to replicate. As users interact with the robot, we generate valuable real-world data across voice, behavior, mobility, and care workflows.
Key Data Advantages:
Elderly Interaction Data: Voice commands, conversation patterns, reminders, and care requests from senior users.
Multilingual Regional Speech Data: Hindi, English, and Indian language accents for better ASR performance.
Mobility & Indoor Navigation Data: Real-world home/hospital movement patterns, obstacle scenarios, and navigation learning.
Personalized Care Data: Daily routines, medication schedules, preferences, and engagement habits.
Human-Robot Feedback Loops: Continuous learning from corrections and user responses.
Why It Matters:
This creates a compounding moat where our AI models become smarter, more personalized, and more accurate over time making the platform harder for competitors to match. |
Our AI component is defensible because it is built on a domain-specific geriatric intelligence layer rather than a generic chatbot or standard robot software. We combine proprietary elderly-care data, multimodal learning, and robotics execution into one system.
Key Defensibility Factors:
Proprietary Dataset: Real-world elderly interaction, speech, and care behavior data that improves model performance over time.
Multimodal Intelligence: Combines voice, vision, sensors, and navigation instead of text-only AI.
Localized AI Models: Optimized for Indian languages, accents, and healthcare use cases.
Personalization Engine: Learns routines, preferences, and health needs for each user.
Hardware + AI Integration: Tight coupling of software intelligence with autonomous robotics creates high replication barriers.
Competitive Edge:
While others may copy individual features, replicating the full AI + robotics + geriatric care ecosystem is significantly harder and takes time, data, and trust. |
We evaluate model performance and reliability using a combination of AI accuracy metrics, robotics safety metrics, and user experience outcomes to ensure dependable geriatric care assistance.
AI Model Metrics:
ASR Accuracy: Word Error Rate (WER) for speech recognition across accents and noisy environments.
Intent Detection Accuracy: Correct understanding of user commands and requests.
Response Quality: Relevance, coherence, and latency of AI-generated responses.
Multilingual Performance: Accuracy across Hindi, English, and regional language inputs.
Robotics & Navigation Metrics:
Task Success Rate: % of commands completed correctly.
Obstacle Avoidance Accuracy: Safe navigation without collisions.
Localization Accuracy: Precision of mapping and indoor positioning.
Response Time: Time taken from command to action.
Reliability & User Metrics:
System Uptime: Continuous operational availability.
Error Recovery Rate: Ability to handle failures or unclear commands safely.
User Satisfaction Score: Feedback from elderly users, families, and caregivers.
Engagement & Retention: Frequency of continued daily usage.
Goal:
Our focus is not just AI accuracy, but safe, trustworthy, and real-world dependable performance in elderly care environments. |
We handle data privacy, compliance, and security through a privacy-first architecture designed for sensitive healthcare and elderly user environments. User trust and safety are core to our system design.
Data Privacy:
Consent-Based Data Collection: Explicit user/family consent before collecting voice, health, or usage data.
Data Minimization: Collect only essential data required for functionality.
User Control: Users can access, manage, or delete their data.
Anonymization: Personal identifiers removed for model training and analytics.
Security Measures:
End-to-End Encryption: Data encrypted in transit and at rest.
Secure Access Controls: Role-based access for caregivers, admins, and healthcare partners.
Edge Processing: Sensitive real-time data processed locally on-device where possible.
Regular Audits: Continuous monitoring, vulnerability testing, and system updates. |
For security, all data is protected through end-to-end encryption in transit and at rest, with role-based access controls for caregivers, admins, and healthcare partners. Wherever possible, sensitive real-time information is processed locally through edge computing to reduce exposure. We also conduct regular monitoring, vulnerability testing, and system updates to maintain a secure and reliable platform. |
If we scale 10x, the main challenges will be latency, infrastructure load, personalization complexity, and operational reliability across thousands of users and devices.
Potential Break Points:
Inference Latency: More simultaneous users can slow real-time voice responses and decision-making.
Cloud Compute Costs: Large-scale LLM and speech processing can become expensive.
Personalization at Scale: Managing unique routines, preferences, and care data for many users becomes complex.
Device Reliability: Maintaining uptime, updates, and performance across many deployed robots.
Data Security Risk Surface: More users and endpoints increase cybersecurity requirements.
Support & Maintenance: Hardware servicing and field operations grow significantly.
How We Solve It:
Hybrid edge + cloud AI to reduce latency and cost.
Smaller optimized models for on-device tasks.
Scalable microservices and automated fleet management.
Strong cybersecurity architecture and monitoring.
Standardized hardware support and remote diagnostics. |
Yes, we have an in-house AI/ML team with strong expertise across artificial intelligence, robotics, speech technology, and product development. The team combines academic research depth with hands-on engineering execution.
Core Team Expertise:
Dr. Deepak K T – Expertise in AI, humanoid robotics, autonomous systems, R&D leadership, and deep-tech innovation, with experience in funded research projects and startup building.
Dr. Prakash Pawar – Expertise in speech technologies, embedded systems, IoT, machine learning, and intelligent systems, with multiple sponsored research projects.
Urooj Akhtar Khan – Focused on AI product strategy, startup execution, applied machine learning, and market deployment.
MR Swapnil – Specialized support in advanced AI model development, experimentation, and research execution.
Interns & Engineering Contributors – Working on data annotation, prototyping, testing, robotics integration, and software development.
Combined Strength:
Our in-house team covers the full stack: LLMs, speech AI, computer vision, robotics navigation, embedded systems, and commercialization, giving us the capability to build and scale proprietary AI products quickly. |
We are currently using primarily owned and internally generated datasets, giving us stronger control, customization, and long-term defensibility. These datasets are created specifically for our humanoid geriatric care use case.
Current Datasets (Owned):
Voice Interaction Dataset: In-house recordings of commands, conversations, and elderly assistance scenarios in Hindi, English, and regional accents.
Navigation & Environment Dataset: Proprietary indoor mapping, obstacle, and movement data collected from home, hospital, and care-center simulations.
User Behavior Dataset: Interaction logs, routines, reminder usage patterns, and task preferences for personalization.
Multimodal Dataset: Combined speech, text, and sensor data for user identification, command understanding, and contextual decision-making.
Testing & Safety Dataset: Edge cases for fall-risk alerts, emergency triggers, noisy environments, and low-light conditions.
Ownership & Licensing:
Core datasets are fully owned and generated in-house.
Any third-party pretraining or benchmark datasets used during research are only supplemental and properly licensed.
Strategic Advantage:
Owning our data allows faster iteration, better privacy control, and models optimized specifically for elderly care robotics. |
We plan to continuously improve model performance through a closed feedback loop of data, deployment, and retraining. As more users interact with the system, the AI becomes smarter, safer, and more personalized over time.
Improvement Strategy:
Continuous Data Collection: Use consented real-world voice, navigation, and interaction data to identify gaps.
User Feedback Loops: Learn from corrections, failed commands, and satisfaction signals.
Periodic Retraining: Regularly fine-tune speech, language, and decision models on new owned datasets.
Personalization Engine: Adapt models to user routines, accents, and preferences.
A/B Testing: Compare model versions for accuracy, latency, and engagement before rollout.
Edge + Cloud Updates: Push lightweight on-device improvements while using cloud for advanced model upgrades.
Safety Monitoring: Continuously track errors, drift, and reliability in live environments. |
We are building for both India and global markets, with India as our initial launch market and validation ground. India offers a large and growing elderly population, rising healthcare demand, multilingual needs, and a strong opportunity for affordable AI-powered care solutions.
Go-to-Market Approach:
Phase 1 – India: Focus on homes, hospitals, and senior care centers with localized languages, cost-effective deployment, and strong healthcare partnerships.
Phase 2 – Global Expansion: Scale to aging markets such as the US, Japan, Europe, and Middle East, where demand for elderly care automation is rapidly increasing. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/YI9Mp43rxZVC-Smart-Humanoid-in-Geriatrics.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/YI9Mp43rxZVC-Smart-Humanoid-in-Geriatrics.pdf |
https://drive.google.com/file/d/1pow9X1kR2xoIlP-3rz7WUsWduUZcEu5D/view?usp=sharing |
Yes, we are building a mission-driven and impact-focused startup centered on improving the quality of life for elderly individuals through AI-powered robotics. Our mission is to make senior care more dignified, safe, affordable, and accessible by combining intelligent assistance, companionship, and healthcare support in one platform.
We address critical challenges such as caregiver shortages, rising elderly populations, loneliness, and safety risks. By enabling seniors to live more independently while reducing the burden on families and healthcare providers, we create both social and economic impact.
Our long-term vision is to build a scalable elderly care ecosystem that not only transforms healthcare delivery in India but also serves aging populations globally. |
Yes We are Incubated in IDRP IIIT Dharward |
NA |
checked |
| Apr 30, 2026 @ 8:26 PM |
KISHORE BS, VADIRAJ A |
kishore@hectomart.com |
https://www.linkedin.com/in/kishorebsonnappa |
https://www.hectomart.com |
9535705206 |
Kishore is an MBA graduate from the Indian Institute of Management, Indore, with over 7 years of experience in the tech services industry. He began his career building products for Fortune 500 companies, later advising executives on commercial excellence and working with private equity firms on large-scale commercial due diligence engagements (over $100M). In his most recent role, he served as a Practice Director at an engineering services firm. He is also a 3x founder, having built ventures across sectors including a fountain beverage company, home construction services, rapid commerce, and e-learning. Vadiraj brings over three decades of experience, having led multiple 0–1 and 1–10 journeys and achieved successful exits as a 2x founder. He has held leadership roles across organizations including Mindtree, Tele brahma, Pine Labs, ONN Bikes, and Acko Insurance, among others. At Acko, he served as VP of Product Management and built the product function from scratch. He is also associated with EK-Step Foundation and has been involved in accelerator programs. Together, Kishore and Vadiraj have previously collaborated on ventures and bring a strong combination of product strategy, operational expertise, and entrepreneurial experience to build and scale their company. |
Kishore and Vadiraj first met around eight years ago through the Founder Institute in Bangalore, where Vadiraj was leading the accelerator and Kishore was a participant. Since then, they have tested ideas across domains including data-driven learning, e-learning, rapid commerce, construction, and consumer products.
After several iterations and learnings, they are now focused on building a large, population-scale problem in the SMB and MSME space. |
All |
Our biggest strength as a team lies in the complementary nature of our experience and execution capabilities. Vadiraj brings deep expertise in building products from the ground up, with decades of experience across 0–1 and 1–10 journeys, strong product intuition, and a proven ability to lead and scale high-performing teams. Kishore complements this with a strong foundation in strategy and operations, shaped by his experience advising CXOs, working on commercial excellence, and executing large-scale diligence engagements. As a computer science engineer, he also brings hands-on product-building experience for Fortune 500 firms and his own ventures. Together, we combine product depth with strategic and operational rigor, enabling us to not only build and scale robust products but also attract strong talent and execute with clarity and speed. |
Baba AI |
www.hectobaba.com |
Kolar, Karnataka |
We build AI coworkers that automate daily business operations for SMB & MSME owners, saving time and improving efficiency. |
Most SMB and MSME owners run their businesses themselves without adopting software or hiring teams. They manage procurement, inventory, customer communication, and operations manually, creating a constant operational burden that limits their ability to scale and keeps them stuck in day-to-day execution. |
Small businesses do not need more software. They need a coworker that can think and act like a business owner. We are building an AI coworker for SMB and MSME owners, an agentic system that handles day-to-day operations across functions such as procurement, logistics, marketing, and order management. These AI agents work under the owner’s direction to execute tasks, coordinate workflows, and make routine decisions, helping businesses operate more efficiently at a fraction of the cost. |
Our defensibility comes from proprietary data and deep workflow integration. In India, less than 5% of SMB and MSME businesses are meaningfully present online, which means the vast majority are not discoverable on search engines or AI assistants. As a result, structured data on these businesses simply does not exist today and is extremely difficult to acquire. By onboarding and working closely with these businesses, we are building a proprietary, high-quality data layer that compounds over time. In addition, our product is embedded directly into daily operations through a high-touch onboarding and execution model. SMB owners require trust and handholding to adopt new technology, and once our AI coworkers are integrated into their workflows, we become a critical part of how the business runs. This creates strong retention and makes it difficult for competitors to replace us, even if they offer similar technology. |
MVP |
Pilots |
Our target customers are SMB and MSME owners operating physical, hyperlocal businesses with small teams, typically fewer than five employees. These owners face a shortage of human capital and manage repetitive day-to-day operations themselves. We serve a broad range of local businesses including retail stores such as kirana, grocery, jewellery, and agri shops; service providers such as plumbers, electricians, carpenters, and construction workers; healthcare providers including clinics, hospitals, and diagnostic labs; as well as small dealerships and real estate operators. |
1.2 trillion |
900 billion |
90-180 billion |
We operate on a hybrid revenue model. SMB and MSME owners can choose between a low-cost subscription plan, priced at approximately ₹150 per month, or a commission-based model where we take around 10% on transactions facilitated through the platform. We also generate additional revenue through advertising, enabling brands to promote their products to distributors and end customers within our ecosystem. |
There is no direct competitor offering a full-stack AI coworker for SMB and MSME operations, which makes this an emerging category. Our primary competition today is the status quo, where business owners manage operations manually without automation. We also compete indirectly with SaaS tools, marketplaces, and fintech platforms that address parts of the workflow but still require active effort from the owner. Over time, larger platforms such as Amazon, Flipkart, PhonePe, and Paytm could expand into this space. However, our focus on building a proprietary data layer and deep workflow integration with SMBs positions us strongly to serve this segment. |
We acquire customers through a field-driven, on-ground distribution model. SMB and MSME owners in India typically require trust and direct engagement before adopting new technology, so our approach focuses on in-person onboarding and activation. We deploy local field teams to identify, engage, and onboard businesses within a specific city and category, following a city-based beachhead strategy. Once we achieve density within a segment, we expand through referrals and local network effects. Early interactions indicate a customer acquisition cost in the range of ₹200 to ₹600 per business, making this model scalable across Tier 2 and Tier 3 cities |
Our go-to-market strategy is built around a city-based beachhead approach. We start by entering a specific city and focusing on a single, high-density business category that is conducive for execution. Within that category, we aim to onboard a large share of businesses and deeply integrate our AI coworkers into their daily operations. Once we establish strong penetration and workflow integration, we expand to adjacent categories within the same city. After achieving multi-category presence in a city, we replicate this playbook across new cities. This approach allows us to build density, strong network effects, and a defensible local presence before expanding. |
Our long-term vision is to build a full-stack AI operating layer for SMB and MSME businesses. We start by providing AI coworkers that execute day-to-day operations, and progressively evolve into an AI advisor that helps business owners make better decisions.
Over time, we aim to build a digital twin of each business, capable of understanding, operating, and optimizing the business like an owner. This enables small business owners to scale beyond their personal bandwidth.
In the long term, we envision extending this intelligence into physical interfaces, bringing AI-powered systems directly into offline businesses. |
Incorporated |
NA |
Yes |
We are applying to Avinya because of its focus on supporting founders building for underserved geographies. While we are not originally from these geographies, our product is specifically designed for SMB and MSME businesses in these markets, which are often overlooked by traditional software and technology solutions. This alignment in vision gives us confidence that Avinya understands both the opportunity and the execution challenges in these segments. We believe this shared perspective will help us build and scale our solution more effectively. |
During the programme, we aim to execute our city-based beachhead strategy by entering a Tier 2 or Tier 3 city in Karnataka, such as Hubli, Dharwad, or Kolar. We will focus on activating a key business segment and onboarding local SMB and MSME owners onto our platform. Our goal is to deploy our AI coworkers across multiple business types, including retail shops, agri-input stores, and healthcare providers such as diagnostic labs, and validate their ability to automate day-to-day operations. By the end of the programme, we aim to demonstrate strong adoption, validate core workflows, and establish a repeatable go-to-market playbook for scaling to other cities. |
Yes |
Yes |
Core engine |
Our AI architecture is built as an agent orchestration system that enables AI coworkers to manage multi-step business workflows. Each agent is modular and designed for specific business functions such as onboarding, operations, support, and data validation.
The system maintains persistent state across sessions, allowing businesses to pause and resume workflows seamlessly. It operates within isolated, multi-tenant environments to ensure data security and contextual accuracy for each business.
We use dynamic schema validation and context-aware routing to adapt workflows based on business type and requirements. This allows us to scale across different SMB segments and rapidly expand into new verticals such as retail, healthcare, and services without rebuilding core infrastructure. |
Our primary data advantage comes from the large portion of SMB and MSME businesses in India that operate offline. Less than 5% of these businesses have a meaningful online presence, which means structured, high-quality data about their operations is largely unavailable to existing platforms and AI systems.
As we deploy our AI coworkers into these businesses, we generate and collect structured operational data across workflows such as procurement, inventory, customer interactions, and transactions. This data is proprietary, continuously updated, and tightly coupled with real business operations.
Over time, this creates a compounding data advantage, enabling our system to better understand, automate, and optimize business workflows. This data layer becomes increasingly difficult to replicate without direct, on-ground integration with these businesses. |
At the current stage, our defensibility is driven by execution rather than static technology. While AI infrastructure is becoming increasingly accessible, our advantage comes from deploying AI coworkers directly into SMB and MSME workflows.
As we onboard businesses, we build deep integration into their day-to-day operations, generating proprietary data, increasing engagement, and improving automation over time. This creates a compounding flywheel where better data leads to better performance, which drives higher usage and retention.
Over time, this combination of proprietary data, workflow integration, and distribution creates a defensible moat that is difficult to replicate without similar on-ground execution. |
We evaluate model performance and reliability across latency, accuracy, and task completion. We monitor system latency, error rates, and fallback usage across agents to ensure consistent performance.
As our system matures, we are tracking key metrics such as intent classification accuracy, task completion rate, average resolution time across workflows, and token efficiency.
All interactions are logged with session-level context, enabling us to continuously analyze performance, debug failures, and improve agent behavior over time. This allows us to iteratively refine both model outputs and orchestration logic to ensure reliability in real-world business workflows. |
We follow a multi-tenant architecture with strict data isolation, where each business operates within its own secure environment. All data access is scoped to the specific business, preventing cross-tenant leakage.
We implement role-based access control to manage permissions across users within each business, ensuring that only authorized actions are allowed. Sensitive data is protected using industry-standard practices such as secure authentication, hashed credentials, and controlled access policies.
We maintain audit logs for key actions to ensure traceability and accountability. As we scale, we are building toward compliance with global and local data protection standards, including support for data residency and enhanced encryption for sensitive data fields. |
There are no immediate regulatory barriers to our product, as we function as an AI layer within SMB and MSME operations |
At 10x scale, the primary challenges will be around orchestration complexity, latency, and consistency across multiple agents and workflows. As the number of concurrent users and tasks increases, coordinating multiple agents in real time while maintaining reliability becomes more complex.
We also expect challenges around maintaining context accuracy and ensuring consistent performance across diverse business types and workflows.
To address this, we are designing our system with modular orchestration, robust observability, and scalable infrastructure, allowing us to monitor performance, optimize workflows, and progressively improve reliability as we scale. |
Currently, AI and ML development is led by the founding team. Both founders are closely involved in product and AI system design, including architecture, workflows, and model behaviour.
We are supported by early-stage contributors on the AI/ML side, and plan to expand into a dedicated in-house team as we scale. At this stage, maintaining tight founder control allows us to iterate quickly on both product and AI systems |
At the current stage, we primarily use publicly available data and synthetic datasets to build and test our AI systems, including product catalog information across categories. |
We're not fine-tuning or retraining yet, that's post-launch. Right now we're at the "instrument and measure" phase. Once we have 6 months of production data, we'll know where to invest in model improvements. |
We are starting with India, focusing on SMB and MSME businesses in Tier 2 and Tier 3 cities, where the need for operational automation is high and software adoption is low.
Over time, we plan to expand to global markets with similar characteristics, including Southeast Asia, the Middle East, and other emerging economies with large offline business ecosystems.
Our long-term vision is to build a globally scalable AI coworker platform for small businesses. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/1s9J4RepagUl-BABA-AI-STORY-Long-Deck-v4001.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/1s9J4RepagUl-BABA-AI-STORY-Long-Deck-v4001.pdf |
http://NA |
NA |
NA |
NA |
checked |
| Apr 30, 2026 @ 7:58 PM |
Anandan Shankaran, Arjun Rao |
anand@mera.health |
https://www.linkedin.com/in/anandans |
https://mera.health |
7349024342 |
Anand has worked at Microsoft, Amazon, Dir (Product and Tech) at a Unicorn and have 2 failed startups. Arjun has worked at LinkedIn, Thumbtack, Head (Eng) at Sugarbox Networks and 1 failed startup. |
Arjun and I have been close friends since 2002, when we were college roommates. Over the past two decades, we attempted several side projects together, so building a company as co-founders has always been the plan, not an accident.
Last year, we both independently decided to wind down our respective startups within months of each other. For the first time in 20 years, we were simultaneously free to work on something full-time. We saw this as a rare window and deliberately chose a problem we both care deeply about personally, which is how mera.health was born. |
All |
Twenty-four years of actual friendship. Not just professional respect, not a LinkedIn connection. Arjun and I have been each other's sounding board through careers, relationships, kids, failures. We've seen each other at our best and worst.
We're both obsessive about technology and products, but we came up through completely different ecosystems. I was deep in Microsoft, he was Linux all the way. Some of our best memories are night-long arguments about .NET vs Java, Windows vs Mac, Android vs iPhone. We have always had strong, different opinions and we've never been shy about saying so.
What that's built over time is the ability to really push each other without it affecting the personal relationship. We know when the other person is right, we know when they're being stubborn, and we're honest about both. That's harder to find than any skill set. |
mera.health |
mera.health |
Bangalore |
mera.health is a preventive health platform built to help families through AI powered clinical intelligence |
Fragmented storage of patient medical information forces every stakeholder in the healthcare chain to pay a different and compounding price for the same missing data and context. |
We are building a platform which combines longitudinal family health data with proactive clinical intelligence. mera.health builds that: a private AI-powered Patient 360 that tracks health across the whole family over time, surfaces what matters, and puts the right information in front of the doctor at the right moment. |
Our USP is a clinical AI pipeline built on Indian-specific datasets (NidaanKosha, LOINC) that doesn't just store health records but reads them, surfaces trends across a family's history, and produces structured doctor briefings before a consultation. We've taken this from concept to a live Android product with 500+ users, with the technical and compliance groundwork (DPDP, ABDM) for multi-stakeholder distribution already underway. We are also in the process of working with hospitals to integrate this to any of their existing EMR systems that way this Patient 360 data is enriched by all parties involved in the care journey. |
Users |
Users, Signups |
We are starting with urban adult who is managing their health, spouse's health, kids and parents health, who care about convenience and care to spend to avoid a health crisis. |
6B USD |
500M USD |
50M USD |
Freemium + B2C subscription (family health records). B2B2C pipeline via hospitals, labs, and corporates. |
eka.care, hospital PHRs, paper/WhatsApp (the default today) |
Paid digital acquisition (Meta/Instagram) + diagnostic lab and hospital partnerships for offline reach. |
Start with urban families managing chronic conditions via diagnostic labs and hospitals. Expand to corporates and insurers for group health record access. |
Every Indian family should have a single, trusted place where their complete health history lives, is understood, and is always accessible. Not just records, but context, continuity, and the ability to make better decisions across generations.
Today, health information in India is fragmented across hospitals, labs, apps, and WhatsApp forwards. Families manage chronic conditions reactively, often without the full picture. We want to fix that, starting with giving families ownership of their own health story.
Long term, mera.health becomes the health memory layer for Indian families. But the bigger opportunity is what that layer enables at scale. When millions of families are actively maintaining their health records, you have something that has never existed in India before: a longitudinal, ground-truth dataset of how disease actually progresses in real populations.
That data layer, combined with AI, changes the game for everyone in the system. Doctors get genuine patient context before a consultation, not a blank slate. Hospitals get intelligence on outcomes and population patterns they can actually act on. Insurers get real risk signals instead of proxies. And at a national level, you can start to predict where chronic disease is heading before it becomes a crisis.
We start by solving a real problem for families. We build trust and a habit. And in doing so, we quietly assemble the dataset that makes Indian healthcare smarter for everyone. As ABDM matures, mera.health is positioned to be the consumer-facing layer that makes that infrastructure real for hundreds of millions of Indians. |
Pvt Ltd Inc in progress. |
NA |
Yes |
We are building in a space that sits at the intersection of consumer behaviour, healthcare infrastructure, and AI. Avinya's focus on deep-tech and health makes it a natural fit. Beyond the programme itself, we want to be in a community of founders and mentors who understand the complexity of building in regulated, trust-sensitive markets in India. |
Three things: sharpen our go-to-market for the first 100,000 users, establish our first hospital or diagnostic lab partnership, and close our seed round. We are at the stage where structured mentorship and the right introductions can compress 12 months of learning into 3. |
Yes, fully open to hybrid or virtual. We are Bangalore-based so in-person works too whenever preferred. |
Yes |
Core engine |
mera.health runs three AI pipelines behind a Supabase edge-function backend.
1. Document extraction: 4-step pipeline on every uploaded report. GPT-4o vision validity check, GPT-4o vision verbatim extraction (markers, values, units, ranges, meds, diagnoses), deterministic LOINC enrichment via Postgres lookup, GPT-4o-mini classification into panels with status assignment. All steps use OpenAI tool-use for structured JSON. A validation layer drops classifier outputs not grounded in the verbatim extraction.
2. Health snapshot: Gemini 2.5 Flash synthesizes a personalized summary from marker history and family profile. Triggered async via pg_net, batches sibling uploads in a 36s window, output post-translated to Hinglish.
3. Doctor-visit prep: GPT-4o-mini generates 3 to 5 contextual questions in English and Hinglish from changed markers. |
The LLM calls are not the moat. Anyone can call GPT-4o. Defensibility lives in three layers around the model.
1. Privacy-first PII scrubbing. We are building a server-side preprocessing layer (OCR plus a small language model running on our infrastructure) that scrubs PII before any cloud LLM call. Names, phone numbers, addresses, and identifiers stay within our systems; only de-identified clinical content is sent to OpenAI or Gemini. Most generic AI wrappers do not do this, and Indian families are increasingly sensitive to where their health data goes.
2. Indian-context normalization. LOINC mapping built from real Indian lab outputs, gender and age-bucketed Indian reference ranges (which differ from US/EU ranges), and a validation pass that drops ungrounded classifier outputs. Codified institutional knowledge, not a prompt. Compounds with every report.
3. ABDM tailwind. As ABDM matures, hospitals and labs will emit FHIR-compatible records. Our LOINC direction means every step of rollout makes our normalization layer more valuable.
A fourth layer accrues over time: switching cost from accumulated family history. Forward-looking, not a moat we have today. |
The LLM calls are not the moat. Anyone can call GPT-4o. Defensibility lives in three layers around the model.
1. Privacy-first PII scrubbing. We are building a server-side preprocessing layer (OCR plus a small language model running on our infrastructure) that scrubs PII before any cloud LLM call. Names, phone numbers, addresses, and identifiers stay within our systems; only de-identified clinical content is sent to OpenAI or Gemini. Most generic AI wrappers do not do this, and Indian families are increasingly sensitive to where their health data goes.
2. Indian-context normalization. LOINC mapping built from real Indian lab outputs, gender and age-bucketed Indian reference ranges (which differ from US/EU ranges), and a validation pass that drops ungrounded classifier outputs. Codified institutional knowledge, not a prompt. Compounds with every report.
3. ABDM tailwind. As ABDM matures, hospitals and labs will emit FHIR-compatible records. Our LOINC direction means every step of rollout makes our normalization layer more valuable.
A fourth layer accrues over time: switching cost from accumulated family history. Forward-looking, not a moat we have today. |
Today, early-stage:
* Per-step pipeline logs (extraction_pipeline_logs): model, tokens, duration, error class, retry count. Tracks cost, latency, per-step error rate.
* Hallucination guard in code: classifier output diffed against raw extraction; ungrounded metrics dropped. Mismatches logged.
* LOINC match rate: % of extracted markers mapping to a known LOINC code. Proxy for extraction quality and alias coverage.
* PostHog funnels: record_uploaded → record_analysis_completed vs failed; snapshot_generated rate; time-to-snapshot.
* Slack alerts on extraction failures.
In progress: Langfuse integration for full LLM-call observability (traces, prompt versioning, cost and latency per step, eval runs).
Planned: per-step error-rate dashboards, OpenAI 429/402/latency tracking, value-type mismatch detection, Sentry for client crashes, and a regression eval set of de-identified reports with labelled ground truth. |
Privacy is a core product principle, not a checklist.
* Server-side PII scrubbing (in build): OCR plus a small language model running on our infrastructure scrub names, phone numbers, addresses, and identifiers before any cloud LLM call. PII stays within our systems; only de-identified clinical content reaches OpenAI or Gemini.
* Encryption: AES-256 at rest (Supabase), TLS in transit.
* Row-level security on every table (medical_records, snapshots, family_members, vitals, profiles). Users access only their own rows. Storage bucket enforces per-user-id paths.
* Auth: phone OTP via Supabase Auth.
* AI processing: documents sent to OpenAI and Gemini for extraction and synthesis only, never for training. DPA with OpenAI in progress; ZDR (Zero Data Retention) agreement planned next.
* Data minimization: only what users upload. No third-party ad/analytics SDKs read PHI.
* Deletion: in-app flow removes account and all data within 30 days.
* Standards: HIPAA-style controls, designing toward DPDP Act 2023 ahead of May 2027 deadline.
Not currently HIPAA or DPDP certified. Planned: edge-function rate limits, WhatsApp webhook signature verification, Sentry, formal access-audit log. |
Yes, three we actively track.
1. DPDP Act 2023. Health data is sensitive personal data. Once Rules and the Data Protection Board are operational, we will need formal consent artifacts, breach-notification SLAs, and possibly a DPO at scale. Compliance deadline May 2027. Deletion, consent, and access controls already designed for this.
2. ABDM and ABHA interoperability. Government push toward Ayushman Bharat Digital Mission may make ABHA-linked records and FHIR table-stakes. Our LOINC and structured-data direction positions us for this; we do not yet integrate.
3. Medical-device risk. If our AI output crosses from "summarization and organization" into "diagnosis or treatment recommendation," it could be construed as Software-as-Medical-Device under CDSCO or FDA. We are deliberately scoped to organize, summarize, and prepare. Disclaimers and product copy reinforce this.
Secondary: cross-border data-transfer rules, OpenAI enterprise terms changes. |
We have thought about this concretely.
1. OpenAI rate limits and cost. Vision calls dominate; current limits become the bottleneck. Mitigations: Gemini fallback (already wired and in production for snapshots), batching, LLM cache, routing typed PDFs through a cheaper non-vision path.
2. Synchronous extraction inside edge functions. Multi-page PDFs take 30 to 60 seconds; cold starts and execution time become a problem. Move to a queue-backed async worker (Supabase queues or pg-boss) with progress streaming.
3. pg_net snapshot trigger. Default 5s timeout and no dead-letter handling will silently drop work. Need a durable job queue with retries and observability.
4. Storage growth. Originals plus base64 copies in logs add up. Lifecycle policies and tiering needed.
5. Postgres query plans. Denormalize markers into a typed lab_markers table keyed by LOINC, add indexes, shard logs.
6. No real eval harness. Regressions become invisible at 10x without one. Near-term roadmap. |
No dedicated in-house AI/ML team. The two co-founders lead AI work directly, both senior engineers with applied LLM systems experience.
We have shipped the full AI stack ourselves: a 4-step extraction pipeline (vision validity check, verbatim extraction, deterministic LOINC enrichment, classification with hallucination guard), LOINC normalization over 6.84M rows of Indian lab data via the NidaanKosha dataset, and async snapshot generation. Multi-provider in production today: OpenAI for extraction, Gemini 2.5 Flash for synthesis.
This is deliberate. The moat in vertical health AI is not in training models, it is in the normalization layer, the eval harness, and the product surface around the model. These are systems-engineering problems. We plan to add a clinical advisor and a senior data engineer once the eval corpus justifies them, not ML researchers. |
* User-uploaded medical records. Owned by the user, processed under our terms with explicit consent. Right-to-process for the service; we do not train models on it.
* LOINC. Open standard, freely licensed by Regenstrief Institute including commercial use. Used for marker code mapping.
* NidaanKosha (Eka Care). 6.84M rows of Indian lab data, CC-BY-SA-4.0. Used for LOINC mapping coverage and Indian reference ranges. Attribution requirements honored.
* Indian reference ranges. Compiled from public clinical sources (textbook ranges, ICMR guidelines, lab-published ranges) and NidaanKosha into our indian_reference_ranges table.
* Test-name alias list. Derived from our production data and NidaanKosha. Owned by us.
* OpenAI GPT-4o and GPT-4o-mini, Gemini 2.5 Flash. Paid API access. We follow each provider's enterprise data policy (no training on inputs).
No licensed third-party medical datasets (UpToDate, MIMIC). No web-scraped corpora. |
A four-loop plan.
1. Eval harness. Labelled regression set of de-identified reports (~1200 to start, growing) covering the long tail of Indian lab formats. Run on every prompt or model change in CI. Track per-marker precision/recall and end-to-end snapshot quality with a rubric. Langfuse integration in progress to anchor this.
2. Production telemetry to dataset. Mismatches caught by our hallucination guard, low LOINC match rates, and explicit user "this is wrong" feedback flow into a triage queue. Triaged examples become eval cases or alias-table additions.
3. The deterministic layer beats the prompt. Many quality wins come from extending the LOINC mapping, value-type validation, and reference-range tables. Gains compound and do not drift with model versions.
4. Model fungibility. Already multi-provider in production (OpenAI + Gemini). Periodically benchmark new models on the eval set and switch when one is meaningfully better at a step's cost/quality point.
Future: fine-tuned classifier, user-in-the-loop correction. |
India-first. Specifically: Indian families managing care across multiple members and multiple labs, with output in English and Hinglish.
The product is shaped by Indian realities: fragmented lab ecosystem, WhatsApp as the default health-records inbox, Hindi/English code-switching, family-led healthcare decision-making, and gender/age-specific Indian reference ranges. We win this market by being native to it, not by being a localized US product.
Global expansion is plausible later. The architecture (LOINC, FHIR-compatible normalization, multi-language) is portable, but it is not on our 18-month roadmap. The sequencing is: own the Indian household, then evaluate Southeast Asia and the diaspora. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/I4hkiUX7hm12-merahealth-acsel.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/I4hkiUX7hm12-merahealth-acsel.pdf |
https://youtu.be/WXp7TieU5go |
Yes, deeply. And I can point to the exact moment the mission became personal.
In December 2020, at the peak of COVID, my father had a stroke in Chikmagalur. He was rushed by ambulance to a hospital in Bangalore. My mother and sister reached the hospital as fast as they could. When the treating doctor asked for his medical history, his medications, his prior test results, there was nothing. No records. The doctor was making critical decisions with minimal context, and my mom were standing there unable to help.
That moment is why mera.health exists.
India has over a billion people who interact with a fragmented, mostly paper-based healthcare system. The cost of that fragmentation is paid in missed diagnoses, repeated tests, delayed decisions, and in moments exactly like the one my family went through. Most of it falls on families, not institutions.
If we can give every Indian family a single, trusted health memory, we reduce that friction. And at scale, the aggregate data from millions of active health records can genuinely improve how care is delivered, how risk is assessed, and how public health is managed. We are not just building a product. We are building infrastructure for a system that desperately needs it. |
NA |
Ramjee Ganti |
checked |
| Apr 30, 2026 @ 7:54 PM |
Tejaswini, Rathika Rani Ponnusamy |
tejaswini@anveshakhub.com |
http://www.linkedin.com/in/tejaswini-490a2285 |
http://www.anveshakhub.com |
08971726699 |
Tejaswini - Managing Director at Anveshakhub, Ex-NXP Semiconductor, Ex-Harman International, Ex-Wipro. Rathika - Director at Anveshakhub Ex-Cassadara Ex- ICT Academy Trainer, Marketing |
We met through a common connection. Having come from smaller towns like Shimogga and Coimbatore we realize that we have similar interests in terms of bridging the gap between Academia and Industry specifically MSME's but not limiting to.
We have worked together for 7 months now. |
2 |
Both founders share different skillsets. Tejaswini comes from Technical background and handles Technology and Business scale, Rathika handles Marketing and Sales. This is the biggest strength. |
Anveshak Hub Private Limited |
www.anveshakhub.com |
Bangalore |
Deep Tech, research and innovation focused startup acting as a catalyst to bridge the gap between Industry particularly MSME's and Academic Institutions |
India's R&D spend is ~0.64% of GDP, much much lower than China ( ~2.6% of GDP) , US (~ 3.6% of GDP) , Japan ( ~3.4 of GDP ). Not all industries/Enterprises/organizations have access to academic research capabilities to carry out world class research and Innovation, particularly MSME's and smaller hospitals or diagnostics centers and hence there is a huge gap in original innovation indigeneous in India. Additional hindrances to such innovative ecosystem is the lack of knowledge that such sources exist. Even If its know, industries don't have that kind of time at hand to establish and connect to such sources themselves. |
Anveshak Hub acts as an Institutional R&D Partner that synchronizes the intellectual capital of academia with the commercial urgency of the industry.
I. Strategic R&D & Talent Funding
We identify critical industrial/diagnostic problem statements and map them to the expertise of professors and labs at BITS and IITs. Unlike traditional models, Anveshak funds specific outcomes, not just resources. We fund and manage PhD/MTech scholars to solve these problems, providing a structured bridge from university labs to the factory floor.
II. End-to-End TRL Management
We de-risk the innovation process by providing professional Program and Project Management.
Bridging the Gap: We handhold projects from TRL 1 (Basic Principles) to TRL 9 (System Proven in Operational Environment).
Expert Oversight: Technical mentorship is provided by a network of Post-Docs and Industry Veterans to ensure academic rigor meets industrial standards.
III. IP Commercialization & Knowledge Governance
Anveshak eliminates the legal friction that often kills industry-academic collaborations. We facilitate and manage:
IP Strategy: From provisional filing to final patent grants.
Equitable Licensing: Drafting frameworks that protect the interests of both the university and the industrial partner to ensure a smooth path to product realization.
IV. Future-Proofing via Upskilling
To ensure the technology we build is actually adopted, we provide a top-to-bottom upskilling framework. We bring university faculty and industry experts together to train everyone from shop-floor technicians to C-suite executives, ensuring the workforce is ready for the "Technology Tsunami." |
Offering research resources to MSME's is unique in itself because these organizations have fewer resources for tech building and are constantly in pressure to show better revenue and profit. Offerings tailored to MSME's can level up the innovation, quality and process . Many of these are suppliers to large OEM's when the re is innovation and quality in its supllier base we naturally become more competitive to global markets with indigenous solutions. We are also collaborating with Government bodies who are supporting thsi segment with various schemes like |
MVP |
Revenue, Pilots |
Industrial Manufacturing, Medical devices, Automotive companies etc |
₹ 5,500 Crores ( Total ~55000 PhD's - Engineering PhD's across all Universities/colleges can be mapped to Industry sponsored project) with each PhD we can make a margin of 20 % to 40 % |
375 Crores ( 25% of 15000 PhD's - Engineering PhD's in BITS and other Top 25 colleges ) |
3.75 Crores ( 10% of SAM = 1500 PhD's) |
Research as a Service via PhD sponsorship with BITS as lead institute ( 55%) + Upskilling and Training ( 10%) + Deep Tech Consulting(10) + IP and Royalty ( 25%) |
THub |
Known connects + Marketing reach |
Onboarded GTM consultants on success fee based model |
Foundational years ( Year 1 to Year 3 ) -
The first three years are dedicated to building a robust operational and technical baseline:
Infrastructure: Developing custom software architecture and hardware facilities.
Team: Onboarding 3-4 full-time mentors (including MD) to oversee an initial cohort of 18-24 PhD students.
Anchor Phase: Securing BITS Pilani as the anchor institute and signing the first wave of anchor industry clients.
Grants: Fast-tracking development through ANRF, ICMR, and other government R&D grants.
Scaling years ( year 4 to year 10 ) - Scaling the RaaS model to its Serviceable Addressable Market (SAM) at 3,750 students. Revenue streams diversify into training, consultancy (20%), and IP commercialization (25%). |
Micro |
Applied for 38 Crores with ANRF and ICMR |
Yes |
Incubation and Grants |
Build AI stack for Machine health prediction using available data sets |
Yes |
Yes |
Core engine |
Problem Translation & Research Layer (Academic Foundation):
Industry-specific challenges from MSMEs (e.g., predictive maintenance, process optimization, quality inspection, supply chain anomalies) are formalized into well-scoped research problems. These are addressed via sponsored PhD programs using rigorous methodologies including:
Deep learning architectures (CNNs, Transformers, Graph Neural Networks)
Time-series forecasting (LSTM, GRU, Temporal Fusion Transformers)
Reinforcement Learning or Multi-Agent Systems for optimization problems
Hybrid physics-informed neural networks (PINNs) where domain knowledge is critical
Research outputs emphasize explainability (SHAP, LIME, attention visualization) and generalizability across constrained industrial datasets.
Example 1 : Current sensor development focuses on reading machine OEE characteristics, showing the machine status on a Digital App and showing the predictive analytics on a subscription basis.
Model Development and Training Layer -
Models are developed using modern frameworks such as PyTorch (primary) and TensorFlow, with heavy reliance on:
fine-tuning of pre trained models (e.g., ResNet, EfficientNet, Vision Transformers, BERT variants for industrial text/logs)
Self-supervised and semi-supervised learning techniques to handle limited labeled industrial data
Federated or distributed learning approaches when data privacy is a concern
Training occurs on a mix of on premise GPU/TPU clusters, cloud resources, and academic high-performance computing facilities, optimized for reproducibility and versioning (using tools like DVC and MLflow).
Edge AI Optimization layer and deployment -
Solutions are engineered for real-world MSME environments characterized by limited compute, intermittent connectivity, and harsh shop-floor conditions.
System adopts an "Edge-First" approach rather than a "Cloud-first" one. Architecture relies on
Model Compression: Using techniques like quantization and pruning to fit models onto local compute units.
Hardware-Level Optimization: Using runtimes (like TensorRT or GGUF) that are specifically tuned to edge hardware's instruction sets.
Hybrid Inference: A system that handles latency-sensitive tasks (like real-time feedback) on device using these compact models, while offloading only the most complex, non-time-critical reasoning tasks to the cloud.
Example 2 : Development of a digital twin of the laser cutting process to
predict the quality of the outcome and reduce waste for a leading manufacturer. Prototypes of new-generation electrical cores with enhanced performance characteristics, such as reduced core losses and increased power density. |
Our long-term proprietary edge is depth of data, relevance, and commercial validation. In the Indian MSME context, where high-quality labeled industrial data is scarce and general public datasets perform poorly due to domain shift, our ability to generate and own problem-specific, deployment-tested data plus IP provides a defensible moat.
We treat data as part of the IP & Knowledge Management pillar, with structured processes for protection, licensing, and commercialization. This aligns with our diversified revenue model (core RaaS + training/consultancy plus IP monetization).
We are actively investing in internal research capacity to accelerate dataset creation, versioning (using tools like GitHub), and MLOps practices to strengthen this advantage over the next 3–7 years. |
We maintain a strong IP clause in our agreements with clients and Academic Institution. models that are trained on specific data and quantized for edge becomes our assets. We accumulate cross project insights and improve future solutions via transfer learning. |
MAE (Mean Absolute Error): Measures the average magnitude of errors in the same units as your sensor data. It is highly interpretable and less sensitive to outliers, making it a "stable" view of performance.
RMSE (Root Mean Squared Error): Gives a higher penalty to large errors due to the squaring effect. In sensor AI, this is critical if large deviations represent a failure state or dangerous condition, as it highlights models that are "consistently wrong" in extreme cases.
MAPE (Mean Absolute Percentage Error): Useful for relative comparison, as it shows error as a percentage of the actual value. This is especially helpful if your sensor values vary significantly in magnitude (e.g., pressure readings across different industrial ranges). |
Edge-First Processing
Architecture is designed to process data locally on the edge device whenever possible. By keeping raw sensor data local and only transmitting insights or metadata to the cloud. |
1. Maintaining detailed technical documentation and "Audit Logs"
2. Edge First approach there by reducing the exposure of raw data compared to cloud only solutions. |
1. Latency in data transmission could occur when scaled to 10x
2. Data storage and computational capacity (in case ANRF funding doesn’t come through) |
Yes. AI/ML - Experience on regression models, classification, Clustering and Timeseries
Deep Leaning and CV - CNN, RNN.LSTM, Object detection.
NLP : Lexical and Semantic Processing , Word2Vec, Seq2Seq, Encoder-Decoder, Transfer learning |
Raw data of Current, Voltage and power from machines in the shopfloor. These data sets are owned. |
We don't just rely on static training sets, we maintain a "Data Pipeline" that continuously captures sensor data to refine our features, ensuring the model evolves with the physical environment. |
Both India and Global markets. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/NqMpCEifPUM3-Anveshak-Hub.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/NqMpCEifPUM3-Anveshak-Hub.pdf |
https://anveshakhub.sharepoint.com/:f:/s/AnveshakHubTeam/IgA64KbTN2lyTKIpS9LYV5hiATIJChD47UiaOJ9JSvChF7o?e=wCjVal |
Our mission is to democratize access to AI-driven industrial intelligence to MSMEs in the Manufacturing sector by connecting them to Academic research solutions. |
Yes. Women Founders only. |
Ajay Krishnan from KDEM |
checked |
| Apr 30, 2026 @ 7:40 PM |
Debashree Basu, Deboshree Roy |
debashree@kreatr.in |
https://www.linkedin.com/in/meetdebashree/ |
https://www.linkedin.com/company/kreatr-ai%20,%20https://www.kreatr.in/%20 |
9901888443 |
Debashree Basu - CEO & CTO, Deboshree Roy - CPO and CBO |
We were pursuing our own startup ideas and have been connected through different startup communities. While Debashree pivoted the idea from K-12 learning to AI Video Localization, Deboshree Roy started resonating with the new idea space, and recently joined in. We have been working together for about 2-3 months. |
2 |
Yes, a passionate team with strong expertise in AI Creative tools and models along with the sound & video editing and domain expertise on content types and narrative styles. |
Kreative Media AI Labs Private Limited |
https://www.kreatr.in |
Bengaluru |
Seamless Storytelling Localization at Scale |
The "Language Gap" is the single biggest blocker, for access, and for new markets at scale where less than 20% of the population worldwide speaks or understands English and closer home, in India, almost 90% of the population is not comfortable with English. The Short-form video market for Storytelling, Microdrama, Ads and Game Dubbing, is expected to reach $200B by 2030 (above 30% CAGR!). Explosive internet adoption in developing nations has unleashed a massive new audience onto Digital Learning and short-video platforms / OTTs. Dubbing has been a very painful process, tied with extremely tight timelines with massive volumes of data which frequently need to be converted to multiple languages. Hiring translators, linguists and voice-over artists, for every language and delivering with the same quality respectively, is a massive blocker in scaling content dubbing. AI Agentic workflow can be a great enabler to such volumes at scale. |
Content is created or produced once, but it needs to be localised for a wider multilingual market at scale, to different geographies.
Kreatr provides an AI-powered Video Localization Platform, including Translation, Dubbing and Lipsyncing that
Key Features - Kreatr AI Localization Platform:
- 80% Automated: Integrates an end-to-end pipeline in just one Workflow, with Human-in-the-loop
- Realistic Quality: Produces realistic, consumable localized content
- Automatic A/V Sync: Audio-Video sync for each dialogue, in any language
- Expressive narration: With emotion, character and voice nuances
- More than 50+ languages and dialects covered
- Speed: End-to-end achieved in 10-15 minutes for a short-form video (1-3 minutes)
-Bulk mode of Translation of content
-High-Accuracy: 90% Accuracy with Automated QC for translation quality and correctness with Semantic Scoring
We are early revenue in the Digital Learning segment and can be classified as TRL7. |
Highly Accurate tranlate with Automated QC , Bulk translanslation for scale and Human-in-the-loop orchetration |
Revenue |
Revenue |
Digital Learning providers, Animation & storytelling |
$200B |
$20B |
$2B |
B2B partnerships |
Deepdub, Dubverse |
1. Strategic Outbound: * Ideal Customer Profile (ICP) * Personalized Outreach 2. Product-Led Growth (PLG) Elements (Iterative Approach): * Interactive Demos & Proof of Value 3. Leveraging Industry Presence & Community Engagement: * Relevant Events * Digital Community |
Our GTM strategy for B2B mid-market SaaS companies is a multi-channel approach focused towards demonstrating clear, quantifiable value and building targeted relationships |
A strategic framework for Kreatr AI’s long-term growth
1. Dominate High-Volume, High-Frequency Niches First
2. Transition from "Tech-Enabled Service" to "Enterprise API"
3. Build a Proprietary Data Moat |
Yes |
NA |
Yes |
Mentorship, Market Access and Funding |
PMF and the right direction to choose |
yes |
Yes |
Automation layer |
Kreatr's Agentic AI Localization Platform generates multi-speaker localized content, from translation, dubbing to lipsync, by integrating multiple AI tools, like Elevenlabs, Gemini TTS, multimodal llms for Visual understanding, Nano Banana, Kling 2.0, Veo 3.1, Heygen with a audio and video processing framework . We integrate with prompt intelligence to generate the expressive, realistic narratives and stitch them together for a superior quality content in minutes |
Complex Multispeaker localization |
Multispeaker localization with microsecond accuracy, scene wise analytics and bulk translation with 90% accuracy and automated cultural and contextual QC |
NA |
yes, through model based harm conditions and prompt intelligence based multi-stage checks |
No |
ye sproduction pipeline will need a rejig at that scale |
Yes, with strong expertise in AI, video/audio frameworks and domain understanding |
NA |
Fine-tuning to bring in domain intelligence |
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/bSFXSuXrL5h3-Business-Pitch-Deck-Kreatr-AI-1.zip, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/bSFXSuXrL5h3-Business-Pitch-Deck-Kreatr-AI-1.zip |
https://youtu.be/YTUneWHESTg |
Our mission is to make localization seamless. In a world where a plethora of languages exist, creation being a tedious, time-taking process, has remained stuck at one language, which was primarily English. Our moto is create local, Go Global, with seamless localization! |
Women |
NA |
checked |
| Apr 30, 2026 @ 5:28 PM |
H Shreyas ,Dr Raja N Moorthy, A Vaidyanathan |
shreyas@anilyx.com |
https://www.linkedin.com/in/shreyas-h-27197a402/ |
http://www.anilyx.com |
7349777288 |
A. Vaidyanathan – Co-Founder & Director : Tech entrepreneur with two successful exits and extensive experience in product strategy, business development, and technology commercialization. Advisory Team H. Shreyas – Co-Founder & Director : Electronics Engineer with strong experience in product innovation, embedded systems, and hardware design. Responsible for system architecture, sensor integration, and smart helmet product development. Dr. Raja N Moorthy – Co-Founder & Director : PhD with over 40 years of experience in Research & Development, innovation leadership, and technology commercialization. Dr. Moorthy has led multiple deep-technology initiatives and provides strategic direction for the Anilyx platform. Dr. Manas Gaur – Co-Founder : PhD (AI/ML) and Advisor – University of Maryland, Baltimore County (UMBC), USA. His expertise includes Artificial Intelligence, dat |
The founding team came together through a shared interest in deep-tech innovation, environmental health, and smart mobility solutions. The founders bring complementary expertise across technology commercialization, electronics engineering, AI/ML, and research & development. Collectively, the team has worked together across product innovation, hardware design, and technology strategy for several years while building Anilyx. |
All |
Our biggest strength lies in the combination of deep domain expertise, technical execution, and commercialization capability. The team brings together experience in AI, embedded systems, R&D, environmental sensing, and startup building. This multidisciplinary background enables us to rapidly move from concept to prototype while maintaining strong product-market alignment. |
Anilyx Labs Pvt Ltd |
www.anilyx.com |
Bangalore |
Anilyx develops AI-powered smart environmental monitoring solutions integrated into wearable systems to protect individuals from pollution exposure while generating hyperlocal environmental intelligence. |
Urban air pollution is one of the most severe climate-linked health risks, particularly for two-wheeler riders. It is estimated that the pollution expsoure is equivalent to 8 to 14 Cigerettes per day* due to exposure to PM2.5, Carbon Monoxide and other gaseous emissions. India has over 260 Mn two wheelers on the road growing at 9% p.a. Existing protection solutions such as Masks, Balaclava are uncomfortable for long rides, ineffective against multiple pollutants, and impractical to integrate with helmets; Pollution combined with Heat Stress (high temperatures exposure in scalps), causes Hair loss & skin irritation. The Standard Helmets protect the head from impact, but do not solve problems created by Air Pollution and Heat Stress |
The Anilyx Smart Helmet aims to significantly reduce the daily air pollution exposure of two-wheeler riders, particularly in densely populated urban environments across India where air quality often exceeds safe limits defined by the World Health Organization. By providing real-time protection against particulate matter, toxic gases, and biological contaminants, the system can help lower long-term risks of respiratory illnesses, cardiovascular diseases, and pollution-related health complications among commuters and delivery riders. Beyond individual health protection, Anilyx also contributes to broader climate and environmental impact by creating a distributed network of mobile air quality sensors. Aggregated data from connected helmets can generate hyperlocal pollution insights that support urban planning, traffic management, and environmental research. By combining personal protection with crowd-sourced environmental intelligence, the solution has the potential to improve public health outcomes while enabling data-driven strategies for cleaner and more sustainable cities. |
Anilyx combines wearable pollution protection with a distributed environmental intelligence network. Unlike traditional helmets or masks, the Anilyx Smart Helmet integrates advanced filtration, environmental sensing, AI-driven analytics, and connected IoT capabilities into a single user-friendly platform. The defensibility comes from proprietary sensor integration, data collection capabilities, AI-driven predictive analytics, and the creation of hyperlocal pollution datasets generated from real-world rider movement. |
MVP |
Pilots |
Urban two-wheeler riders Delivery and logistics personnel Gig economy workers Health-conscious commuters Fleet operators and mobility companies Government and smart city initiatives for environmental monitoring |
India has over 260 million two-wheelers, representing a massive opportunity for pollution-protection and smart wearable safety solutions. |
Urban two-wheeler users across high-pollution metropolitan cities in India, estimated at over 80 million riders. |
Initial penetration targeting metro cities and delivery ecosystems, estimated at 1–2 million users over the first few years |
Hardware sales of smart helmets Subscription-based environmental monitoring dashboard Data analytics services for smart city and environmental agencies Enterprise partnerships with fleet and delivery operators |
Traditional helmet manufacturers Air filtration mask companies Smart wearable health-tech products Emerging IoT-based environmental monitoring startups |
Direct-to-consumer online channels Partnerships with delivery and mobility companies Collaborations with helmet distributors and retailers Smart city pilots and institutional partnerships |
Anilyx plans to enter the market through pilot deployments with delivery fleets and urban riders, followed by D2C expansion through online sales channels and strategic partnerships. The focus is on high-pollution metro cities where awareness and adoption potential are highest. |
To become a global environmental intelligence platform that combines wearable protection, real-time sensing, and AI-powered analytics to improve public health and create cleaner, safer cities. |
Private Limited Company – Incorporated in India. |
NA |
Yes |
We are applying to Avinya to gain mentorship, ecosystem access, pilot opportunities, investor connections, and strategic support to accelerate commercialization of our climate-focused smart wearable platform. |
Validate product-market fit Strengthen AI and IoT product architecture Build strategic partnerships Secure pilot deployments Prepare for fundraising and scaling |
Yes |
Yes |
Supporting feature |
The Anilyx platform uses an AI-driven architecture that combines IoT sensor streams, environmental monitoring data, and predictive analytics models. Sensor data collected from helmets is processed through cloud infrastructure where AI models analyze pollution patterns, rider exposure levels, and environmental trends. The system generates actionable insights through dashboards and predictive alerts. |
Anilyx generates proprietary hyperlocal environmental datasets collected from real-world rider movement and wearable sensor networks. This creates a unique data layer that is difficult for competitors to replicate. |
Defensibility comes from proprietary environmental datasets, hardware-software integration, sensor calibration techniques, and continuously improving prediction models trained on real-world mobility data. |
We evaluate model performance using accuracy, anomaly detection precision, sensor calibration validation, prediction confidence scores, and real-world validation against environmental benchmarks |
Data is anonymized and encrypted during transmission and storage. The platform follows privacy-first principles, secure API communication, cloud access control, and compliance-ready data governance practices. |
Potential risks include wearable device compliance, environmental monitoring regulations, and data privacy requirements. However, the solution aligns with increasing demand for climate-tech and public health innovations. |
Scaling challenges may include increased sensor data ingestion, cloud infrastructure costs, model retraining complexity, and real-time processing requirements. These are mitigated through scalable cloud architecture and modular AI pipelines. |
Yes. The team includes AI/ML expertise through advisory and technical leadership focused on predictive analytics, intelligent systems, environmental data processing, and scalable cloud-based AI deployment. |
We use a combination of publicly available environmental datasets, sensor-generated internal datasets, and pilot-collected data streams. Over time, proprietary data ownership will become a significant asset. |
Continuous model retraining, sensor calibration feedback loops, real-world validation, and expanded data collection will improve prediction accuracy and system intelligence. |
Initially India-focused, with long-term expansion into global urban mobility and environmental health markets. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/sY4Coc4XkhdJ-Avinya_2026.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/sY4Coc4XkhdJ-Avinya_2026.pptx |
https://drive.google.com/file/d/1dGidF1wVxDiYUi5WJSD7TXuIs5I8BHvy/view?usp=sharing |
Yes. Anilyx is mission-driven and focused on reducing pollution-related health risks while contributing to environmental intelligence. By protecting riders and generating hyperlocal pollution data, the platform supports healthier cities, climate awareness, and data-driven urban planning. |
NA |
NA |
checked |
| Apr 30, 2026 @ 4:58 PM |
Adarsh S Punit |
ceo@sonative.in |
https://www.linkedin.com/in/adarsh-s-punit-322b901b9/ |
https://www.sonative.in/ |
8929558491 |
Adarsh S Punit - CEO (15+ yrs exp in health & fitness, Manipal University), Sumit Singh - CMO (8+ yrs exp in FMCG and retail market), Raju Prajapati - CSO (Food Tech Strategies and IIT Guwahati), Devesh Anand - R&D (Food Engineering, IIT Kharagpur) |
Adarsh S Punit initiated the company based on his experience in health and fitness and was soon joined by his friend and batchmate, Sumit Kumar Singh, who aligned with the vision from the outset. Raju Prajapati later came on board after meeting Adarsh at an event and resonating with the problem and approach. He subsequently brought in Devesh Anand, recognizing a strong fit in terms of technical expertise and alignment with the company’s goals. Together, the team combines complementary strengths across domain knowledge, market understanding, and product development. |
All |
Our biggest strength is our complementary expertise across health, FMCG, and food technology, which allows us to move seamlessly from idea to product to market. Combined with strong alignment and quick execution, this enables us to build practical, science-backed solutions efficiently. |
SoNative Pvt Ltd |
NA |
Bengaluru |
Nhance.health is an AI-powered science-backed platform delivering personalized nutrition solutions tailored to individual needs and lifestyles. |
Nutrition today is generic and one-size-fits-all, failing to meet individual health needs and leading to ineffective outcomes. |
Nhance.health delivers guided, outcome-based nutrition by combining user health data with AI-driven insights and expert consultation to create personalized nutrition plans. It then provides targeted supplementation and continuously tracks progress through a real-time dashboard, ensuring measurable and clinically relevant health outcomes. |
Our defensibility comes from combining AI-driven personalization, expert validation, and continuous outcome tracking into a closed-loop system—linking data, recommendation, product, and results. This integrated approach not only improves effectiveness but also builds proprietary user data and feedback loops that strengthen accuracy and long-term differentiation. |
Idea |
Testimonials |
Age 25 to 40. Health-conscious consumers and Fitness enthusiasts |
$ 75,00,000 |
NA |
NA |
Subscription based and D2C sales |
HealthifyMe, Qua Nutrition, Nutrova, iThrive |
We acquire customers through on-ground awareness initiatives such as corporate and park activations focused on personalized nutrition, where we distribute free consultation coupons to drive trials. In parallel, we partner with quick commerce platforms to distribute coupons alongside product purchases, enabling direct conversion from product users to personalized nutrition services. |
Our GTM follows a trust → conversion → retention funnel: we acquire users through corporate wellness programs and community activations with low-cost trials, convert them via platform onboarding with health scores and dashboards, and retain them through personalized nutrition kits and subscription plans, scaling further through corporate, healthcare, and residential networks. |
To become the most trusted brand for guided nutrition, empowering individuals and families to take control of their health. |
Incorporated |
NA |
No |
We are applying to Avinya Acsel Technology Forum to gain mentorship, industry connections, and strategic support to refine our technology, accelerate validation, and scale our personalized nutrition platform effectively. |
During the programme, we aim to validate our product-market fit, strengthen our technology and personalization engine, onboard early users through pilot programs, and build the right partnerships to scale our platform efficiently. |
Yes |
Yes |
Core engine |
Our planned AI architecture will take user health data (lifestyle, diet, and goals) as input and use a combination of basic rules and machine learning models to generate personalized nutrition recommendations. Over time, user feedback and outcomes will be used to improve the system and make the recommendations more accurate. |
We currently do not have proprietary data, but we plan to build a strong data advantage by capturing user health and outcome data over time, and by exploring access to anonymized clinical care data through third-party collaborations to improve personalization and accuracy. |
Our AI becomes defensible through a closed-loop system that continuously learns from user data, preferences, and outcomes, while being grounded in strong scientific evidence and trained on clinical research and trial data. As we scale, this builds a proprietary dataset and improves personalization accuracy, making it increasingly difficult for competitors to replicate. |
We evaluate model performance using metrics like recommendation accuracy, user adherence, and improvement in health outcomes. Reliability is tracked through user consistency, feedback ratings, and expert validation, ensuring that recommendations remain safe, relevant, and effective over time. |
We follow a privacy-first approach by collecting only necessary user data, storing it securely with encryption, and ensuring access is strictly controlled. We plan to comply with relevant data protection regulations, use anonymized data wherever possible, and maintain transparency with users on how their data is used. |
Yes, there are potential regulatory risks around data privacy, health claims, and nutritional products. We plan to mitigate these by complying with data protection laws, ensuring all recommendations are evidence-based, and aligning our products and communication with relevant food and healthcare regulations. |
At 10x scale, challenges could arise in maintaining recommendation quality, handling large volumes of user data, and ensuring consistent personalization across diverse users. We plan to address this by strengthening our data pipelines, improving model robustness, and incorporating continuous monitoring and expert validation to maintain accuracy and reliability. |
No |
We are currently not using any proprietary datasets, as the system is in the early planning stage. We intend to build our models on publicly available scientific literature and licensed datasets where required, and over time develop our own proprietary dataset through user data and outcomes. |
We plan to continuously improve model performance through a feedback loop that incorporates user data, adherence, and health outcomes. This will be supported by periodic model updates, integration of new scientific research, and expert validation to ensure recommendations remain accurate and relevant over time. |
We are initially building for India, focusing on urban users, with a long-term vision to expand into global markets as we refine our personalization models and platform. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/e4DEiWjzZoSK-Nhance.health-2.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/e4DEiWjzZoSK-Nhance.health-2.pdf |
http://NA |
We are a mission-driven startup. Our mission is to make personalized, science-backed nutrition accessible and actionable, helping individuals and families make better health decisions every day. By moving away from generic, one-size-fits-all approaches, we aim to improve trust, adherence, and real health outcomes at scale. |
No |
NA |
checked |
| Apr 30, 2026 @ 4:45 PM |
Varsha Agarwal, Manish Agarwal |
varsha@assess.plus |
https://www.linkedin.com/in/varsha-agarwal-change-maker/ |
https://assess.plus |
09880463460 |
Varsha Agarwal - Co-Founder and COO - 25 years of experience spanning engineering, technology leadership, and enterprise operations. As Engineering Director at VMware, she led the engineering organisation responsible for $1B+ subscription revenue systems — including AI-driven platforms and enterprise customer data infrastructure — delivering measurable outcomes at scale with full compliance and governance. At Assess.Plus, she owns operations, enterprise relationships, and go-to-market execution. Her network in enterprise hiring circles, combined with her role as Karnataka State Vice President for the AI/ML stream at WICCI, positions her as the bridge between the product and the market it serves. | Manish Agarwal - Co-Founder, CEO and CTO - Computer Science graduate from IIT with 23 years of experience across the full technology ladder — engineering, architecture, and product — at global organisations including Motorola, Samsung, and Mavenir. Conceived the idea of a structured assessment platform in 2023, built and launched an early version as an MCQ-based assessment infrastructure, and recognised the opportunity to rebuild it ground-up with full AI capability — evolving it into Assess.Plus in late 2025. Leads product vision, platform architecture, and technical execution at Assess.Plus. Patent applications have been filed on the core methodology and intelligence pipeline. |
We are siblings, which means the relationship predates the startup by decades. We began working together formally when the decision was made to rebuild the assessment platform from the ground up as a fully AI-enabled product, combining technology leadership with enterprise and operational experience to build and take it to market together. |
2 |
Two things that reinforce each other. The first is complementary expertise with close to 50 years of combined experience; one founder owns the technology, architecture, and product, the other owns enterprise operations, market relationships, and execution. Together, the team covers the full journey from building to scaling without a gap. The second is decision-making clarity that comes from a sibling relationship built over decades. Every opinion, hesitation, and disagreement is laid on the table openly, there is no ego management, no agenda behind any position, and no concern about how something will be received. Decisions are fast, honest, and final. |
Assess.Plus |
https://assess.plus |
Bengaluru |
We are the KYC for the hiring pipeline — Know Your Candidate, before they walk in. |
Hiring pipelines are drowning in resumes — and the problem is getting worse. Candidates now use AI to make every resume look perfect, making it nearly impossible for HR teams to separate real talent from polished noise. The burden falls on Hiring Managers, who burn 50 hours per hire — an entire work week — on telephonic rounds just to find who is actually worth meeting. The result: best people are exhausted, hiring pipeline is slow, and wrong hires still happen. The problem was never too many resumes. It was never having a verdict at the entry point. |
Assess.Plus changes the fundamental unit that flows through the hiring pipeline. Instead of resumes, verdicts flow.
The product centres on mapIQ, a proprietary AI engine that takes any job description, ingests company-specific hiring intelligence, and maps it across 65+ parameters spanning assessment purpose, role seniority, domain knowledge, cognitive complexity, competency frameworks, and role-specific psychological traits. This is not a question picker pulling from a static library — it is a live intelligence engine that re-engineers itself from scratch for every JD, every company, every hire. The output is a precisely calibrated assessment, built for that exact role, industry, and seniority level.
The assessment is then distributed frictionlessly — via link, QR code, email, or embedded directly in a job posting. Candidates scan and attempt from any device. The evaluation engine scores every response in real time. Each candidate receives instant, personalised feedback on their performance.
The talent team receives a verdict in three layers: an overall score that gives an instant read on suitability, a skill-wise breakdown that shows exactly where a candidate is strong and where they fall short, and a psychometric trait report that maps their personality against the traits that actually predict performance in this specific role. Not a generic personality profile, but a role-calibrated signal. Together, the three layers give talent teams a complete, evidence-based picture of every candidate.
The result is a bulletproof shortlist that arrives at the hiring team's desk without a single discovery call to validate basics. The pipeline no longer begins with a resume. It begins with proof. |
Every competitor starts from a question library and filters toward a role. Assess.Plus starts from the role and generates the assessment. That architectural inversion — JD in, calibrated assessment out — is the core of what makes this structurally different and structurally defensible. mapIQ, our proprietary AI engine, re-engineers each assessment from scratch for every JD, every company, every seniority level. A FinTech Sales Manager and an FMCG Sales Manager receive fundamentally different assessments. This cannot be replicated by connecting a language model to a question database — the library is the constraint, and incumbents are built on it. Three further edges compound this: Breadth without compromise. 23 role buckets across Finance, Healthcare, Legal, Manufacturing, Creative, Public Sector, and more. A hospital hiring a Clinical Officer gets an assessment as precise as a startup hiring a backend engineer — something no library-based tool can deliver across non-tech domains. Role-calibrated psychometrics. Every existing tool gives the same generic personality questionnaire regardless of role. Assess.Plus identifies and assesses only the traits that predict performance in that specific role. The output is actionable signal, not a score teams have to interpret themselves. Most candidates who attempt assessments elsewhere receive nothing back. Assess.Plus gives every candidate instant, personalised performance feedback. That experience is uncommon enough that candidates remember it — and the hiring companies that offer it stand out. Word spreads organically, without a marketing budget. Defensibility is structural: proprietary lightweight models with patent applications filed, and an assessment quality engine that enforces scientific integrity even when recruiters attempt to override it. Incumbents cannot close this gap by retraining — they would need to rebuild from scratch. |
MVP |
Pilots |
Assess.Plus targets companies running high-volume hiring where senior talent teams are spending their highest-value hours in screening and telecons, and the cost of a wrong hire is high, but the hiring infrastructure is thin. Our primary segment is mid-to-large enterprises in BFSI, Business Services, Healthcare, and Manufacturing sectors, characterised by recurring mass recruitment, structured role profiles, and HR teams that rely almost entirely on resume screening and telecons. These organisations hire hundreds to thousands of candidates annually, yet have no assessment layer in their hiring infrastructure. The pain is acute, the budget exists, and the switching cost is zero because there is nothing to switch from. Our secondary segment is growth-stage technology companies scaling beyond 100 employees, past the point where ad hoc hiring breaks down, and actively looking to standardise on a serious assessment platform for the first time. The buyer is the Head of Talent Acquisition or CHRO, someone who feels the hiring cost and inefficiency directly and has the authority to drive technology adoption. |
INR 13000 Crore (India Market) |
INR 6500 Crore (India Market) |
INR 100 Crore (India Market) |
Assess.Plus operates on a SaaS subscription model with consumption-based top-ups, designed to grow with the customer's hiring volume. Base Packages — Monthly Subscription Two product tracks, each available in tiered plans: • Skill Assessment + Psychometric — full-verdict package: JD-generated skill assessment + role-calibrated psychometric report • Skill Assessment Only — skill verdict without the psychometric layer Each plan carries two caps: (1) number of JDs for which assessments can be generated in a billing cycle, and (2) number of candidate attempts allowed. Higher-tier plans carry higher caps and lower per-unit effective cost — incentivising volume commitment. Top-Up Add-Ons Clients who exhaust either cap mid-cycle can purchase top-ups: additional JD slots, additional attempt credits, or both — without changing their base plan. This keeps the base plan predictable while capturing upside from hiring spikes (campus drives, mass recruitment rounds). White-Label Package For enterprise clients who want to deploy Assess.Plus under their own employer brand — or for HR consulting firms running hiring on behalf of corporates. Priced at a premium. Pilot-to-Convert Layer A 15-day pilot with 2 JD slots and 50 attempts, structured as a risk-free entry point for first-time buyers. The pilot is designed to generate a live verdict before purchase commitment — the product proves itself before money changes hands. Note: The talent intelligence layer, queryable historical score data across all assessed candidates creates long-term data value that increases switching costs and supports future product monetization. |
Mercer Mettl, PMaps, iMocha, Xobin, Talview. Assess.Plus wins against all of them: Every competitor above either sources from a static question library, applies a JD filter on top of pre-built tests, or delivers the same generic psychometric questionnaire regardless of role. Assess.Plus is architecturally different; it generates each assessment fresh from the JD itself, and psychometric traits are calibrated specifically to what predicts performance in that exact role. That is not a feature gap that incumbents can close by retraining on more data. Their foundation is a question library. Adding intelligence on top of that foundation does not change the foundation. Assess.Plus was built from the opposite direction. and that inversion is the moat. |
Direct Outreach + Network Partnerships: The primary acquisition channel today is structured direct outreach targeting Heads of Talent Acquisition and CHROs in the priority sectors. This is complemented by success-fee partnerships with individuals who carry deep networks inside target companies — paying commission on conversion rather than retainer. This keeps customer acquisition costs low and aligns partner incentives with closed deals rather than introductions. The pilot programme (15 days, 2 JDs, 50 attempts) functions as the conversion mechanism. The ask is not a budget commitment — it is 15 days and a live role. The product proves its verdict before purchase. |
Our GTM is vertical-first — BFSI, Business Services, Healthcare, and Manufacturing. These sectors share recurring mass recruitment with thin or no assessment infrastructure. Where tools exist, they serve only the largest enterprises. The mid-market majority runs on unstructured telephonic rounds — no contract to wait out, no switching cost to overcome. Our entry is adoption from zero, not displacement. Growth-stage technology companies, actively standardising their hiring for the first time, are a secondary market we address with the same motion. Every acquisition begins with a 15-day pilot — two live JDs, fifty candidate attempts, zero financial commitment. Not a demo. A live hiring outcome. A hiring manager experiences a ranked shortlist without a single screening call. That experience is our conversion engine, and reference customers accelerate adoption within each sector. We build sector density first, product depth second. As traction develops in one sector, expansion into the next follows on the back of reference credibility. The talent intelligence layer — making accumulated skill-score data queryable and predictive — is introduced after sector density, transforming Assess.Plus from a per-hire tool into an infrastructure investment that becomes harder to leave the longer it is used. Platform integrations and peer referrals within sectors are the natural growth mechanics that follow. |
The resume was invented in 1482. Every attempt to improve hiring since has been built on top of it, not instead of it. Assess.Plus is building what replaces it.
Our vision is to make the verified competency profile the new currency of the talent economy.
We pursue this in three horizons. In the near term, we replace telephonic screening for high-volume Indian enterprises — every JD generates a calibrated assessment, every candidate produces a verdict, hiring begins with proof rather than paper.
In the medium term, the platform extends from external screening to internal assessment. Companies that trust Assess.Plus to tell them who to hire naturally ask next: who should we promote, who is ready for a stretch role, where are our skill gaps? The same engine applies. Screening data and internal assessments together create a living competency graph per company — driving internal mobility, succession planning, and targeted learning investment.
In the long term, Assess.Plus sits on intelligence no other platform generates: real hiring behaviour at the competency level, across industries and roles. Which sectors struggle to find quality talent. Which roles are hardest to fill and why. Where companies are lowering the bar because supply isn't there. How hiring standards vary and shift over time. That intelligence is commercially valuable as workforce insights for enterprises, benchmarking products for companies, programme signals for EdTech platforms, and policy data for institutions tracking skill gaps. The platform that runs the assessments becomes, over time, the operating system of the talent economy. |
Incorporation in Progress |
NA |
Yes |
Avinya stood out for a specific reason: the combination of structured AI-focused mentorship and direct access to corporate pilot partnerships. Both matter to us right now. On the market access side, we need live deployments across BFSI, Healthcare, Manufacturing, and Business Services to generate the case evidence that converts the next wave of enterprise buyers. Avinya's pilot facilitation through corporate and institutional networks is a meaningful accelerant for that. On the mentorship side, the IIT alumni network, domain experts, and strategists give us structured access to people who have navigated the challenges we are about to face — scaling a product, building a sales motion, and stress-testing a business model. That kind of honest, experienced feedback is hard to find at this stage. Being embedded in Karnataka's AI ecosystem, the same geography where our early conversations are already happening, means the connections made during the programme have immediate relevance rather than being theoretical. We are applying because what Avinya offers aligns with where Assess.Plus is right now — early enough to benefit meaningfully from mentorship, and ready enough to make real use of the pilot and network opportunities the programme provides. |
Three concrete outcomes. First, use Avinya's corporate partnerships to get Assess.Plus into live deployments across at least two or three target sectors. The product is at a stage where real enterprise usage will surface what works, what needs refinement, and what the selling process actually looks like in practice. Pilots are the fastest way to find that out. Second, build the evidence layer from those deployments. Every live assessment run generates real data — shortlist quality, time saved, hiring team feedback. By Demo Day, we want to walk in with documented outcomes rather than projected ones. That evidence is what makes the next conversation with an enterprise buyer, or an investor, a different conversation. Third, stress-test the business fundamentals with the mentors and domain experts in the programme — unit economics, pricing structure, and how the business scales. We have conviction in the product direction, but there is a difference between building a product and building a business. The programme is an opportunity to pressure-test assumptions with people who have seen what breaks at the next stage, before we find out the hard way. If we exit the programme with paying customers, documented outcomes, and a clearer picture of the business, we will have used Avinya well. |
Hybrid |
Yes |
Core engine |
The core of Assess.Plus is mapIQ, a proprietary AI engine that takes a job description as input and generates a fully calibrated assessment as output. This is not a retrieval or filtering system. mapIQ runs a multi-stage pipeline: it first analyses the JD to extract role intent, seniority signals, domain context, and competency requirements. It then maps these across 65+ parameters spanning knowledge domains, cognitive complexity, and role-specific psychological traits.
The pipeline is built on several proprietary intelligence layers, each handling a specific stage: JD analysis, competency calibration, cognitive complexity mapping, psychometric trait identification, and answer quality enforcement. These layers are designed and calibrated by us, they define what the assessment must contain, how difficult it should be, which traits to assess, and what a quality response looks like for that specific role. An additional quality enforcement layer validates scientific integrity before any assessment is delivered, and overrides recruiter customizations that would compromise assessment validity.
The language generation at the final stage is handled through LLM APIs, orchestrated and constrained by our proprietary pipeline. The LLM does not decide what to assess, how to calibrate difficulty, or which psychometric traits are relevant; our intelligence layers do. The LLM executes the generation within those boundaries. Simultaneously, the pipeline constructs a structured evaluation framework calibrated to that same role, domain, and complexity level — so assessment generation and response evaluation are derived from the same intelligence layer, not built independently. This is where the defensibility sits: the intelligence that shapes and constrains the output is ours. Replacing the LLM at the generation layer does not change what the system produces because the proprietary layers upstream remain intact. |
As candidates attempt role-calibrated assessments, Assess.Plus accumulates structured competency response data across roles, industries, and seniority levels. This covers skills, knowledge depth, behavioural patterns, and psychometric trait distributions. This data does not exist in any public dataset and cannot be sourced or purchased externally.
The data works on two levels. At the product level, it makes the system progressively more intelligent. Response patterns across large candidate pools reveal what strong, average, and weak performance actually looks like at the competency level for each role type, which sharpens both assessment generation and scoring accuracy over time. At the market level, it builds benchmark hiring intelligence. Consistent patterns across companies and industries reveal where skill gaps are structural, which roles are hardest to fill, and how hiring standards vary across sectors. Because the assessments are role and domain-specific at the point of generation, the benchmark data that emerges is precise and actionable rather than generic. This benchmark intelligence layer has standalone commercial value beyond the core assessment product.
Every assessment run makes the system more accurate and more valuable. No competitor generating assessments from a static library accumulates data of this structure, depth, or commercial utility. |
Every competitor builds on a question library; AI is used to tag, filter, and retrieve from that library. Adding more intelligence on top of a library-based foundation does not change the fundamental constraint: the assessment is limited to what the library contains.
As described in the architecture, mapIQ derives each assessment from the JD itself through proprietary intelligence layers, with LLM APIs constrained entirely by what that pipeline has already determined. A competitor connecting an LLM directly to a question database is not building the same thing; they are automating retrieval. We are generating from intent.
The moat is the intelligence that sits upstream of the LLM: the calibration logic, the parameter mapping, the quality enforcement, and the competency framework built across 23 role buckets spanning Finance, Healthcare, Legal, Manufacturing, Creative, Education, Public Sector, Technology, and more, and the evaluation framework that is generated simultaneously with every assessment from the same intelligence layer. Competitors who automate retrieval from a question library have no equivalent — they cannot generate an evaluation framework because they never generated the assessment in the first place. Replicating this requires rebuilding the intelligence pipeline from scratch. Patent applications have been filed on the core methodology. |
Performance and reliability evaluation operates at two levels: the quality of what mapIQ generates, and the reliability of how candidate responses are scored.
On the generation side, the primary mechanism is automated conformance validation. The intelligence pipeline specifies precisely what the assessment must contain before generation happens — role intent, competency parameters, cognitive complexity level, and psychometric traits relevant to that specific role. Once the assessment is generated, automated checks run against these specifications to validate that the output faithfully reflects what the pipeline determined. Assessments that fall outside acceptable thresholds are flagged and regenerated before delivery — no human intervention required.
For every generated assessment, the system simultaneously constructs a structured evaluation framework calibrated to that role, domain, and complexity level. Candidate responses are then evaluated by an AI engine built on this framework, reading answers the way an expert examiner would — understanding meaning, rewarding depth, and catching flawed reasoning, instantly, consistently, and at any scale.
At the system level, candidate response distributions across a given assessment serve as a continuous performance signal. A well-constructed assessment produces a meaningful spread across performance levels. Distributions that cluster at either extreme trigger a review of the generation for that role, creating a feedback loop that catches issues at scale without waiting for human observation. |
Data privacy and security are treated as foundational.
Candidate data is accessible only to authorised personnel within the hiring company — no candidate results are shared across companies. Data is encrypted in transit and at rest. Platform access is secured through encrypted authentication, and backend systems operate under controlled and authorised access protocols. A privacy policy and terms of use are published on the platform.
As the product moves into commercial deployment, consent frameworks, data retention policies, and formal compliance structures are being established in parallel. |
The primary regulatory consideration is India's Digital Personal Data Protection Act, which governs how personal data — including candidate assessment responses — is collected, stored, and processed. Compliance structures are being built into the data pipeline as the product moves into commercial deployment.
Psychometric assessment in hiring contexts carries inherent responsibility around fairness, bias, and scientific validity. Assess.Plus addresses this at the foundation. Role-specific personality traits are mapped to the Big Five OCEAN framework — the Five Factor Model developed through decades of empirical research by Lewis Goldberg at the Oregon Research Institute and Paul Costa and Robert McCrae at the National Institutes of Health, and recognised globally as the scientific standard for personality measurement. Psychometric questions and evaluation frameworks are generated using the International Personality Item Pool, a public domain instrument also developed by Lewis Goldberg and managed by the Oregon Research Institute, with validity demonstrated across hundreds of international studies. Building on these established scientific standards means the psychometric output is not proprietary opinion — it is grounded in peer-reviewed, institutionally validated methodology, making it defensible to enterprise buyers, their legal teams, and regulators alike.
No sector-specific regulatory barrier currently restricts AI-based pre-employment assessment in India's target sectors, and India's regulatory environment is actively supportive of AI innovation in enterprise applications. |
Two areas require proactive attention at scale.
First, concurrency management. At 10x volume, every stage of the pipeline scales simultaneously — assessment generation, candidate attempts, and real-time evaluation all run in parallel across a large user base. Ensuring response time remains within acceptable thresholds at that scale requires distributed processing infrastructure, and this is being considered.
Second, LLM API dependency. At 10x volume, dependence on third-party LLM APIs introduces cost, rate limit, and reliability variables that are not acceptable in the long term. The mitigation is architectural: as assessment generation data accumulates across industries and role types, we fine-tune lightweight domain-specific language models trained on our own pipeline outputs to replace LLM API calls for generation. A Finance hiring model, a Healthcare hiring model, and a Manufacturing hiring model, each trained on real competency assessment data from our own pipeline, running on our own infrastructure. The intelligence layers upstream remain unchanged. The generation layer becomes proprietary, cost-efficient, and fully controlled. At that point, the fine-tuned models themselves become a standalone asset with value well beyond the core product. |
The founding team brings hands-on engineering capability in building and deploying AI pipelines, with direct experience in LLM API orchestration, prompt engineering, and intelligence layer design. For specialized requirements, the team works with contractual AI and NLP resources brought in as needed — a deliberate model that keeps the core architecture founder-driven while accessing deep technical expertise at the right stages.
As the product scales toward fine-tuned domain-specific language models, the team is actively building toward the AI and ML capability required for that stage, through ongoing self-development and identifying the right technical partners and hires. |
The datasets used are for psychometric assessment — the OCEAN framework and the International Personality Item Pool (IPIP), both public domain instruments validated across hundreds of international studies. We do not rely on scraped or unlicensed data sources. |
Improvement operates across three areas.
First, the intelligence pipeline is continuously refined through recruiter behaviour signals. When recruiters adjust, override, or customise generated assessments, those signals reveal where the pipeline output needs calibration for specific industries, roles, or company contexts. The quality enforcement layer ensures overrides that compromise scientific validity are not accepted, but valid adjustments become inputs for pipeline refinement.
Second, candidate response distributions across assessments serve as a continuous signal for evaluation framework quality. Distributions that fall outside expected ranges trigger a review of the generation pipeline.
Third, at scale, the roadmap moves towards self-owned domains and industry-specific language models. These models will be validated against the current LLM API output for quality, coverage, and calibration before deployment. |
Both, but India-first. The immediate focus is on high-volume hiring in India's BFSI, Healthcare, Manufacturing, and Business Services sectors, and growth-stage technology companies, industries, and organizations where assessment infrastructure is largely absent despite acute hiring pain. The architecture is not India-specific; the JD-to-assessment pipeline works across languages and geographies, but the go-to-market, competency calibration, and commercial focus are deliberately India-focused in this phase. Global expansion is a natural next step once the India model is proven. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/46xidGCp4Q3R-AssessPlus_Pitch_Avinya.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/46xidGCp4Q3R-AssessPlus_Pitch_Avinya.pdf |
https://drive.google.com/file/d/1VWQ32dwL0ujuRfnNMbxurvQQ4zEe11zE/view |
Yes. The mission is straightforward: hiring should begin with proof, not paper. But the impact runs deeper than the product.
Every year, millions of candidates in India apply for jobs and hear nothing back, no feedback, no indication of where they stand, no way to understand what they need to improve. The hiring process treats candidates as inputs to a filter, not as people deserving a fair evaluation. Assess.Plus changes that. Every candidate who attempts an assessment receives instant, personalized feedback on their performance, what they demonstrated, where they fell short, and how they compare against what the role requires. At scale, this is not a feature. It is a dignity signal delivered to every person who engages with the hiring process through our platform.
On the employer side, structured evidence-based hiring reduces the role of bias, gut feel, and network proximity in who gets shortlisted. A candidate from a tier-two city with no referral network gets the same fair evaluation as someone with every advantage. The shortlist is built on what they demonstrated, not who they know. |
Yes. One of the co-founders is a woman, bringing 25 years of experience in enterprise technology leadership — including a senior engineering leadership role at VMware — and currently serving as Karnataka State Vice President for the AI/ML stream at WICCI. |
Vivek Jha |
checked |
| Apr 30, 2026 @ 3:02 PM |
Sneha Kiran Manded |
snehamanded@gmail.com |
https://www.linkedin.com/in/sneha-manded-427378271/ |
https://medora.dev/ |
7411730876 |
Co Founder |
We met as classmates during our college studies, where we got to know each other through academic and project collaborations. Over time, we developed a strong understanding of each other’s skills and work styles. It has now been around one year that we have been working together closely on various projects. |
2 |
Our biggest strength as a team is our strong coordination and complementary skill sets. We understand each other’s working styles well, communicate effectively, and divide tasks based on our strengths. This allows us to work efficiently, solve problems faster, and consistently deliver high-quality results. |
Ayeeni Healthtech |
https://medora.dev/ |
Belagavi |
We are creating a multimodal AI system that decodes the full spectrum of healthcare data from symptoms to scans to speech into actionable medical intelligence. |
Healthcare today is fragmented, time-consuming, and difficult to access patients struggle to understand symptoms, doctors spend excessive time on documentation, and critical medical data (reports, prescriptions, scans) remains underutilized and disconnected. |
We are building a unified AI health intelligence system that replaces fragmented healthcare workflows with a single, intelligent layer.
What It Does
Understands Patients Instantly
Captures symptoms through an interactive interface and uses adaptive questioning to generate clinically relevant insights within seconds
Thinks Like a Clinical Assistant
Recommends the right specialists, tests, and next steps—guiding users toward faster and more accurate care
Reads and Interprets Medical Data
Processes prescriptions, lab reports, and medical imaging (X-rays, CT scans) into structured, usable information
Listens and Documents Automatically
Converts doctor–patient conversations into structured clinical reports, eliminating manual documentation |
Our solution goes beyond a single-feature AI it is a multimodal intelligence system that understands and reasons across symptoms, medical data, imaging, and clinical conversations within one unified platform, enabling a holistic view of the patient rather than fragmented outputs. We are not building just a feature, but a digital healthcare infrastructure that combines these capabilities into a continuous decision-support system. As the platform scales, it continuously learns from symptom patterns, doctor decisions, report interpretations, and real-world outcomes, creating powerful data network effects. Over time, this evolves into a proprietary healthcare intelligence graph that becomes increasingly difficult for competitors to replicate, forming a strong and defensible moat. |
Users |
Pilots |
We start with patients and clinics for adoption, and expand into hospitals as an integrated healthcare infrastructure layer. |
₹12–15 lakh crore |
₹1–1.5 lakh crore |
₹50 – 80 crore (initial 3–5 years) |
SaaS model for healthcare providers, a freemium model for patients, and transaction commissions on diagnostics and consultations |
Symptom Checkers : Ada Health Babylon Health , Telemedicine Platforms : Practo Tata 1mg, Medical Documentation & Scribing : AI Suki, AI Nuance Communications, We don’t compete with one category we replace the need for multiple disconnected tools. |
1. Doctor-Led Adoption (Primary Entry Point) ,2. Institutional Partnerships (Hospitals,Medical Colleges ) ,3. Patient Acquisition (Pull Strategy), 4. Hyperlocal Expansion (Local Clinics, Doctors ), 5. Referral & Network Effects |
Our GTM strategy is to start hyperlocal with a doctor-first approach, use clinics as distribution channels, drive patient adoption through a digital-first experience, and scale through partnerships and network effects. |
MEDORA is built on the belief that the future of healthcare is driven by intelligence, not just digitization.
Our vision is to create the intelligence layer of healthcare a unified AI system that continuously learns from and enhances every patient–doctor interaction across the care journey.
In 5 to 8 years we aim to evolve into a real-time healthcare decision engine that enables predictive care, augments clinical decisions, and connects fragmented healthcare systems.
Over time, MEDORA will become a global deeptech healthcare infrastructure, making intelligent, data-driven care accessible, personalized, and scalable for everyone. |
INCORPORATED |
NA |
Yes |
I am applying to the AVINYA cohort by ACSEL because MEDORA is at a critical stage where the right mentorship and ecosystem can significantly accelerate our path to scale. As a multimodal AI healthcare platform, we are focused on solving real-world gaps in accessibility and clinical intelligence, and AVINYA’s strong blend of technical depth, industry access, and founder-driven support aligns perfectly with our needs. I see this cohort as a strategic catalyst to refine our product-market fit, strengthen execution, and build a scalable, high-impact healthcare solution. |
During the AVINYA programme by ACSEL, my objective is to compress years of learning and execution into a few months of focused, high-quality iteration. I want to achieve three outcomes: Clinical-grade validation: Prove that MEDORA delivers measurable impact in real healthcare environments Production-ready AI system: Build a robust, scalable multimodal architecture capable of handling real-world variability Repeatable growth engine: Establish a clear GTM motion with early adopters, partnerships, and a path to scale By the end of the programme, I aim for MEDORA to move from a high-potential prototype to a validated, deployable, and investable deeptech healthcare company. |
Yes we are open for both hybrid and Virtual participation |
Yes |
Core engine |
Our platform is built as a distributed, multimodal AI system combining specialized models with a central reasoning engine.
1. Data Ingestion & Preprocessing Layer
Pipelines: Kafka / REST APIs for real-time + batch ingestion
Speech: Audio → processed via ASR models (fine-tuned Whisper/wav2vec for Indian accents)
Text: Tokenization, normalization, medical entity tagging
Images: Preprocessing (resize, denoise, DICOM handling for scans)
2. Modality-Specific Models
NLP Stack (Text Understanding)
Transformer-based models (e.g., fine-tuned BioBERT / ClinicalBERT)
Tasks:
Named Entity Recognition (NER)
Symptom & condition extraction
Relation mapping (symptom → disease → test)
Speech Intelligence
ASR → Whisper / wav2vec
Post-processing with clinical NLP for:
SOAP note generation
Context-aware summarization
Computer Vision Stack
CNN / Vision Transformers (ViT)
Tasks:
X-ray / CT scan classification
Report parsing (OCR + layout understanding)
Handwriting recognition
3. Multimodal Fusion Layer (Core Engine)
Combines embeddings from:
Text
Speech
Images
Techniques:
Cross-modal attention
Embedding alignment
Late + hybrid fusion strategies
4. Clinical Reasoning Engine
Built using:
Knowledge Graphs (symptom–disease–treatment relationships)
Probabilistic models (Bayesian inference for likelihood estimation)
LLM-based reasoning (fine-tuned for medical context)
Generates:
Differential conditions (assistive)
Next-step recommendations
Confidence scores
5. Learning & Feedback Loop
Human-in-the-loop:
Doctor corrections
Outcome validation
Techniques:
Continuous fine-tuning (LoRA / PEFT)
Reinforcement Learning from Feedback (RLHF-like loop)
6. Infrastructure & Deployment
Backend: Microservices architecture (Docker + Kubernetes)
Model Serving: TorchServe / FastAPI
Storage:
Structured → PostgreSQL
Unstructured → Object storage
Embeddings → Vector DB (FAISS / Pinecone)
Security & Compliance:
HIPAA-like standards
Encryption (data at rest + in transit) |
Proprietary Data Advantage
Currently, we are in the early stage, so we do not rely on large proprietary datasets yet. However, our core advantage lies in how we capture and compound data from day one.
We are building a system that continuously generates high-quality, structured clinical data from real-world interactions, including:
Symptom patterns and user inputs
Doctor–patient conversations converted into structured reports
Medical reports, prescriptions, and imaging interpretations
Doctor feedback, corrections, and outcomes
This data is not just collected—it is contextualized and connected across modalities, forming a longitudinal patient and clinical intelligence graph.
Over time, this creates a powerful data network effect, where:
More usage → better insights
Better insights → higher adoption
Higher adoption → richer proprietary data
This compounding loop builds a defensible data moat that is difficult for competitors to replicate, especially due to our deep integration into clinical workflows. |
Our AI is defensible because it is a multimodal, continuously learning system embedded directly into real clinical workflows, not a standalone model.
It integrates symptoms, medical data, imaging, and doctor–patient conversations into a unified patient context, enabling deeper and more accurate understanding than fragmented point solutions.
As the platform scales, it builds proprietary data flywheels from real-world interactions continuously improving accuracy and personalization.
1. Multimodal Intelligence & Unified Context
Integrates symptoms, reports, imaging, and conversations into a single system for holistic patient understanding
2. Proprietary Data Flywheel
Continuously learns from real-world usage, outcomes, and doctor feedback to improve over time
3. Workflow Integration & Reasoning Advantage
Embedded in clinical workflows with context-aware reasoning, creating high switching costs and deeper defensibility
This combination of multimodal fusion, continuous learning from proprietary data, and deep workflow integration ensures that the system becomes more accurate, more valuable, and increasingly difficult for competitors to replicate as it scales. |
We evaluate our AI using a multi-layered evaluation framework, since the system operates across text, speech, images, and clinical reasoning.
1. NLP & Clinical Understanding (Text + Reports)
Accuracy / F1 Score → for symptom & entity extraction
Precision & Recall → ensures correctness vs completeness
Clinical Concept Accuracy → mapping symptoms to correct medical entities
2. Speech-to-Text & Documentation
Word Error Rate (WER) → transcription accuracy
Medical Term Accuracy → correctness of clinical vocabulary
Summarization Quality → BLEU / ROUGE scores + human evaluation
3. Medical Imaging (X-rays / CT scans)
AUC-ROC → diagnostic classification performance
Sensitivity & Specificity → critical for medical reliability
False Positive / False Negative Rates → risk assessment
4. Decision Support & Reasoning Layer
Top-K Accuracy → whether correct condition appears in top suggestions
Clinical Validation Score → evaluated with doctors
Explainability Checks → traceable reasoning paths
5. End-to-End System Metrics (Most Important)
Time Saved per Consultation
Doctor Adoption & Retention
Correction Rate (Human-in-the-loop feedback)
Outcome Alignment (AI suggestion vs final diagnosis trend)
6. Reliability & Safety Measures
Confidence scoring on outputs
Human-in-the-loop validation for critical decisions
Continuous monitoring & model retraining |
We design our platform with privacy-by-design and security-by-default principles, ensuring patient data is protected at every stage.
1. Data Privacy (User Control & Minimization)
Collect only necessary medical data (data minimization)
Explicit user consent for data usage
Anonymization/pseudonymization for training and analytics
Role-based access (patients, doctors, admins)
2. Security Architecture
End-to-end encryption (data in transit via TLS, at rest via AES-256)
Secure authentication (OAuth / multi-factor authentication)
Regular security audits and vulnerability testing
Isolated environments for sensitive medical data
3. Compliance Readiness
Designed to align with global standards like HIPAA and GDPR
Prepared for India’s Digital Personal Data Protection Act 2023
Audit logs and traceability for all data access
4. AI-Specific Safeguards
Human-in-the-loop for critical decisions
Model outputs with confidence scores
Continuous monitoring to detect errors or bias |
1. Medical Device Classification
If our system is considered a Software-as-a-Medical-Device (SaMD), it may require approval from regulators like Central Drugs Standard Control Organization.
Mitigation: We position the product as clinical decision support (assistive, not diagnostic) in early stages and design with regulatory pathways in mind.
2. Data Privacy & Protection Laws
Handling sensitive health data must comply with regulations like Digital Personal Data Protection Act 2023.
Mitigation: Privacy-by-design, user consent, encryption, and strict access controls.
3. Clinical Liability Risk
Incorrect AI suggestions could lead to medical risk or liability concerns.
Mitigation:
Human-in-the-loop (doctor oversight)
Confidence scores + explainability
Clear disclaimers (decision support, not replacement)
4. Data Localization & Storage
Regulations may require health data to be stored within India.
Mitigation: Use India-based secure cloud infrastructure and compliant storage policies.
5. Bias & Model Reliability Regulations
AI systems in healthcare are increasingly scrutinized for bias and safety.
Mitigation:
Continuous validation with doctors
Diverse dataset training
Monitoring + audit logs |
1. Model Latency & Compute Bottlenecks
With 10x users, real-time processing (speech, imaging, NLP) can slow down.
Mitigation: Model optimization (quantization, distillation), GPU scaling, and async pipelines
2. Multimodal Pipeline Complexity
Coordinating text, speech, and image models at scale can create failures or inconsistencies.
Mitigation: Modular microservices architecture + robust orchestration + fallback mechanisms
3. Data Quality & Noise Explosion
More users = more noisy, inconsistent, and low-quality inputs
Mitigation: Input validation, data filtering, and human-in-the-loop correction systems
4. Model Drift & Reliability Issues
Real-world data changes over time → model performance can degrade
Mitigation: Continuous monitoring, retraining pipelines, and feedback loops
5. Infrastructure & Cost Scaling
Compute and storage costs can grow rapidly with multimodal data
Mitigation: Efficient model serving, tiered storage, and cost-optimized inference
6. Security & Compliance Risks at Scale
More users = higher risk of breaches and regulatory exposure
Mitigation: Strong encryption, access control, audit logs, and compliance frameworks |
We are a deeptech-driven founding team building at the intersection of AI and healthcare, with strong execution capability and proven innovation experience.
Urooj Akhtar Khan — Founder & CEO
Leads AI vision, product strategy, and system architecture
Background in robotics, multimodal AI systems, and healthcare innovation
National-level hackathon winner and leading initiatives in innovation ecosystems
Shadab Khan — AI/ML Engineer
Focused on building and scaling AI systems
Experience in backend architecture, model deployment, and performance optimization
Works on integrating multimodal models into production systems
Sneha Manded — ML Engineer
Specializes in data pipelines, model training, and evaluation
Contributes to improving model accuracy, robustness, and validation workflows
Deepak K T — Technical Advisor
Provides guidance on scalable architecture and advanced AI systems
Supports critical technical decisions and system design
4 Interns
A growing team working on:
Data annotation & preprocessing
Model experimentation
Feature development |
Currently, we are in the early stage and rely on a combination of publicly available, and synthetic datasets to build and validate our models.
1. Public & Research Datasets (Initial Training & Benchmarking)
We use well-known open datasets for model development, such as:
MIMIC-III / MIMIC-IV → clinical notes & structured data
NIH Chest X-ray Dataset → imaging models
PubMed → medical text understanding
2. Synthetic & Augmented Data
Generated datasets for:
Symptom flows
Clinical conversations
Data augmentation for robustness
3. Early Proprietary Data (Growing Asset)
As we onboard users, we start collecting:
Anonymized symptom interactions
Doctor–patient conversation transcripts
Structured reports and feedback |
We improve model performance through a closed-loop learning system that combines real-world data, feedback, and continuous retraining.
1. Human-in-the-Loop Feedback
Doctors can correct outputs (reports, suggestions)
Corrections are captured as high-quality labeled data
Directly improves model accuracy over time
2. Real-World Data Flywheel
Continuous inflow of:
Symptom interactions
Clinical decisions
Reports and outcomes
Builds a proprietary dataset that compounds with usage
3. Continuous Training Pipeline
Periodic model retraining using new data
Techniques:
Fine-tuning (LoRA / PEFT)
Domain adaptation
Keeps models aligned with real-world scenarios
4. Active Learning & Error Sampling
Identify low-confidence or incorrect predictions
Prioritize them for review and retraining
Improves performance where it matters most
5. Model Monitoring & Drift Detection
Track performance metrics in production
Detect data drift and performance drops
Trigger retraining when needed
6. Multimodal Optimization
Improve fusion between text, speech, and image models
Refine cross-modal reasoning
Enhances overall system intelligence |
We are building for India first, with a clear path to global expansion.
In India (Primary Focus – Phase 1)
Large, underserved population
Fragmented healthcare system → high need for integration
Strong fit for our AI + accessibility-driven approach
Allows us to validate, iterate, and build proprietary data
Global (Expansion – Phase 2)
Expand to markets with similar challenges:
Southeast Asia
Middle East
Africa
Eventually target developed markets with advanced integrations
Leverage our multimodal AI + data advantage at scale |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/48FixxEKX8xL-MEDORA-PPT.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/48FixxEKX8xL-MEDORA-PPT.pdf |
https://drive.google.com/drive/folders/1A2sH_0ynJtw2WBpKJXqFhYdx8Eq7SpWy?usp=drive_link |
Yes
We are building a mission-driven, impact-focused startup aimed at making healthcare more accessible, intelligent, and efficient.
Our mission is to reduce the gap between patients and quality healthcare by using AI to simplify how people understand symptoms, access care, and interact with medical systems. Today, healthcare is fragmented patients struggle to navigate it, and doctors spend significant time on manual processes instead of care. We aim to solve this by creating an intelligent layer that connects patients, doctors, and medical data into a seamless system.
Our impact is centered on:
Accessibility → enabling early health insights even in underserved and semi-urban regions
Efficiency → reducing doctor workload through automation of documentation and data interpretation
Quality of Care → supporting better clinical decisions through structured, data-driven insights
In the long term, we aim to shift healthcare from reactive to proactive, enabling earlier detection, better outcomes, and more personalized care at scale. |
We are Incubated in VTU's VRIF |
Cipherion Pvt Ltd |
checked |
| Apr 30, 2026 @ 1:56 PM |
Vaishnavi G, Jayanth Babu C |
hello@nirmaan.education |
https://www.linkedin.com/in/vaishnavi-g-krishnan-profile/ |
http://NA |
8431234711 |
Vaishnavi Gopal, Co-founder, leads Strategy, Curriculum, and Education. Former founder of 12Thirty, UPSC educator, and Founders Office at Leumas. She brings deep insight into how students learn and what institutions need.Jayanth Babu (Jay), Co-founder, leads Growth, Tech, and Finance. Former investment banker at Dexter Capital, MBA from IIM Lucknow. He brings financial rigor and operational execution to Nirmaan. |
Vaishnavi and Jayanth met at Alliance University, Bengaluru, where both studied Electrical and Electronics Engineering. What started as a shared frustration with how broken the education system was around them grew into a shared mission.
They have worked together for over 3 years, first validating the problem on the ground through offline learning centers, teaching 900 students across 3 centers before concluding that the only way to truly scale personalized learning was through AI.
Nirmaan is not a startup they decided to build. It is a problem they lived inside long enough to know exactly how to solve it. |
one |
Most edtech startups are built by technologists who have never sat inside a struggling classroom or educators who have never shipped a product.
We have done both.
Jay sees the business architecture, the unit economics, the fundraising, and the scale. Vaishnavi sees the child, the teacher, the gap, the curriculum. Together we see the full picture, which is rare and which is exactly what this problem demands.
Our biggest strength is that we did not start with a product. We started with a classroom. Three learning centers. 900 students. Years of watching real children fall through the cracks before we wrote a single line of code.
That means our AI is not built on assumptions. It is built on evidence.
We argue fast, align faster, and move without ego. One of us always knows something the other doesn't. That tension is not a weakness. It is the engine. |
Nirmaan |
https://nirmaan.education/ |
Bengaluru |
Nirmaan is an AI-native education stack that profiles each student's learning DNA, gaps, behaviroal patterns and teaches through multimodal AI (voice conversations, notes, visuals) and adapts continuously, delivering a human like 1-to-1 tutor experience at scale. |
Nirmaan solves the problem of lack of personalized learning in classrooms where one teacher often manages 30-40+ students, making individual attention impossible. This creates major learning gaps, as students who miss foundational skills like reading, basic math, and concept understanding early on are unable to progress further. A Grade 5 student may still struggle to read a Grade 3 text, which affects their entire learning journey. India has 25 crore school students where traditional education is still one-size-fits-all, exam-driven, and focused on rote memorization rather than true understanding. Students are taught the same way regardless of their pace, learning style, or confidence level. This leads to poor academic outcomes, low confidence, weak communication skills, and long-term employability challenges. We are solving this through personalized (1 tutor 1 student), multimodal AI learning that ensures every child learns at the right level |
We solve the problem of one-size-fits-all education by using a hyper-personalized, multimodal AI education stackWe are an AI-powered personalized learning platform that acts as an intelligent tutor for every student — diagnosing exactly where they are, adapting to how they learn, and teaching in a way that actually sticks.
Through our SIM (Student Intelligence Model), we continuously build learner profiles using voice, text, assessments, and behavioral signals. Our Whiteboard AI then teaches through conversational explanations, illustrations, and step-by-step problem solving, just like a teacher on a board.
For students, we replace the fear of falling behind with the confidence of moving forward ,at their own pace, in their own style.
For schools, we give teachers something they've never had — real visibility into every student's learning gaps, without adding to their workload.
This ensures Teaching at the Right Level (TaRL), |
- While every other platform delivers content , video lectures, practice questions, recorded lessons . Nirmaan builds a living intelligence model of each student. We call it the Student Intelligence Model (SIM). It doesn't just track what a child answered. It maps why they struggle, where the foundational gap actually is, and how they learn best. - Our second USP is multimodal teaching. Nirmaan doesn't communicate through text alone, it teaches through voice, visuals, and interactive conversation, delivering a genuinely human-like learning experience. The child isn't reading instructions. They're having a dialogue with a tutor. - Our third USP is edge computing. Nirmaan's core models run directly on-device - no internet dependency, no cloud LLM costs. A child in a low-connectivity village gets the same AI tutoring experience as one in urban Bengaluru. This is what makes true scale in India possible and affordable. |
two |
two, Pilots |
Nirmaan targets students aged 10–23 from aspirational middle- and lower-middle-class families (₹5–25L income, mobile-first) needing affordable personalized learning, along with budget to mid-tier schools and colleges (₹35K–₹1.2L fees) lacking diagnostic infrastructure—starting in Bengaluru and scaling across Tier 1 and 2 cities in India. |
₹52,177 Cr |
₹12,753 Cr |
NA |
D2C: Direct to Students and Parents Students and families subscribe monthly or annually. Pricing is tiered by access level: Basic: ₹299/month Standard: ₹599/month Premium: ₹1,499/month Freemium entry ensures low friction. Conversion happens once the child's personalized learning gap report is generated. B2B: Schools and Colleges Institutions pay an annual per-student license fee: Affordable private schools: ₹1,500/student/year Mid-fee schools: ₹1,800/student/year Premium schools: ₹2,000+/student/year |
Our main competitors are generic LLMs like ChatGPT and specialized AI tutors like Studdy.ai, Tegore, and Khanmigo. These tools answer questions well but none of them truly know the student. They respond. Nirmaan understands, maps, and adapts continuously to each child's unique learning journey. |
D2C: Through freemium diagnostics that hook parents the moment they see their child's learning gaps visualized. Hyperlocal WhatsApp community marketing, school gate activations, tuition center partnerships, and peer referral loops built into the product. B2B: Through direct school outreach, outcome-led demos, and word of mouth between school principals. One school contract onboards hundreds of students instantly. |
We start D2C, acquiring students and parents directly through freemium diagnostics, hyperlocal community marketing, and peer referrals. As our user base grows, we walk into schools with proven outcome data and convert them into B2B partners, making schools our distribution engine. Bengaluru first, then Tier 1 and Tier 2 cities across India. |
Our vision is an India where learning is so personalized, so engaging, and so effective that students don't just pass exams. They build real understanding, discover real strengths, and grow into careers that actually move the nation forward.
Education is the single largest lever for GDP growth. A generation that truly learns, thinks, and leads does not just find jobs. It builds industries, solves problems, and compounds national progress.
Nirmaan is not building another EdTech app. We are building the foundation for that generation. |
LLP |
NA |
one |
Nirmaan is at the stage where the product is built, the problem is validated, and the next move is scale. What we need now is structured mentorship, real pilot environments inside schools, and a network that accelerates what cold outreach cannot. ACSEL's focus on responsible, impactful AI in education aligns exactly with what Nirmaan stands for. Karnataka is our home market and being part of Karnataka's premier AI ecosystem, with IIT alumni networks and government connections, gives us the right foundation to go from Bengaluru to Bharat. We are applying because Avinya backs founders solving real problems. That is precisely who we are. |
Three things. First, close our first 10 school partnerships in Bengaluru and generate enough outcome data to make every subsequent school an easy decision. Second, refine our D2C acquisition engine through real user feedback, converting freemium diagnostics into paying subscribers at a repeatable cost. Third, walk into our next fundraise with proof. Not just a product and a vision, but numbers, pilots, and a network that validates both. Avinya's mentorship, pilot infrastructure, and investor access are exactly what we need to compress 12 months of learning into 4. |
Yes |
one |
one |
Student Intelligence Model (SIM): using competence-based knowledge graphs, maps learning gaps, pace, and style per student
Whiteboard AI: multimodal teaching engine via voice, visuals, conversation
Runs on-device via ONNX Runtime, INT8 quantization, XNNPACK acceleration
No cloud dependency. |
Behavioral interaction data from 900+ students across offline learning centers
Every session generates labeled learning gap data no public dataset captures
Proprietary student response patterns across Indian curriculum and regional languages
Data flywheel: more students = smarter SIM |
On-device architecture cannot be replicated by cloud-first competitors at our price point
SIM's behavioral data moat compounds with every user
Built for Indian curriculum, languages, and low-bandwidth reality
Content can be copied. Infrastructure cannot. |
Learning outcome improvement per student per session
Diagnostic accuracy: gap identified vs. gap confirmed by assessment
Session completion and re-engagement rates
Teacher-reported visibility scores in B2B deployments |
All student data processed on-device, minimizing cloud exposure
No personally identifiable data stored on external servers
Compliant with India's Digital Personal Data Protection Act 2023
Parental consent flows built into onboarding |
DPDP Act 2023 compliance for child data is closely monitored
NEP 2020 alignment reduces policy risk
EdTech regulation in India is evolving but directionally favorable
No known hard blockers at current stage |
On-device model synchronization across diverse Android hardware
SIM personalization latency at high concurrent users
Data pipeline for aggregating behavioral signals at scale
Mitigation: federated learning architecture planned for Phase 2 |
Krishnendu Dasgupta, Head of AI: Independent researcher, UN vetted AI engineer, MIT Executive Education, 2x founder
Core model development handled in-house |
Proprietary: behavioral and interaction data from offline learning centers
and some open-source models like Whisper and Kokoro (open-source MIT license)
NCERT curriculum data: publicly available
All third-party data used within licensed and public domain terms |
Every student session feeds labeled data back into SIM
Regular model retraining cycles based on outcome metrics
Teacher and student feedback loops built into the product
Phase 2: federated learning for privacy-preserving improvement at scale |
India first, always
Built for Indian curriculum, languages, infrastructure, and price points
Global expansion in Phase 3 targeting similar emerging markets
Southeast Asia, Africa, and Latin America as natural next markets |
|
https://drive.google.com/file/d/1LEWcuyX-ytJ4vr37zdDxOemFkl9Mps8o/view?usp=sharing |
Nirmaan was not started to build a product. It was started because we spent years watching children fall through the cracks of a system not designed for them.
Before writing a single line of code, we ran 3 learning centers and taught 900 students ourselves. We saw firsthand what a child looks like when they finally understand something they have been pretending to understand for years. That expression is why Nirmaan exists.
India will not become a developed nation by 2047 by producing exam-passing graduates. It will get there by producing people who truly understand, who can think, communicate, and lead. Nirmaan is built to create that generation.
Every product decision we make runs through one filter: does this actually help the child learn better? Not does it look good in a demo. Not does it impress an investor. Does it help the child.
That is mission. Everything else is execution. |
NA |
NA |
checked |
| Apr 30, 2026 @ 10:10 AM |
Ganesh Bhat |
hyperbharatgroup@gmail.com |
https://www.linkedin.com/in/innovatoraaryan/ |
https://www.wamimedia.com/ |
08970777835 |
Founder |
We both worked in our previous Company in udayavani |
2 |
Domain knowledge , technical expertise |
Dzentral Media Private Limited |
https://www.wamimedia.com/ |
Udupi |
WAMI is a district-first, language-first digital platform connecting citizens, businesses, and government through local news, public information, and economic opportunities. |
India has decentralized governance across 776 districts, but local information remains fragmented. Citizens struggle to access reliable district news, government contacts, schemes, jobs, and business opportunities in their own language. |
WAMI builds a district-first digital ecosystem that delivers verified local news, events, government directories, schemes, jobs, and business visibility in regional languages through one unified app. |
Our moat is a district-by-district network effect: localized content + vernacular brands + government utility + SME monetization. We are building digital infrastructure for Bharat where governance actually happens, creating a scalable model across all 776 districts. |
Users |
Users |
Citizens seeking local news, events, and government information SMEs, freelancers, and local businesses seeking visibility and customers Employers and job seekers through Wami Work Government departments for awareness campaigns and public communication |
₹63,000+ Cr |
₹6,000+ Cr |
₹236+ Cr |
Local business ads and sponsored promotions Premium listings and subscriptions Job postings and hiring services (Wami Work) Government awareness campaigns and notices Event promotions and ticketing commissions Future cross-sell from Homes, Market, Learn, TV and other verticals |
Dailyhunt, Public App, Inshorts, local newspaper groups, Justdial, Apna, WorkIndia, Facebook local groups, fragmented govt portals. |
District-first launch model using hyperlocal content, referrals, WhatsApp sharing, influencer partnerships, campus/community outreach, business onboarding, and vernacular social media marketing. |
Launch in Udupi & Dakshina Kannada, prove engagement and monetization, then replicate state-by-state using local language brands and a repeatable district expansion playbook. |
To build India’s district operating system — a unified digital infrastructure connecting citizens, businesses, and government across all 776 districts, scalable globally. |
20 March 2025 |
NA |
Yes |
We are applying to Avinya to accelerate WAMI’s AI-driven district-first platform through expert mentorship, product guidance, strategic network access, and ecosystem support. As we build AI-powered multilingual news, government utility, and local commerce solutions, Avinya can help us scale faster from Karnataka to national level. We also seek access to grants and non-dilutive support that can help strengthen product development, AI capabilities, and early market expansion. |
Strengthen our AI roadmap (AI anchors, personalization, recommendations, automation) Refine product-market fit and growth strategy Build investor readiness and fundraising strategy Secure grants or ecosystem support for growth and innovation Expand from Udupi & Dakshina Kannada to more districts Connect with mentors, partners, and the Karnataka startup ecosystem |
Yes, we are fully open to hybrid, in-person, and virtual participation and committed to actively engaging throughout the programme. |
Yes |
Supporting feature |
We use AI APIs and cloud models for content summarization, multilingual translation, recommendations, search assistance, moderation, and AI news anchor generation. Our app layer collects user signals and district preferences to personalize feeds and alerts. |
We are building district-level structured data across news, events, government contacts, services, schemes, and local business information. Over time, this creates a unique Bharat-focused dataset not easily available in one place. |
Our advantage is district-first localized data, multilingual usage patterns, and deep integration with governance + business + citizen needs. Competitors may have models, but not our local data network. |
We track translation quality, click-through rate, user retention, recommendation relevance, moderation accuracy, response time, and user feedback. We also manually review sensitive outputs. |
We follow privacy-by-design practices, secure cloud hosting, encrypted storage where required, limited data access, consent-based permissions, and clear privacy policies. Sensitive data is handled with minimum retention principles. |
Yes. News/content regulations, data privacy laws, intermediary compliance, AI content labeling, and advertising standards may evolve. We plan to stay compliant and adapt quickly. |
Inference cost, latency, moderation load, and content pipeline speed may become bottlenecks. We plan to solve this through caching, better infrastructure, model routing, and partial in-house systems. |
Currently we are a lean founding team using external AI tools and APIs. We plan to build an in-house AI/ML team focused on personalization, language models, and automation as we scale. |
We use publicly available data, internally created content, user interaction signals, and structured local datasets we are building ourselves. Third-party APIs are used under their respective terms. |
Through feedback loops, A/B testing, better prompts, fine-tuning where needed, expanding local datasets, and monitoring real user behavior district by district. |
We are building for India first, with a long-term vision to expand globally in other multilingual emerging markets. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/HHUpBEgxdO8X-WAMI_DzentralMediaPrivateLimited_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/HHUpBEgxdO8X-WAMI_DzentralMediaPrivateLimited_compressed.pdf |
https://youtube.com/shorts/5k6jJKSA3ZI?feature=share |
Yes. WAMI is a mission-driven startup focused on solving India’s grassroots information gap. While governance happens at the district and taluk level, reliable local information, public services access, and economic opportunities remain fragmented. We aim to bridge this gap through a district-first digital platform that connects citizens, businesses, and government in regional languages.
Our impact focus includes:
* Creating district-level jobs in media, tech, and sales
* Helping MSMEs and freelancers grow through digital visibility
* Improving access to schemes, officials, and public services
* Supporting 22 Indian languages and preserving local identities
* Promoting innovation, accountability, and informed communities
Our long-term vision is to build inclusive digital infrastructure for Bharat, starting from Karnataka and scaling across all 776 districts of India. |
NA |
Swasthika Devadiga | Founder & CEO | Calyirex Technologies Private Limited |
checked |
| Apr 30, 2026 @ 9:57 AM |
Anubhav Elhence |
kvgaitech@gmail.com |
https://www.linkedin.com/in/anubhavelhence/ |
https://kvgai.com/ |
7388889308 |
Founder |
I am the solo founder of kvgai tech |
1 |
Our biggest strength lies in strong ownership, fast execution, and a clear problem-solving mindset. Even as a lean team, we are highly focused, adaptable, and committed to building impactful AI-driven solutions. We move quickly from idea to implementation, continuously iterate based on feedback, and maintain a high level of accountability. Our ability to learn fast, stay resilient, and execute efficiently allows us to compete effectively despite limited resources |
Kvgai Tech |
https://kvgai.com/ |
Bengaluru |
KVGAI Tech is an AI-powered platform that automates evaluation, feedback, and intelligent insights for education and assessment systems. |
Traditional evaluation systems are slow, inconsistent, and highly manual, leading to delayed feedback, human bias, and inefficiencies for educators and institutions. This limits scalability and reduces the quality of learning outcomes, especially in large-scale or digital-first environments. |
KVGAI Tech provides an AI-driven evaluation engine that automatically analyzes responses, generates accurate results, and delivers instant, actionable feedback. Our platform integrates seamlessly with existing systems and supports scalable, real-time assessment workflows for institutions, educators, and organizations. |
Our core strength lies in our proprietary AI models optimized for evaluation accuracy, contextual understanding, and feedback generation. Unlike generic tools, KVGAI focuses on intelligent assessment workflows, combining automation with meaningful insights. Continuous learning loops, domain-specific tuning, and real-world testing create strong defensibility over time. |
Users |
Users, Revenue |
Our primary customers are educational institutions, coaching centers, edtech platforms, and organizations conducting large-scale assessments. We also target educators and enterprises that require automated evaluation and feedback systems. |
The global edtech and assessment market exceeds $300B+, with AI-driven evaluation rapidly growing as digital learning adoption increases. |
Our initial focus is the Indian edtech and institutional assessment market, estimated at $10B+, particularly targeting schools, colleges, and test-prep platforms adopting digital evaluation. |
In the next 3–5 years, we aim to capture $5M–$10M worth of this market by onboarding institutions and scaling through SaaS distribution. |
We operate on a SaaS-based model with tiered subscription plans for institutions and organizations. Additional revenue streams include enterprise integrations, API access, and usage-based pricing for large-scale evaluations. |
Our competitors include traditional assessment tools and emerging AI-based platforms such as Gradescope, Turnitin, and other edtech evaluation systems. However, most lack deep AI-driven feedback and real-time automation capabilities. |
We acquire customers through direct outreach to institutions, partnerships with edtech platforms, pilot programs, and leveraging founder-led networking and LinkedIn-based distribution. |
We follow a bottom-up approach starting with pilot programs in institutions, demonstrating value through improved efficiency and accuracy, and then scaling via referrals, partnerships, and SaaS adoption across similar organizations. |
To become the default AI infrastructure for evaluation and feedback globally, powering intelligent assessment systems across education and enterprise, and transforming how performance is measured and improved. |
Registered in India |
NA |
Yes |
We are applying to Avinya to accelerate KVGAI’s growth through structured mentorship, strategic guidance, and access to a strong ecosystem of founders, investors, and industry experts. The program aligns with our vision of building a scalable AI-driven evaluation platform, and we believe Avinya can help us refine our product, validate our market, and scale efficiently. |
During the program, we aim to strengthen our product-market fit, onboard pilot customers, and improve our AI models for higher accuracy and scalability. We also want to build strong partnerships, prepare for fundraising, and establish a clear go-to-market strategy to scale KVGAI effectively. |
YES |
Yes |
Core engine |
Our system uses a hybrid AI architecture combining transformer-based language models with task-specific fine-tuning for evaluation and feedback generation. The pipeline includes input preprocessing, semantic understanding, scoring logic, and feedback synthesis.
We leverage APIs like OpenAI/LLM providers along with custom prompt engineering and lightweight fine-tuned models. The backend is built with scalable cloud infrastructure and modular microservices for inference and evaluation workflows. |
We are building a growing dataset of evaluated responses, feedback patterns, and scoring benchmarks generated through real usage. Over time, this creates a proprietary feedback loop that improves model accuracy and domain-specific understanding, especially in evaluation contexts. |
Our defensibility comes from domain-specific optimization for evaluation workflows, continuous fine-tuning based on real user data, and feedback loops that improve accuracy over time. Unlike generic AI tools, we focus on structured evaluation + actionable feedback, creating higher switching costs as data and models improve. |
We evaluate performance using accuracy, consistency in scoring, response relevance, and qualitative feedback alignment. Additional metrics include error rate, latency, and user satisfaction from pilot users. Continuous benchmarking is done against expected outputs and human-evaluated samples. |
We follow best practices including secure data storage, encryption, and restricted access controls. Sensitive user data is anonymized where possible, and we ensure compliance with applicable data protection standards. We avoid storing unnecessary personally identifiable information and maintain strict data handling policies. |
Potential risks include data privacy regulations and AI governance policies. We mitigate this by minimizing sensitive data usage, ensuring transparency in AI outputs, and aligning with evolving compliance standards. |
At 10x scale, challenges may arise in inference latency, cost optimization, and maintaining consistent accuracy across diverse inputs. We are addressing this by optimizing model usage, introducing caching layers, and planning for fine-tuned smaller models to reduce dependency on heavy APIs. |
Currently led by the founder with strong hands-on experience in AI integration, prompt engineering, and system design. We plan to expand with dedicated ML engineers as we scale to strengthen model development and optimization. |
We primarily use publicly available datasets, synthetic data, and internally generated data from platform usage. As we grow, we are building proprietary datasets through real-world evaluation scenarios. |
We use a feedback loop approach where real user data and corrections are used to refine prompts, improve evaluation logic, and fine-tune models. Regular testing, iteration, and benchmarking ensure continuous improvement. |
We are building for both India and global markets, starting with India as our primary market and expanding internationally as the product matures. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/aUi7dMiMR73l-Smart-QnA-Tech-Pitch-Startup-India-1.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/aUi7dMiMR73l-Smart-QnA-Tech-Pitch-Startup-India-1.pptx |
https://smartpapercheck.com/ |
Yes. KVGAI Tech is focused on improving the quality, speed, and fairness of evaluation systems in education. By automating assessments and providing instant, unbiased feedback, we aim to reduce manual workload for educators and improve learning outcomes for students at scale. Our long-term mission is to make high-quality evaluation accessible, efficient, and data-driven across institutions globally |
NA |
NA |
checked |
| Apr 30, 2026 @ 1:08 AM |
Shruti Basu, Sudipto Basu |
shruti.basu@souloxy.com |
https://www.linkedin.com/in/shruti-basu/ |
https://www.souloxy.com |
9916772916 |
Shruti Basu – Co-founder & CPTO (Full-time) M.Tech from BITS Pilani, MBA, Diploma in Counselling, and ex-VMware. Shruti is the architect of Souloxy’s proprietary AI-powered behavioral intelligence system (patent pending). She leads product, AI, and engineering, builds the core technology in-house, writes core code, and manages the engineering team. Sudipto Basu – Co-founder & CEO (Full-time) BE from Jadavpur University, Executive Management from Indian Institute of Management Bangalore, ICF-certified Coach, and ex-Oracle. Sudipto leads institutional sales, partnerships, fundraising, pricing strategy, and overall business operations. He drives university and corporate relationships and is responsible for scaling Souloxy’s market presence. |
e are childhood friends turned life partners, and now business partners. Having known each other since childhood, we have grown together through different phases of life and have always shared deep trust, understanding, and complementary strengths. Over the years, our friendship evolved into partnership in life, and eventually into building Souloxy together.
We have been working together professionally since Souloxy was conceptualized, combining our distinct expertise to bring the vision to life. Sudipto brings experience in business strategy, partnerships, and go-to-market, while Shruti brings deep expertise in product, AI, and technology development. This strong personal foundation, combined with complementary professional skills, enables us to make decisions quickly, navigate challenges with resilience, and build with long-term alignment. |
2 |
Our biggest strength as a team is the combination of deep trust, complementary expertise, and a shared long-term vision. Having known each other since childhood and worked closely together since Souloxy was conceptualized, we have a strong personal and professional understanding of each other’s strengths and decision-making styles. This allows us to move fast, stay aligned, and navigate uncertainty with resilience.
Functionally, our skills are highly complementary—Sudipto leads business strategy, partnerships, fundraising, and go-to-market, while Shruti leads product, AI, engineering, and innovation. Together, we cover both the “build” and the “sell” sides of the business.
Beyond the founding team, our in-house psychologist and network of senior mental health experts add strong clinical and ethical depth, enabling us to build responsible and scalable AI-powered wellbeing infrastructure. |
Souloxy |
https://www.souloxy.com |
Bengaluru |
Souloxy is building AI-powered preventive mental wellbeing infrastructure for educational institutions and enterprises |
We are solving the growing mental health and wellbeing crisis among students and employees, where stress, anxiety, burnout, and disengagement often go unnoticed until they become severe. Existing support systems such as counseling and EAPs are largely reactive, underutilized, and difficult to scale, while there is a shortage of mental health professionals. Institutions lack real-time visibility into wellbeing risks and struggle to provide timely, personalized support. Souloxy addresses this gap by enabling early detection, preventive intervention, and institutional intelligence through AI-powered behavioral insights. |
Souloxy is an AI-powered preventive wellbeing infrastructure platform for educational institutions and enterprises. Our first layer is a structured resilience-building program designed to drive high engagement through guided reflection, self-awareness exercises, and wellbeing journeys embedded into existing institutional workflows. This creates continuous behavioral and engagement data that powers our AI engine.
Our AI then analyzes behavioral signals, interaction patterns, and sentiment to detect early signs of stress, burnout, or disengagement, and delivers personalized interventions such as contextual nudges, guided journeys, and early alerts. At the institutional level, Souloxy provides anonymized dashboards and predictive insights to identify cohort-level risks, track wellbeing trends, and enable timely action.
By combining engagement-led programs with AI-driven behavioral intelligence, Souloxy helps institutions shift from reactive mental health support to scalable, data-driven prevention. |
Souloxy is unique because we are not building just another wellness app or reactive chatbot—we are building AI-powered preventive wellbeing infrastructure. Unlike traditional counseling services and EAPs that are reactive, or generic AI mental health apps that rely on voluntary engagement, Souloxy is embedded into institutional workflows through structured resilience-building programs that drive high and continuous engagement. This allows us to capture longitudinal behavioral and engagement signals over time. Our proprietary AI engine uses this data to build individual behavioral baselines, detect early deviations and wellbeing risks, and deliver personalized interventions through a continuous Detect → Personalize → Act → Learn loop. At the institutional level, we provide anonymized cohort-level dashboards, predictive insights, and stressor intelligence, creating value not just for individuals, but also for institutions. Our defensibility comes from: 1. Proprietary behavioral intelligence engine with a patent-pending closed-loop behavioral state modeling and intervention optimization system. 2. Data moat—more engagement leads to better data, better models, better outcomes, and stronger retention. 3. Embedded distribution model via institutions, creating higher adoption than voluntary wellness apps. 4. Clinical and ethical rigor through in-house psychology expertise and senior mental health advisors. This combination of AI, behavioral data, institutional integration, and clinical expertise makes Souloxy difficult to replicate and highly scalable. |
Users |
Pilots |
Our primary target customers are higher education institutions—universities and colleges—looking to proactively improve student wellbeing, engagement, and retention. We work with institutional decision-makers such as Directors, Deans, Student Affairs teams, and counseling/wellbeing departments. Our secondary target customers are enterprises and corporates seeking scalable preventive wellbeing solutions to reduce employee stress, burnout, and disengagement while improving productivity and retention. The end users of our platform are students and employees who receive personalized wellbeing support and interventions. |
TAM (Total Addressable Market): ₹1,000+ Cr ARR Across higher education institutions and enterprises in India, with long-term expansion into global education, enterprise, and public sector markets. |
SAM (Serviceable Available Market): ₹170+ Cr ARR Focused 5-year India-first execution market: ₹84 Cr from 600 target higher education institutions + ₹90 Cr from 600 mid-to-large enterprises. |
SOM (Serviceable Obtainable Market): ₹8–15 Cr ARR Realistic near-term obtainable market through converting 50–100 institutions and initial enterprise clients over the next 3–5 years. |
Souloxy operates on a B2B SaaS subscription model. We charge higher education institutions an annual per-student subscription fee (₹500–₹1,000 per student/year), typically through multi-year contracts integrated into institutional systems. For enterprises and corporates, we offer an annual per-employee subscription model (starting at ~₹1,500 per employee/year) for preventive wellbeing programs and AI-powered engagement reinforcement. Our go-to-market follows a pilot-to-contract approach: starting with paid or strategic pilots, expanding to full cohort deployments, and converting into recurring multi-year subscriptions. Over time, as our AI-driven infrastructure scales, margins improve through automation and deeper institutional integration. |
Our competitors fall into three broad categories: Traditional counseling services and therapy centers – provide reactive, one-to-one support but are difficult to scale and often accessed only during crisis. Employee Assistance Program (EAP) providers and corporate wellness vendors – offer support programs but often see low engagement and limited personalization. Wellness and mental health apps / AI chat-based platforms – such as meditation apps, self-help apps, or AI companions, which rely on voluntary engagement and lack institutional integration or longitudinal behavioral intelligence. Souloxy differentiates itself by being embedded into institutional workflows, driving continuous engagement through structured resilience-building programs, and using AI-powered behavioral intelligence for early detection, personalized intervention, and institutional insights. |
We acquire customers primarily through a B2B institutional sales model. Our initial wedge is higher education institutions, where we leverage founder-led outreach, warm introductions through academic networks, LinkedIn outreach, and partnerships with professors, deans, and institutional decision-makers. Our go-to-market follows a pilot-led acquisition strategy: we start with department-level or cohort-based pilots, demonstrate engagement and outcome metrics, and then expand into institution-wide, multi-year contracts. We are also building credibility through thought leadership, workshops, and wellbeing programs with universities, which help generate inbound interest and referrals. For enterprises, we plan to acquire customers through direct outreach, HR/wellbeing partnerships, and strategic corporate pilots. |
Our go-to-market strategy is a pilot-led B2B expansion model across both higher education institutions and enterprises. We enter through cohort-based resilience-building and wellbeing programs that drive immediate engagement and generate measurable behavioral and outcome data. In higher education, we begin with department-level or cohort-based pilots and expand into institution-wide, multi-year subscription contracts. In enterprises, we start with focused preventive wellbeing programs or pilot deployments for specific teams/business units and scale into organization-wide contracts. Our initial distribution is founder-led through direct outreach, institutional and corporate networks, partnerships, LinkedIn outreach, and referrals. We are also building inbound demand through thought leadership, workshops, and wellbeing initiatives. This pilot-to-contract approach allows us to validate outcomes quickly, build case studies, and scale efficiently across both segments in parallel. |
Our long-term vision is to build the world’s most trusted AI-powered preventive wellbeing infrastructure, helping institutions proactively detect, understand, and address emotional wellbeing risks before they become crises.
We aim to transform mental health support from a reactive, one-to-one service model into a scalable, data-driven prevention layer embedded across universities, enterprises, healthcare systems, and public institutions.
Over time, Souloxy will evolve into a comprehensive behavioral intelligence platform, combining continuous signal capture, personalized interventions, and institutional insights to improve wellbeing, engagement, retention, and performance at scale.
Our mission is to impact millions of lives globally by making preventive mental wellbeing accessible, personalized, and integrated into everyday institutional systems. |
Private Limited Company |
NA |
Yes |
We are applying to Avinya because Souloxy is at an inflection point where the right mentorship, ecosystem access, and strategic support can significantly accelerate our growth. As an AI-powered behavioral intelligence platform, we believe Avinya’s focus on AI-first startups can help us strengthen our product, sharpen our GTM strategy, and scale responsibly. We are particularly excited about access to mentors, investors, and industry networks that can help us accelerate pilots, enterprise partnerships, and institutional adoption. Karnataka’s vibrant startup and technology ecosystem also offers strong opportunities for partnerships and expansion. We believe Souloxy aligns well with Avinya’s vision of supporting impactful AI startups, and we see this cohort as a valuable platform to refine our AI-driven preventive wellbeing infrastructure and scale our impact faster. |
During the programme, we want to achieve three key outcomes: Strengthen our AI and product roadmap by refining our behavioral intelligence engine, improving personalization, and scaling our closed-loop Detect → Personalize → Act → Learn system. Accelerate market expansion and conversions by securing more institutional and enterprise pilots, converting pilots into paid multi-year contracts, and sharpening our go-to-market strategy. Build strategic partnerships and fundraising readiness through access to mentors, investors, and ecosystem partners who can help us scale faster and responsibly. By the end of the programme, we aim to have stronger product-market validation, deeper AI capabilities, and a clearer path to scale across higher education and enterprise markets. |
Yes |
Yes |
Core engine |
AI Architecture – Souloxy
Souloxy is built as a cloud-native, AI-first behavioral intelligence platform designed to deliver real-time, personalized, and preventive mental wellbeing support at scale.
At a high level, the architecture follows a multi-layered, modular AI system combining data engineering, generative AI, and behavioral modeling:
1. Data Ingestion & Signal Layer
Ingests multi-modal data streams:
Structured: assessments, engagement metrics, session history
Unstructured: journaling, conversational inputs, feedback
Behavioral: usage patterns, interaction frequency, response latency
Streaming and batch pipelines process high-frequency, time-series data
2. NLP & Generative AI Layer
Built on transformer-based models and LLM APIs (OpenAI, Gemini, Claude, open-source)
Capabilities include:
Sentiment & emotion detection
Contextual understanding of user inputs
Safe, empathetic conversational responses
Uses prompt orchestration + memory frameworks (LangChain/LangGraph) for multi-turn reasoning and contextual continuity
3. Behavioral Intelligence Engine (Core Differentiator)
Fuses:
Emotional signals (what users express)
Behavioral patterns (how users interact)
Temporal context (how states evolve over time)
Outputs:
Dynamic risk scores
User segmentation
Personalized intervention pathways
Moves beyond static NLP → enables longitudinal behavioral modeling
4. Personalization & Recommendation Layer
Context-aware recommendation engine delivers:
Adaptive intervention journeys
Nudges, exercises, and content
Escalation triggers (AI → human mentor/counsellor)
Ensures closed-loop personalization based on engagement and outcomes
5. Closed-Loop Learning System
Continuously improves via:
User interactions
Feedback loops
Intervention effectiveness data
Supports model refinement, drift detection, and adaptive learning over time
6. Data & Memory Layer
Relational DB (PostgreSQL): structured data
Document stores (Supabase): unstructured content
Vector DB (PgVector): embeddings, semantic search, contextual memory
Enables context retention and personalized AI interactions
7. Backend & Infrastructure
Microservices-based, API-first architecture (Node.js, Python – FastAPI/Django)
Cloud: GCP, containerized via Docker, orchestrated with Kubernetes
CI/CD: GitHub Actions | Monitoring: Prometheus
Designed for elastic scalability across institutions
8. Safety, Security & Responsible AI Layer
Guardrailed AI responses to prevent harmful outputs
Human-in-the-loop escalation for high-risk cases
Encryption (at rest & in transit), RBAC, anonymization
Consent-driven data governance for sensitive wellbeing data
Architecture Summary (Key Strength)
Souloxy’s architecture is not just a conversational AI system—it is a closed-loop behavioral intelligence platform, where:
Data → Emotional & Behavioral Insights → Personalized Interventions → Outcomes → Continuous Learning
This enables the system to evolve from reactive support to predictive and preventive wellbeing intelligence, scalable across campuses and enterprises. |
Souloxy’s core proprietary advantage lies in building a longitudinal, multi-modal behavioral dataset focused on mental wellbeing—an area where high-quality, contextual data is extremely scarce and difficult to replicate.
Unlike generic AI systems that rely primarily on static or public datasets, Souloxy continuously captures three interconnected layers of signals:
Emotional signals: journaling, conversational interactions, qualitative feedback
Behavioral signals: app usage patterns, engagement frequency, response latency, drop-offs
Intervention outcomes: participation in programs, mentor sessions, and resulting wellbeing shifts
This creates a closed-loop dataset that links user state → intervention → outcome, enabling us to understand not just how users feel, but what actually works for whom and in which context.
Additionally:
Data is longitudinal, capturing how emotional and behavioral patterns evolve over time rather than point-in-time snapshots
The system aggregates cohort-level, anonymized institutional insights, creating a unique layer of population-scale wellbeing intelligence
Continuous interaction through AI + human touchpoints generates high-frequency, real-world data, improving model accuracy and personalization
Over time, this results in a defensible data moat, as:
Emotional-behavioral datasets are hard to source, sensitive, and context-dependent
The value compounds with scale through continuous feedback loops and intervention learning
This proprietary dataset enables Souloxy to move beyond generic recommendations toward context-aware, outcome-driven, and continuously improving wellbeing intelligence, creating a strong competitive advantage. |
Souloxy’s AI is defensible not because of a single model, but due to a compounding system of data, intelligence, and domain integration that is difficult to replicate.
1. Proprietary Longitudinal Behavioral Dataset
We are building a closed-loop, multi-modal dataset combining emotional signals (journaling, conversations), behavioral patterns (engagement, usage), and intervention outcomes.
This creates a unique understanding of what works, for whom, and in which context, forming a strong and growing data moat.
2. Behavioral Intelligence Engine (Core IP)
Unlike generic LLM-based chatbots, Souloxy integrates:
Emotional + behavioral + temporal signals
Longitudinal user profiles
Context-aware risk scoring and intervention mapping
This behavioral intelligence layer enables predictive and personalized support, which is significantly harder to replicate than standalone conversational AI.
3. Closed-Loop Learning System
Our system continuously learns from:
User interactions
Intervention effectiveness
Feedback loops
This creates a self-improving system, where performance and personalization increase over time—leading to compounding advantage with scale.
4. Human + AI Hybrid Model
We combine AI-driven insights with human mentor escalation and validation, especially in high-risk scenarios.
This hybrid model improves:
Data quality
Trust and safety
Outcome effectiveness
It also creates unique supervised learning signals competitors without human integration cannot easily access.
5. Domain-Specific Design (Mental Wellbeing)
Mental wellbeing is:
Highly contextual
Sensitive
Non-linear
Our system is built with domain-specific guardrails, frameworks, and intervention logic, making it far more specialized than generic AI tools.
6. Deep Institutional Integration
Souloxy integrates at the institution level (universities, enterprises), capturing:
Cohort-level trends
Program effectiveness
Organizational insights
This creates network effects + switching costs, strengthening long-term defensibility.
Summary
Souloxy’s defensibility comes from a compounding moat:
Proprietary data + Behavioral intelligence + Closed-loop learning + Human-AI integration + Domain depth
Together, this makes the system increasingly personalized, effective, and difficult to replicate over time. |
Souloxy evaluates AI performance through a holistic, outcome-driven framework that goes beyond traditional model accuracy metrics.
In a sensitive domain like mental wellbeing, accuracy alone is insufficient. We therefore assess our system across four critical dimensions:
Understanding Accuracy: How well the AI interprets emotional and contextual signals (sentiment, intent, behavioral patterns)
Safety & Reliability: Whether responses are safe, non-harmful, and correctly trigger escalation in high-risk scenarios
Real-World Effectiveness: Whether AI-driven interventions lead to measurable improvements in user wellbeing, engagement, and help-seeking behavior
System Robustness at Scale: Whether the platform performs reliably under real-world conditions (latency, uptime, consistency)
A key differentiator is our focus on closed-loop outcome validation—we don’t just measure predictions, but also track intervention effectiveness over time, linking AI recommendations to actual user outcomes.
Additionally, human-in-the-loop validation (experts reviewing outputs and edge cases) ensures that the system remains clinically appropriate, context-aware, and trustworthy.
Finally, continuous monitoring of model drift, bias, and performance degradation ensures that the system adapts as user behaviour and data evolve. |
Souloxy follows a privacy-by-design, security-first architecture, essential for handling sensitive mental wellbeing data. Our approach integrates consent-driven governance, strong encryption, and advanced data protection practices aligned with Indian and global standards.
1. Privacy-by-Design & User Consent
All data collection is explicitly consent-driven, with clear disclosure of purpose and usage
Data minimization principles ensure only necessary data is collected
Users retain control over their data, aligned with consent and usage policies
2. PHI / PII Data Protection & Masking
Sensitive data, including Personally Identifiable Information (PII) and Personal Health Information (PHI), is protected through:
Pseudonymization and anonymization for analytics and model training
Data masking/tokenization to prevent exposure of identifiable attributes
Separation of identity data from behavioral and emotional datasets
Institutional dashboards only access aggregated, non-identifiable insights
3. Data Encryption & Secure Storage
End-to-end encryption:
Data in transit secured via HTTPS/TLS
Data at rest encrypted using industry standards
Secure key management and storage practices
4. Access Control & Governance
Role-Based Access Control (RBAC) to restrict access based on user roles
Secure authentication (OAuth/JWT)
Audit logs and traceability for all critical system activities
5. Secure Infrastructure
Deployed on Google Cloud Platform (GCP) with enterprise-grade security
Containerized (Docker) and orchestrated (Kubernetes) architecture
Continuous monitoring, logging, and anomaly detection
6. Compliance & Regulatory Alignment
Aligned with India’s Digital Personal Data Protection Act, 2023 (DPDPA)
Incorporates global best practices inspired by General Data Protection Regulation (GDPR)
Ensures compliance with principles such as purpose limitation, consent, data protection, and user rights
7. AI Safety & Sensitive Data Handling
Guardrails to prevent unsafe or inappropriate AI outputs
Human-in-the-loop escalation for high-risk scenarios
Strict separation between AI processing layers and identifiable user data
Summary
Souloxy ensures privacy and security through a multi-layered approach combining PHI/PII protection, consent-driven governance, encryption, access control, and regulatory compliance (DPDPA-aligned).
This enables us to build a system that is not only secure and compliant, but also trusted for handling highly sensitive wellbeing data at scale. |
Souloxy follows a privacy-by-design, security-first architecture, essential for handling sensitive mental wellbeing data. Our approach integrates consent-driven governance, strong encryption, and advanced data protection practices aligned with Indian and global standards.
1. Privacy-by-Design & User Consent
All data collection is explicitly consent-driven, with clear disclosure of purpose and usage
Data minimization principles ensure only necessary data is collected
Users retain control over their data, aligned with consent and usage policies
2. PHI / PII Data Protection & Masking
Sensitive data, including Personally Identifiable Information (PII) and Personal Health Information (PHI), is protected through:
Pseudonymization and anonymization for analytics and model training
Data masking/tokenization to prevent exposure of identifiable attributes
Separation of identity data from behavioral and emotional datasets
Institutional dashboards only access aggregated, non-identifiable insights
3. Data Encryption & Secure Storage
End-to-end encryption:
Data in transit secured via HTTPS/TLS
Data at rest encrypted using industry standards
Secure key management and storage practices
4. Access Control & Governance
Role-Based Access Control (RBAC) to restrict access based on user roles
Secure authentication (OAuth/JWT)
Audit logs and traceability for all critical system activities
5. Secure Infrastructure
Deployed on Google Cloud Platform (GCP) with enterprise-grade security
Containerized (Docker) and orchestrated (Kubernetes) architecture
Continuous monitoring, logging, and anomaly detection
6. Compliance & Regulatory Alignment
Aligned with India’s Digital Personal Data Protection Act, 2023 (DPDPA)
Incorporates global best practices inspired by General Data Protection Regulation (GDPR)
Ensures compliance with principles such as purpose limitation, consent, data protection, and user rights
7. AI Safety & Sensitive Data Handling
Guardrails to prevent unsafe or inappropriate AI outputs
Human-in-the-loop escalation for high-risk scenarios
Strict separation between AI processing layers and identifiable user data
Summary
Souloxy ensures privacy and security through a multi-layered approach combining PHI/PII protection, consent-driven governance, encryption, access control, and regulatory compliance (DPDPA-aligned).
This enables us to build a system that is not only secure and compliant, but also trusted for handling highly sensitive wellbeing data at scale. |
At 10x scale, the key challenges are latency, personalization, safety workflows, and cost efficiency.
However, Souloxy is being designed as a modular, cloud-native, and continuously learning system, allowing us to proactively address these bottlenecks through:
Infrastructure scaling
Model optimization
Human-AI workflow design
This ensures that as we scale, the system becomes not just larger—but more efficient, reliable, and intelligent over time. |
Core AI/ML Expertise
Our team brings capabilities across the full AI lifecycle:
NLP & Generative AI:
Experience working with transformer-based models, LLM APIs (OpenAI, Gemini, open-source), prompt engineering, and conversational AI design
Applied Machine Learning:
Building models for sentiment analysis, emotion detection, behavioral pattern recognition, and predictive risk scoring
Behavioral Intelligence Modeling:
Designing systems that combine emotional, behavioral, and temporal data to generate personalized insights and interventions
LLM Orchestration & Systems Design:
Experience with frameworks like LangChain/LangGraph for multi-step reasoning, context management, and safe AI workflows |
What datasets are you currently using? Are they licensed or owned?
Souloxy uses a combination of proprietary (owned), publicly available, and model-provider datasets, designed to balance domain relevance, safety, and scalability.
1. Proprietary (Owned) Data – Core Advantage
Our primary dataset is first-party, consented user data generated through the platform, including:
Assessments and psychometric responses
Journaling and conversational interactions (AI + human-assisted)
Behavioral signals (engagement patterns, usage, response latency)
Intervention data (program participation, mentor sessions, outcomes)
This data is:
Owned by Souloxy (with user consent and privacy safeguards)
Continuously growing and longitudinal
Central to our personalization, risk modeling, and behavioral intelligence
2. Public & Research Datasets (Bootstrapping & Validation)
During early development and model benchmarking, we leverage publicly available datasets, such as:
Sentiment and emotion classification datasets
Mental health and psychological research datasets
Open conversational corpora for language understanding
These are used for:
Initial model training and benchmarking
Improving baseline NLP capabilities
Validating performance across standard tasks
(All datasets are used in compliance with their respective licenses.)
3. Pre-trained & Licensed Model Ecosystems
We build on top of pre-trained LLMs and NLP models (e.g., OpenAI, Gemini, open-source Hugging Face models), which are:
Licensed via API or open-source terms
Fine-tuned and orchestrated with our domain-specific logic and guardrails
4. Expert-Curated Frameworks (Internal IP)
We incorporate:
Psychometric frameworks
Intervention models
Domain-specific rules and safety guardrails
These are internally developed and curated, forming part of our proprietary intelligence layer.
Summary
Souloxy’s data strategy combines:
Owned, longitudinal user data (core moat)
Public datasets (for baseline capability)
Licensed models (for foundational intelligence)
Internal domain frameworks (for differentiation)
Over time, the system becomes increasingly driven by proprietary, real-world behavioral and outcome data, which strengthens both model performance and long-term defensibility. |
How do you plan to continuously improve model performance?
Souloxy follows a closed-loop learning and MLOps-driven approach to continuously improve model performance, ensuring the system becomes more accurate, personalized, and reliable over time.
1. Closed-Loop Learning from Real-World Usage
Continuously capture:
User interactions (conversations, journaling)
Behavioral signals (engagement, drop-offs, response patterns)
Intervention outcomes (what worked vs. not)
Link AI predictions → interventions → outcomes, enabling the system to learn effectiveness, not just accuracy
2. Feedback-Driven Model Refinement
Integrate feedback from:
Users (implicit + explicit signals)
Mentors/counsellors (expert validation)
Use this to:
Correct misclassifications
Improve recommendation quality
Refine risk detection models
3. Continuous Evaluation & Monitoring
Track:
Model accuracy (precision, recall, F1)
Safety metrics (hallucination rate, unsafe outputs)
False positives/negatives in risk detection
Implement drift detection to identify performance degradation across cohorts or time
4. Iterative Model Updates & Experimentation
Regular retraining using fresh, anonymized data
A/B testing of:
Prompts
Models
Intervention strategies
Version control and rollback mechanisms via MLOps pipelines
5. Hybrid Model Strategy
Combine:
Pre-trained LLMs (for language understanding)
Fine-tuned domain-specific models (for emotion, risk, personalization)
Gradually shift toward more efficient, domain-adapted models as proprietary data grows
6. Human-in-the-Loop Learning
Expert review of:
Edge cases
High-risk scenarios
Model errors
Use curated datasets from these reviews to improve model robustness and safety
7. Data Quality & Expansion
Improve dataset richness by:
Increasing user engagement touchpoints
Expanding behavioral and contextual signals
Ensure high-quality labeling through expert-guided annotation frameworks
Summary
Souloxy improves model performance through a continuous learning loop:
Data → Insights → Interventions → Outcomes → Feedback → Model Refinement
This ensures the system evolves from static predictions to a self-improving, outcome-driven intelligence layer, becoming more accurate, personalized, and effective as it scales. |
Souloxy is building for both India and global markets, with an India-first, globally scalable strategy. We are initially focused on India due to the large underserved need in student and employee mental wellbeing, increasing institutional adoption driven by policies like the National Education Policy 2020, and the opportunity to build high-volume, diverse behavioral datasets across languages and cultural contexts. At the same time, our platform is designed from the ground up to be globally deployable, with a cloud-native architecture, multi-language capabilities, and AI models that adapt to cultural nuances in emotional expression. As we validate and scale in India, we plan to expand into Southeast Asia and the Middle East, followed by mature markets like the US and Europe, particularly targeting universities and enterprise wellbeing programs. This approach allows us to combine strong local validation with a clear pathway to becoming a globally scalable mental wellbeing intelligence platform. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/YnOOlhWtzO40-Souloxy-Pitch-Deck-Avinya.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/YnOOlhWtzO40-Souloxy-Pitch-Deck-Avinya.pdf |
https://tinyurl.com/mr42db5c |
Souloxy is a mission-driven and impact-focused startup built around the vision of making preventive mental wellbeing accessible, scalable, and stigma-free for students and working professionals.
Our mission is to shift the current system from reactive, crisis-based intervention to a proactive, data-driven, and continuously supported wellbeing model. Today, most institutions rely on limited counselling infrastructure that activates only after problems escalate. Souloxy addresses this gap by combining AI-driven early detection, personalized interventions, and human support, enabling timely and preventive care at scale.
Our impact is designed at two levels:
Individual level: helping users build emotional awareness, resilience, and healthier behavioral patterns through continuous support
Institutional level: enabling universities and organizations to access anonymized wellbeing intelligence, identify at-risk cohorts early, and design targeted interventions
We measure impact not just by usage, but by meaningful outcomes—improved engagement, increased help-seeking behavior, reduced stigma, and early identification of mental health risks.
At scale, our goal is to build a preventive wellbeing infrastructure layer for campuses and workplaces, positively impacting millions of lives. By integrating AI with human care and responsible design, Souloxy aims to make mental wellbeing accessible, personalized, and proactive, rather than episodic and inaccessible. |
NA |
IIT Alumnus |
checked |
| Apr 28, 2026 @ 8:25 PM |
Prashant Mathur, Anuj Kumar |
prashant@t-teg.com |
https://www.linkedin.com/in/transteg-prashant-narayan-mathur/ |
https://t-teg.com/ |
9971002706 |
Prashant Mathur, CEO, Background:: About two decades in corporate and government system in telecom. solar, automobile ; Anuj Kumar, CTO, Background: Indian Railway, Nokia, BEL for two decades |
Served together Bharat Electronics for 3 years, Later worked together at Reliance Infocomm at CXO level together, Pursuing Transteg Lifecycle since last 7 years together |
two |
Anuj as technical brain and Prashant leads in business side, having great repo in managing growth of the company. I bring common sense and trust towards team building whereas combined we as a team are focused and effective |
Transteg Lifecycle Pvt Ltd |
https://t-teg.com/ |
Gurgaon |
Transteg provides IoTs integrated futuristic technologies to small utilities construction contractors as technology enabler |
Small contractors technological enablement for construction in data led organised operations helping revenue margins |
Utilities deployment on ground by small contractor is not supported by robust technological solutions in India whereas large utilities companies are using top technological solutions for their project management. The gap impact timeline, cost and transparency. We support small contractors. Our method is: Digital mapping of all activities automatically inline with small contractor activities leading to construction process support, cost efficiency, financials, real time project progress and creation of digital twin for maintenance purpose, using innovative IoTs on T-TEG platform. |
It is unique as it is not thought yet in wireline industry for OFC deployment despite OFC deployment is happening since last 30 years in India. Global solutions are very very basic. In India it is ALL manual work and very basic technological support is taken. Transteg is first and unique in India. It is not defensible on idea basis as idea will be copied very soon after seeing the Transteg success but the way it is designed and implemented, it is difficult to copy for quite a good time. |
Revenue |
two |
Small construction contractors in Optical Fiber Cable Outside plant deployment business. Large contractors are eventual customers |
USD 9.6 Bn |
USD 481 Mn |
USD 129 Mn |
We sell hardware IoTs and Software T-TEG platform services. For Small contractors (clients) it is IoT in Capex + SAAS for T-TEG platform. For Larger Clients it is based on deal viz IoTs + SAAS/ Licence/ Wholesale/ Module based etc |
Description below as Company name: How the work: 1. Locusview (US based): Address only Gas & electricity. Data collection manpower oriented 2. Lepton (India based): Only in telecom prominently for GIS survey only. Not a complete solution for Telecom 3. Underground Magnetic (US based): Focus limited to accurate depth only. Very limited services for deployment purpose 4. T-TEG (Transteg): End to end delivery (starting from planning to billing till approval) and supported by required IoTs (specially developed for the purpose). Complete lifecycle for OFC Outside plant deployment. |
By winning the technology and pricing criteria from 1. Previous network connect while working in similar industry (BSNL/ Airtel/ Voda/ Jio) 2. Reaching supply chain/ technology head where we are not having known connect (Lightstorm/ Netz/Gegatel/DEPL etc) 3. Defence/ ISRO/ Railways/ Public Sector through Tender 4. Working as partner on panel with large companies (RVNL/ITI/TCIL/Railtel/Gujratnet/Andhrafiber etc) 5. Large contractors are offloading the work to small contractors - we directly connect them (Alert/ JMD/ Neelakshi etc) |
Networking & Word of Mouth |
Democratisation of OFC civil contract deployment work with 99.99% efficient deployable process covering accurate and automatic work progress data capture creating trust in system and so on small contractors deliveries. T-TEG would be not only serving for telecom, electricity and gas but will be providing services in canal, waste drains, building construction, security civil infra but many more streams of business. |
Private Limited |
NIL |
one |
Mentorship, Network, Learning on Business models |
T-TEG platform quality & audit knowledge, Skill for connecting new customers, How to price the product to every new customer, How to close deal, When to exit with GTM efforts on specific prospect |
Yes – Open |
one |
two |
It is automated Business Process Management, which includes
1. Process Design
2. Process Execution
3. Task Management
4. Monitoring & Analytics
Our platform incorporates a workflow-driven architecture aligned with Business Process Management principles. Each stage of the OFC lifecycle—planning, deployment, inspection, and operations—is digitally orchestrated through task-based workflows. The system ensures controlled execution via task assignment, validation-based closures, and real-time monitoring, effectively functioning as a BPM engine tailored for telecom infrastructure management. |
Our platform implements AAA-based security by enforcing authentication for device and user identity, authorization through role-based access control, and accounting via detailed activity logging. |
Not known. Need to be learn |
No metric used today |
All communications are secured using encryption protocols such as TLS, ensuring data confidentiality and integrity across IoT devices, mobile applications, and backend systems |
Not known yet. Seems it will be in effect after few years as lot of discussion is going on relate to communication system security risk |
The system architecture is designed as a modular, scalable, and secure framework to support telecom and infrastructure lifecycle management across diverse environments. It comprises a data acquisition layer that captures real-time and batch data from field assets, sensors, and operational systems, followed by an application and processing layer that manages core workflows such as quality monitoring, asset tracking, and operational fault detection. A centralized database layer stores structured and unstructured data to support analytics, reporting, and historical analysis. An integrated AI/ML engine is deployed for targeted use cases including predictive maintenance, anomaly detection, and data fraud identification, with scope for further enhancement based on client needs and commercial viability. The integration layer enables seamless connectivity with client systems, OEM tools, and third-party platforms through secure APIs. A user interface layer provides web and mobile dashboards for real-time visibility, control, and decision-making. Additionally, robust security and monitoring components ensure data protection, access control, system reliability, and performance tracking, enabling continuous improvement and scalability of the overall solution. |
Not a team but one resource |
Open-source models (Llama, Google Gemma, etc. with MongoDB. |
Looking for guided mentoring in model performance |
Primarily for India and will go Global in 12 - 18 months when sufficient market is gained in India |
|
|
|
|
|
|
| Apr 28, 2026 @ 6:56 PM |
Brajendar singh Sikarwar |
edenfieldsbiotech@gmail.com |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 28, 2026 @ 5:13 PM |
Sachin Marihal |
chetan@neuraforesight.com |
http://linkedin.com/in/sachin-marihal-bab63917 |
http://www.neuraforesight.com |
9538033363 |
Sachin Marihal- Co-Founder & VP – Business Development Pharma Strategy & Global Market Access. 15+ years experience (Pharma R&D, formulation development, and international business strategy. Aravind P- Co-founder and Chief Technical Officer. Expertise in the pharmaceutical development across different geographies including India and China. 15+ years of experience in Formulation research and development & Consultation. |
The co-founders worked together as colleagues for over 14 years in the industry, each bringing unique expertise within the same domain. |
All |
The co-founders bring diverse expertise in product development, compliance, marketing, sales, and a strong understanding of global markets and strategies with a proactive mindset. As full-time contributors, we are fully committed and motivated by a relentless, results-driven attitude. A cross-functional team with pharma expertise and Al engineering, supported by IIIT Dharwad faculty. Defined roles across strategy, development, and data pipelines have enabled validated early-stage prototypes and strong execution capability. |
NEURAFORESIGHT PRIVATE LIMITED |
www.neuraforesight.com |
Bangalore |
AI-powered pharmaceutical Intelligence Platform |
Pharmaceutical R&D is a complex and data-intensive process. From early-stage drug discovery to clinical trials and regulatory submissions, researchers and decision-makers rely on a vast array of information to guide their work. However, the current landscape is characterized by significant challenges like data fragmentation, reliance on open source, static Insights, lack of domain specificity leading to slow decisions and limited contextual intelligence. |
Neuraforesight aims to address this challenge by aggregating, standardizing, and enriching multi-source data into a single AI-powered platform with a suite of powerful features designed to provide, enabling faster, data-driven decisions and deliver competitive intelligence and analytics. |
The platform leverages the power of Artificial Intelligence (AI) and sophisticated algorithms Machine Learning (ML) to deliver actionable insights that go beyond traditional data analysis. |
MVP |
Testimonials |
Global and regional consulting firms engaged in pharma strategy, market research, competitive intelligence, and regulatory consulting. Big investment houses, venture capital firms, and IPO advisory firms evaluating pharmaceutical and biotech portfolios. Pharma and Biopharma CROs, MNCs, MSMEs, SMEs, and Startups |
$1.48 trillion. |
$50 billion. |
$2.5 billion. |
Subscription-based pricing tiers tailored to customer size and needs. |
Clarivate Cortellis, GlobalData, IQVIA etc. |
Neuraforesight adopts a B2B, relationship-driven customer acquisition strategy focused on pharmaceutical, biotech, and research organizations. 1. Direct Outreach & Industry Networks, 2. Pilot Programs & Proof of Value, 3. Strategic Partnerships |
Phase 1: Validation & Pilot (0–6 months) Target a small group of pharma companies, CROs, and research teams. Deploy MVP with core features (data extraction, intelligence dashboards). Gather feedback, validate use cases, and refine product-market fit. Phase 2: Early Commercialization (6–12 months) Transition pilot users into paid customers via subscription-based SaaS model. Focus on high-value use cases such as competitive intelligence, clinical pipeline tracking, and regulatory insights. Build case studies demonstrating ROI and efficiency gains. Phase 3: Scale & Expansion (12+ months) Expand into global markets with enhanced datasets and AI-driven predictive analytics. Introduce tiered pricing models (enterprise, mid-size pharma, research institutions). Strengthen partnerships with global data providers and integrate multi-source intelligence. |
Neuraforesight aims to become a global AI-powered pharmaceutical intelligence platform that transforms how drug development and strategic decisions are made. We envision building a unified, continuously evolving data ecosystem that delivers predictive insights, reduces manual research effort, and accelerates innovation. Ultimately, our goal is to enable faster, smarter, and more efficient drug research and development worldwide. |
Incorporated on 15 April 2025, CIN: U21001KA2025PTC201243 |
INR 20 lakh |
No |
We are applying to Avinya to accelerate Neuraforesight’s growth through mentorship, industry connections, and structured support. The program’s ecosystem can help us refine our product-market fit, strengthen our go-to-market strategy, and connect with key stakeholders in the pharma and tech space. |
During the programme, we aim to validate our platform with pilot customers, refine our AI-driven features based on expert feedback, and build a strong commercialization roadmap. We also aim to secure early partnerships and prepare for scale. |
Yes, we are open to both hybrid and virtual participation and can actively engage in either format as required by the programme. |
Yes |
Core engine |
Neuraforesight uses a modular AI architecture combining data ingestion, NLP pipelines, and intelligence layers:
1. Data Layer: Automated ingestion from multiple sources (scientific literature, clinical trial registries, regulatory databases, proprietary datasets) using web scraping and APIs.
2. Processing Layer: NLP pipelines built on transformer-based models (e.g., LLM and domain-adapted models like for entity extraction, summarization, and normalization of unstructured pharma data.
4. Knowledge Layer: Structured storage using relational databases and vector databases (for semantic search and retrieval-augmented generation).
5. AI/Inference Layer: SLM or LLM-powered analytics for question answering, insights generation, and trend identification using RAG (Retrieval-Augmented Generation).
6. Application Layer: APIs and dashboards delivering real-time intelligence, visualization, and user-specific insights. |
Our proprietary advantage comes from a continuously evolving, structured pharmaceutical knowledge base built on top of licensed and public datasets:
1. Core Dataset: Access to curated databases
2. AI-Generated Structured Data: Our pipelines transform unstructured data (literature, trials, filings) into standardized, queryable formats (entities, relationships, knowledge graphs).
3. Data Enrichment Layer: Continuous normalization, deduplication, and linking across sources creates deeper context not available in raw datasets. |
Neuraforesight’s AI is defensible through a combination of proprietary data, domain specialization, and continuous learning loops:
1. Proprietary & Curated Data Layer
2. Domain-Specific AI Models
3. Knowledge Graph & Contextual Intelligence.
4. RAG-Based Architecture
Together, these create high data moats, technical differentiation, and increasing returns to scale, making the platform difficult to replicate. |
We evaluate model performance using a combination of quantitative metrics, domain validation, and real-world feedback loops:
1. NLP/Extraction Tasks: Precision, Recall, F1-score for entity recognition, and accuracy of structured data extraction.
2. LLM Outputs: Human evaluation for factual accuracy, coherence, and hallucination rate, along with automated checks for grounding against source data.
3. System Reliability: Latency, response time, uptime, and API success rates.
Domain Validation: Expert review by pharma SMEs to ensure clinical and regulatory correctness. |
Data is there in our own server not accessible by anyone with the database abstracted in a many to many relational db.
We evaluate model performance using a combination of quantitative metrics, domain validation. |
There are no major regulatory risks at present, as we primarily use publicly available and compliant data sources. However, we remain mindful of evolving AI and data regulations. We mitigate this through responsible data usage, clear positioning as a decision-support platform, and continuous monitoring of regulatory developments. |
If the system is scaled 10×, the LLM can become a bottleneck if the infrastructure is not scaled accordingly. This can lead to increased latency and slower response times. Additionally, hallucinations may increase indirectly due to longer or noisier contexts and degraded retrieval quality under load. However, the primary risks are system-level issues such as compute saturation, queuing delays, and reduced pipeline efficiency rather than the model itself “breaking.” |
Yes, we have an in-house AI/ML team supported by collaboration with academic experts from IIIT Dharwad. The team has expertise in NLP, large language models (LLMs), and data engineering, with a focus on processing complex pharmaceutical and scientific data. |
We use a combination of licensed, publicly available, and internally processed datasets:
All data usage strictly adheres to licensing terms and regulatory requirements, ensuring compliance and reliability. Over time, our continuously enriched internal datasets form a proprietary data layer, strengthening our competitive advantage. |
We follow a continuous improvement approach driven by data quality, user feedback, and regular system refinement. By continuously enhancing our data sources, validating outputs with domain experts, and monitoring performance metrics, we ensure the platform becomes more accurate and reliable over time. |
We are building for both India and global markets.
Our platform is designed with a global-first approach, integrating international datasets (clinical trials, regulatory filings, scientific literature) to serve pharmaceutical and biotech companies worldwide. At the same time, we see strong opportunities in the Indian pharma ecosystem, including CROs, generics manufacturers, and research institutions.
This dual focus allows us to validate locally while scaling globally, ensuring relevance, scalability, and broader market impact. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/iPBmB8cXAtkl-Avinya-Cohort-NeuraForeSight-pitch-deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/iPBmB8cXAtkl-Avinya-Cohort-NeuraForeSight-pitch-deck.pdf |
https://1drv.ms/v/c/3fd0ec4c3267d010/IQANyty5nHhHS56EDlmlRNW0ARp5gLeStlbfGlIvIog0cOM?e=fzTSfy |
Yes, Neuraforesight is an impact-driven startup focused on improving efficiency and decision-making in the pharmaceutical ecosystem. By transforming fragmented scientific and clinical data into actionable intelligence, we aim to reduce research time, enhance innovation, and support faster drug development.
Our long-term impact lies in enabling better-informed decisions across R&D and strategy, ultimately contributing to improved healthcare outcomes and broader access to effective treatments. |
Founders from underrepresented geographies of North Karnataka (Dharwad / Bijapur / Haveri) |
Technical guidance of Deepak K T, Manjunath Vanahalli and Dr. Girish G. N., faculty members of IIIT Dharwad |
checked |
| Apr 28, 2026 @ 4:49 PM |
Pushpesh Gaurav |
pushpesh@healjour.com |
https://www.linkedin.com/in/pushpeshgaurav/ |
https://www.healjour.com/ |
8860743544 |
Pushpesh Gaurav: Founder, CEO - (a)12+ years experience in diagnostics operations & sales. (b) Built a network of diagnostics partners across 12k pincodes within 10 months for an aggregation business. (c) Responsible for Sales, Partnerships, Network, Operations & Fundraising. Shobhit Saini: Co-founder, CTO - (a) 5 years experience building SaaS products for health aggregators. (b) Built & deployed SaaS system integrations with some of the largest chains like Apollo Clinics. (c) Responsible for Product development, Coding & Architecture. |
Shobhit & I met at our previous organization - MedPiper, in 2022. Shobhit was a senior software developer, heading the tech team and product development and design. I was heading the operations at MedPiper, where I was responsible for building the structure of team, hiring, defining protocols for validation and quality control.
Shobhit & I worked extensively building the customer journey at MedPiper insur-tech portal, where we collaborated in launching and scaling the business and revenue from 0 to 1 Cr. MRR in span of 8 months. |
two |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 28, 2026 @ 4:33 PM |
Shubham Kumar, Zulkifl Uddin Khairoowala |
shubhambpn1234@gmail.com |
https://www.linkedin.com/in/zulkifl-khairoowala-b5680a194/ |
https://kalpitlabs.com |
7302487854 |
Shubham - CEO & CTO, Zulkifl - CMO |
met though a common friend and is working together since last 2 years |
All |
We're a small, focused founding team with rare full-stack coverage: Shubham brings 2+ years of production ML engineering plus proven offensive security capability, and Zulkifl brings research depth from his Quantum Computing Knowledge, GTM and distribution execution. Builder + storyteller, offense + defense—exactly what AI security demands. |
Kalpit Labs Private Limited |
https://kalpitlabs.com |
Aligarh, Uttar Pradesh |
KalpitLabs is the Cloudflare of AI Security for India—guardrails, AI firewall, and red teaming built for Indian languages and DPDPA compliance. |
Indian AI products are exposed to prompt injections, jailbreaks, and system prompt theft—yet every AI security tool today is built for English and ignores Indian languages, regulations, and attack patterns. With DPDPA now live and penalties up to ₹250 Crore, companies have no native infrastructure to defend their AI or prove compliance |
KalpitLabs is a three-product platform that protects, inspects, and certifies Indian AI systems end-to-end.
Rakshak (AI Guardrail API) is a drop-in REST API that blocks prompt injections, jailbreaks, and PII leaks through a 3-stage pipeline (regex, embedding similarity, LLM classifier)—with native Indic language support.
Kavach (AI Firewall) sits in front of any LLM API for real-time traffic inspection, rate limiting, and threat logging—like Cloudflare WAF, but purpose-built for AI.
Niriksha (Red Teaming + Compliance) delivers adversarial audits of Indian AI products with a full DPDP compliance evidence trail, breaking AI before attackers do.
Rakshak is live with our first pilot underway, and our public disclosures have already established the credibility that drives our GTM. |
Three moats compound over time: Indic adversarial depth. We're the only AI security provider with native support for Hindi, Tamil, Bengali, Hinglish, and 10+ Indian languages—built from months of adversarial research. A global tool can't replicate this overnight by adding a language pack; the dataset, attack patterns, and regulatory context have to be earned. Proof before product. Most AI security companies claim capability through frameworks and certificates. We claim it through CVE-style disclosures on production systems—Sarvam AI's 105B Indus (acknowledged, case study agreed), KissanAI Dhenu, and Le Chat (Mistral). Every red teaming engagement feeds new detection rules into Rakshak, creating a flywheel where offense sharpens defense. India-first architecture. DPDPA compliance, Indian PII formats, Indian pricing, and Indian attack patterns are built in from day one—not retrofitted. Combined with a three-product platform (guardrails + firewall + red teaming) under one roof, switching costs grow with every product a customer adopts. |
MVP |
Pilots |
Indian AI-first startups, fintech and healthcare companies deploying LLMs in production, enterprise IT teams building internal AI applications, and government bodies under DPDPA compliance pressure. Primary ICP: Series A–C Indian companies with AI products in production handling sensitive user data, and DPIIT-recognized AI startups under regulatory scrutiny. |
$50B+ |
₹800 Cr+ |
₹40–60 Cr |
Three streams combining recurring SaaS with services revenue: Rakshak (AI Guardrail API): Usage-based SaaS. Free tier (10K calls/month), Starter ₹4,999/mo, Growth ₹19,999/mo, custom Enterprise pricing. Kavach (AI Firewall): Subscription SaaS with per-endpoint pricing, traffic-volume tiers, SLA guarantees, and on-prem option for enterprise. Niriksha (Red Teaming + Compliance): One-time adversarial audit fees plus annual DPDP compliance retainers; targets enterprise and government SLAs. The three streams reinforce each other: Niriksha audits surface vulnerabilities that drive Rakshak and Kavach adoption. |
Repello AI,Fencio ,AIShield , |
Inbound through public credibility. Our responsible disclosures on Sarvam AI's 105B Indus, KissanAI Dhenu, and Le Chat (Mistral) are published as CVE-style advisories on the KalpitLabs blog. These generate direct inbound interest from Indian AI companies who want their own systems tested before shipping.Outbound to a defined ICP. Targeted cold outreach to Indian AI startups, fintech CTOs, healthcare AI leaders, and compliance heads—using category-specific email templates and a curated lead list of funded AI companies in production.Content distribution. Technical writeups, security advisories, and consistent LinkedIn/X presence converting research into reach. Every disclosure becomes a piece of distribution.Developer funnel. Rakshak's free tier (10K calls/month) creates self-serve onboarding for AI engineers, who later upgrade to paid SaaS as their usage scales.Services-led entry. Niriksha red teaming engagements act as a high-trust wedge—companies that audit with us convert into Rakshak and Kavach subscribers post-engagement. |
Phase 1 — Land first paying customers (Now–Q3 2026): Use Niriksha red teaming as the primary entry point—it's services revenue, builds case studies, and uncovers vulnerabilities that customers want Rakshak and Kavach to defend against. Continue public disclosures to compound the credibility moat.Phase 2 — Build the platform funnel (Q3–Q4 2026): Launch Kavach beta. Convert Niriksha clients into Rakshak and Kavach subscribers (cross-sell). Publish India's first Indic adversarial benchmark dataset to anchor category leadership and drive inbound.Phase 3 — Scale through partnerships (2027): Partner with Indian cloud providers, system integrators, and DPDPA compliance consultancies for distribution. Pursue CERT-In empanelment and government partnerships to position KalpitLabs as default AI security infrastructure for regulated sectors (BFSI, healthcare, government).Channels we will not chase early: paid ads, generic SEO, conference circuits. Our wedge is earned credibility, not bought attention. |
To become the Cloudflare of AI Security—the default infrastructure layer between the world and every AI product in India. Just as Cloudflare became indispensable for web security, KalpitLabs will be indispensable for AI security: inspecting, protecting, certifying, and auditing AI systems at scale.
In the near term, that means making it impossible to deploy a serious AI product in India without KalpitLabs in the stack. Long term, we expand from India to other emerging markets where global English-first tools fail to serve—Southeast Asia, the Middle East, and Africa—building the world's first AI security infrastructure designed for non-Western languages, regulations, and threat models. The end state is a category-defining company that owns AI security infrastructure for the half of the world that English-first tools have ignored. |
Private Limited |
0 |
Yes |
Avinya is uniquely positioned for what KalpitLabs needs at this stage. Three reasons:Ecosystem alignment. Avinya is backed by IIT Alumni Centre Bengaluru, the Karnataka government, and a curated network of AI investors and industry partners. AI security is a category where credibility and distribution matter as much as product, and Avinya offers both at the right stage.Mentorship in our exact gap. We've shipped product, built credibility through public disclosures, and have our first pilot underway—but we need senior guidance on enterprise sales motions, pricing security services, and navigating regulated-sector procurement (BFSI, healthcare, government). Avinya's mentor pods of IIT alumni, founders, and domain experts directly address this gap.Capital efficiency at the right tranche. ₹10 lakhs of seed plus the potential ₹1 crore follow-on aligns perfectly with our ₹25–50L pre-seed plan—it lets us extend runway, sharpen GTM, and prove the platform thesis before a larger seed round.Karnataka access. Bengaluru is the densest concentration of AI buyers in India. Co-working space, pilot opportunities, and proximity to enterprise CTOs accelerate our customer acquisition meaningfully. |
By the end of the 16 weeks, we aim to have: 5+ paying customers across Rakshak (SaaS) and Niriksha (red teaming retainers), validating the platform thesis Kavach (AI Firewall) beta launched with 1–2 design partner pilots in regulated sectors Indic adversarial benchmark dataset v1 published, anchoring KalpitLabs as the category authority Sarvam AI case study published, plus 1–2 new responsible disclosures to compound credibility Seed round preparation complete—deck, metrics, investor pipeline, and warm intros via the Avinya network Enterprise pilot in BFSI or healthcare secured through Karnataka ecosystem partnerships Founder-level clarity on pricing, sales motion, and category positioning through mentor pod engagement |
Yes — open to hybrid. Founder will travel to Bengaluru for all mandatory in-person sessions (mentorships, reviews, Demo Day) while leveraging virtual participation for asynchronous work. Aligarh-based core team remains technically productive in parallel. |
Yes |
Core engine |
KalpitLabs uses a layered AI architecture across three products:Rakshak (3-stage pipeline):
Layer 1 — Regex filtering: Pattern-based detection for known PII (Aadhaar, PAN, UPI, GSTIN), prompt injection signatures, and jailbreak templates. Sub-millisecond latency, zero compute cost.
Layer 2 — Embedding similarity: Sentence-transformer embeddings matched against a curated vector database of known attack patterns and Indic adversarial examples. Catches semantic variants that bypass regex.
Layer 3 — LLM classifier: A small LLM (deployable on commodity GPUs or shared inference) for nuanced edge cases—context-dependent attacks, multi-turn manipulation, and ambiguous intent. Optional, deployed only when accuracy needs justify the latency cost.
Kavach (AI Firewall): API gateway layer with anomaly detection on traffic patterns—rate, payload distribution, temporal sequences. Uses statistical baselines and lightweight ML for outlier detection.Niriksha: Human-led red teaming augmented by automated adversarial generation tools (agentic LLM systems we've built internally for vulnerability discovery).The architecture is intentionally modular—each layer can be deployed independently, tuned per customer, and scaled separately based on threat profile and budget. |
We are building India's first Indic adversarial dataset—a proprietary corpus of prompt injection patterns, jailbreak attempts, and PII-targeting attacks across Hindi, Tamil, Bengali, Hinglish, and 10+ Indian languages. This dataset is built from:
Original red teaming research on Sarvam AI's 105B Indus, KissanAI Dhenu, and Le Chat (Mistral)—each engagement contributes new attack vectors specific to Indian context
Indian PII pattern libraries for Aadhaar, PAN, UPI, GSTIN, voter ID, ration card, and other India-specific identifiers
Hinglish and code-switched attack patterns that bypass English-only tools
DPDPA-aligned compliance signatures mapped to regulatory categories
This dataset cannot be replicated by global tools by simply adding a "language pack"—the attack patterns, cultural context, and regulatory framing took months of adversarial research to build and grow with every engagement. |
Indic adversarial depth. Our dataset and detection models are trained on Indian attack patterns no global tool has. Every red teaming engagement and customer deployment adds new adversarial examples, sharpening detection accuracy in a domain global competitors are blind to.Offense-defense feedback loop. Niriksha (red teaming) discovers vulnerabilities → Rakshak (guardrails) gets new detection rules → Kavach (firewall) gets new threat signatures. Three products feeding each other create a flywheel competitors with single products can't match.Public credibility moat. Our responsible disclosures on Sarvam AI, KissanAI, and Le Chat are documented proof of capability. New entrants can claim AI security expertise; we've published it.Regulatory anchoring. DPDP compliance mapping is built into the architecture from day one—not retrofitted. Indian companies under regulatory pressure choose us not just for security but for audit-ready evidence trails. |
We benchmark against open-source baselines (Guardrails AI, Llama Guard) on Indian attack corpora to demonstrate quantitative superiority in our domain. |
Privacy and compliance are foundational, not features:
DPDPA-aligned by design: Data minimization, purpose limitation, and consent frameworks built into Rakshak's architecture
No customer data retention by default: Inference is stateless; logs are opt-in and customer-controlled
On-prem deployment option: Enterprises and government clients can deploy Rakshak and Kavach within their own infrastructure—no data leaves their environment
Encryption: TLS in transit, AES-256 at rest for any persisted data
Audit trails: Every detection event is logged with cryptographic integrity, providing regulators with tamper-proof compliance evidence
Access control: Role-based access, API key scoping, and IP allowlisting for enterprise tiers
Responsible disclosure ethics: Our own red teaming follows strict 7-day responsible disclosure norms—reflecting how seriously we take security and trust |
Three areas we actively monitor:
DPDPA evolution. The Digital Personal Data Protection Act is live but rules are still being notified. Changes in Significant Data Fiduciary classification, consent frameworks, or cross-border data transfer rules could shift compliance requirements. We treat this as opportunity—every regulatory tightening expands demand for our products.
CERT-In directives. Cybersecurity reporting requirements may extend to AI systems specifically. We're positioning to be CERT-In empaneled, turning compliance into a tailwind.
AI-specific regulation. India's AI governance framework is under active development. New requirements around model auditing, bias testing, or transparency could create both obligations and opportunities for us as the certifying layer.
Cross-border concerns. If we expand to international markets (SEA, MENA), we'll need to navigate GDPR, regional data protection laws, and AI acts like the EU AI Act.
The net read: regulation is a tailwind, not a headwind, for AI security infrastructure |
Multi-tenant isolation. Current single-tenant deployments are simple; multi-tenant SaaS at scale requires hardened tenant isolation, per-customer rate limiting, and noisy-neighbor protection.
Adversarial dataset coverage. New attack patterns emerge weekly. At 10x customer count, manual curation breaks. We're building a semi-automated pipeline using agentic LLMs to generate, validate, and ingest new adversarial examples.
Operational tooling. Monitoring, alerting, and incident response need to mature from founder-led to systematized. |
Shubham Kumar (Founder & CEO) — 2+ years of production ML engineering across fintech, retail, and ecommerce. Built fraud detection systems, GAN models, RAG pipelines, and agentic AI deployments. Personally led offensive ML research that produced disclosures on Sarvam AI's 105B Indus, KissanAI Dhenu, and Le Chat.
Md. Fazil (Cyber Security Analyst) — Extends offensive research bandwidth across red teaming engagements, adversarial dataset construction for Indic languages, and vulnerability research.
Zulkifl Khairoowala (Co-Founder & CMO) — Visvesvaraya Research Scholar in Quantum Computing at ZHCET, M.Tech in Computer Software Engineering. Brings research depth that supports technical credibility and product-research alignment.
The team combines applied ML engineering, offensive security, and academic research—rare for a pre-seed AI security startup. |
Three data sources, all owned or appropriately licensed:
Owned:
Indic adversarial dataset (proprietary, built from our red teaming research)
Indian PII pattern library (proprietary, open-sourced subset on PyPI)
Sarvam AI / KissanAI / Le Chat disclosure findings (our own research output)
Open-source / public domain:
Public prompt injection corpora (e.g., PromptBench, AdvBench) used for benchmarking
Open Indic language corpora (AI4Bharat datasets, where applicable, under their licenses) for embedding training
Customer-derived (with consent):
Anonymized attack patterns surfaced during pilot deployments, contributed back to our dataset under explicit customer agreement
We do not use any unlicensed proprietary data. All training and benchmarking data has clear provenance. |
A multi-loop improvement system:
1. Adversarial discovery loop: Every Niriksha red teaming engagement surfaces new attack patterns, which feed directly into the Rakshak detection corpus.
2. Customer telemetry loop: Anonymized blocked-attack data from customer deployments (with consent) helps identify novel attack vectors and refine detection thresholds.
3. Public research loop: New AI security research, CVE disclosures, and academic papers are continuously monitored and integrated.
4. Active red teaming: Internal red team continuously stress-tests Rakshak against the latest attack techniques. Our own disclosures (Sarvam, KissanAI, Le Chat) feed back into our defense models.
5. Model fine-tuning cadence: Embedding models and Layer 3 classifiers are re-trained quarterly on the updated adversarial corpus.
6. Indic language expansion: Language coverage expands sequentially—Hindi and Hinglish today, Tamil and Bengali next, then Marathi, Telugu, and others based on customer demand.
7. Benchmarking: Quarterly benchmarks against open-source baselines (Guardrails AI, Llama Guard) and commercial alternatives, published as industry-leading category benchmarks. |
India-first, with a clear global expansion path.
Phase 1 (Now–2027): India. We're laser-focused on the Indian market—where the regulatory tailwind (DPDPA), language gap (no global tool serves Indic), and ecosystem density (3,000+ AI startups) create a defensible early market. Becoming the default AI security infrastructure in India is our 2-year mission.
Phase 2 (2027 onward): Emerging markets with similar profiles. Southeast Asia, Middle East, and Africa—markets where global English-first tools fail to serve local languages, regulations, and threat models. Our India playbook (Indic dataset, regulatory-first architecture, services-led entry) translates directly.
Phase 3 (long term): Global category leadership. Once we've established category leadership in non-Western markets, we expand into developed markets as the AI security provider purpose-built for linguistic and regulatory diversity—a positioning no US-headquartered competitor can credibly claim. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/M9rKAcX5RWCx-Kalpit_Labs_Updated.pptx-2_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/M9rKAcX5RWCx-Kalpit_Labs_Updated.pptx-2_compressed.pdf |
https://www.kalpitlabs.com/blog/how-i-found-critical-security-vulnerabilities-in-indus-indias-sovereign-ai-assistant |
KalpitLabs exists because India is deploying AI faster than it's securing it—and the people most at risk are not enterprise users in Bengaluru, but ordinary citizens whose Aadhaar, PAN, and UPI data flows through AI systems they never chose to interact with.
Our mission is to make Indian AI safe by default—for the rural user receiving a fraudulent SMS in Hindi, for the patient whose health data is processed by an English-first chatbot, for the small business owner whose financial information passes through an LLM with no guardrails. Every responsible disclosure we publish (Sarvam AI, KissanAI, Le Chat) is shared openly so the entire Indian AI ecosystem benefits, not just our customers.
We also believe AI security should be accessible, not a luxury. Global tools price Indian startups out of basic protection. By building India-first—with Indian pricing, Indian languages, and Indian compliance—we're democratizing AI safety for the next generation of Indian builders.
Beyond product, we open-source tools like our Indic PII detector to raise the security floor across the industry. Our long-term vision is for KalpitLabs to be the reason India's AI ecosystem becomes a global model for responsible deployment—not a cautionary tale. |
KalpitLabs is being built from Aligarh, Uttar Pradesh—a Tier-3 city far from India's traditional startup hubs of Bengaluru, Mumbai, and Delhi-NCR. Founder Shubham Kumar is a self-taught ML engineer without a premier institute (IIT/IIM) background, building a deep-tech AI security company from outside the established ecosystem. The full team is based in Aligarh and operates without the fundraising networks, investor proximity, or peer mentorship that founders in Tier-1 cities take for granted. |
none |
checked |
| Apr 28, 2026 @ 4:12 PM |
Satish Gupta, Avinash Kumar, Himanshu Shekhar |
satish@q1k.ai |
https://www.linkedin.com/in/satish-gupta-a359746/ |
https://q1k.ai |
9769041807 |
Satish Gupta - CEO - 2nd time founder with 18 years of experience in building 0 to 1 and 1 to 50 businesses out of which 8 years experience in Retail/E-com industry. Avinash Kumar - CTO - 15 years of experience in building scalable tech at companies like Flipkart, Testsigma. Himanshu - CPO - 6 Years of working experience with multiple startups and vast experience in building 0 to 1 products. |
We met 4 years back while working together in the previous company. |
2 |
Deep domain knowledge and experience of Retail/E-com industry. |
Q1k Ai |
https://q1k.ai |
Bengaluru |
Q1k AI is an AI native autopilot for Retail operations. |
Retail operations run on rules—but every day, exceptions break them. Inventory mismatches, return spikes, fulfilment failures. Exceptions occur every day and they can impact up to 5% of topline in lost revenue. Today, teams handle this chaos manually—across emails, spreadsheets, Slack—making exceptions the #1 operational time sink. |
Q1k is an AI exception handling AI autopilot for retail operations. Instead of just sending alerts, the system automatically investigates the exception and identifies the root cause. So instead of teams spending hours figuring out why something happened, the system tells them:
“Here is the issue, here is the likely root cause, and here’s what you should do next.”
The outcome is simple: Exceptions get resolved faster and your teams spend far less time firefighting. |
Our technological edge are the context modeling engine and Behavior tree based agent orchestration that enables long running agents with persistent memory without hallucinations while minimizing token consumption. Our defensibility comes from the customer data intelligence that would make the switching cost expensive for any customer. |
MVP |
Pilots |
Retail / E-commerce companies in the revenue range USD 10 M - USD 500 M in India and USA. |
$50 Bn |
$5 Bn |
$200 Mn |
Subscription + Credits. |
Moxo, Glacis, Blueyonder. |
Early customers - Professional and Personal network. Mid Customers - LinkedIn and Email, WOM. Late Customers - Partnerships and PLG |
Content & SEO - Blogs, LinkedIn posts, Reddit, Quora. Outbound - LinkedIn and Email Marketing. |
To become a goto intelligence solution platform for the entire Retail industry stack - OMS, Procurement, Exceptions, Post Purchase. |
Axionics Innovations Private Limited |
NA |
Yes |
Mentorship, Customer Introductions, Access to future funding rounds. |
Pilot with 5 target companies. Iterate on the product basis pilot feedback to meaningfully start marketing spend on scaling. Raise seed fund. |
Yes |
Yes |
Core engine |
Multi-agent orchestration built around a four-layer pattern — Sense → Think → Act → Learn — where every module is self-contained and independently deployable on-prem.
- *Sense: 5-stage exception detection pipeline (ingestion → normalization → validation → anomaly scoring → classification) running over retail / ecommerce data flows (orders, inventory, fulfillment, payments).
- *Think: 26 specialized agents reasoning over a per-tenant knowledge graph (built on our open-source @q1k-oss/context-engine). Agents are stateless and routed via Temporal for durable, replayable workflows.
- *Act: Resolution agents executing remediations against connected systems (ERP/WMS/OMS/marketplaces) through an idempotent connector layer.
- *Learn: Outcome feedback updates the knowledge graph and per-tenant SOPs codified as reusable Exception Cards.
Cross-cutting: *MINT* serialization (~70% token reduction vs JSON), model tiering (small models route, large models reason), aggressive caching, and self-hosted auth (EdDSA JWT, Argon2id, WebAuth/Passkeys, HMAC-SHA256 SDK signing) so the full stack runs air-gapped if required. |
We don't own customer raw data — customers do. Our advantage is in derived structure:
- *Per-tenant knowledge graphs capturing the implicit business rules that emerge from how each customer actually resolves exceptions.
- *Exception Card library — an abstracted, anonymized template catalog of exception patterns + resolution playbooks that compounds across deployments without raw data ever leaving tenant boundaries.
- *Resolution outcome labels (success/failure/escalated/reversed) used to refine routing and tool-use policies.
Cross-customer learning happens at the pattern level, not the data level — which is exactly what makes on-prem deployability viable. |
Our defensibility compounds from integration depth, not model novelty:
- *Time to Clone: rebuilding the 5-stage detection pipeline, 26-agent orchestration, knowledge graph schema, connector library, MINT format, and Exception Card framework — and convincing customers to migrate trust-critical workflows — is a multi-quarter effort even for a strong team.
- *Structural cost moat: MINT + model tiering put us on a ₹34 → ₹5 per-order trajectory. Competitors on vanilla JSON and single-tier inference can't match this without rebuilding their stack.
- *On-prem deployability: most agentic AI products are cloud-only. We win regulated/enterprise buyers others structurally can't serve.
- *Connector + workflow lock-in: once wired into a customer's ERP/WMS/OMS with audited resolutions, switching cost is high.
- *Open-core flywheel: context-engine, kiban, and MINT are open-source — the orchestration, agents, and Exception Card library remain proprietary. Builds developer pull while keeping the moat intact. |
Three layers of metrics:
- *Detection: precision/recall on exception classification, false-positive rate, time-to-detect (event → classified exception).
- *Resolution: auto-resolution rate, escalation rate, MTTR, resolution accuracy (vs ground truth or human review), reversal rate.
- *System: agent step success rate, tool-use accuracy, structured-output hallucination rate, cost per order (₹), tokens per workflow, p50/p95/p99 latency.
Reliability enforced via Temporal (durable + replayable), idempotent writes, per-tenant sandboxing, and "golden test sets per Exception Card" that run on every prompt/model change before promotion. |
Q1k architecture is privacy-by-isolation:
- *On-prem / single-tenant deployable: the full stack — detection, agents, knowledge graph, auth — can run inside the customer's VPC or datacenter. No customer data needs to leave their perimeter.
- *Self-hosted auth: EdDSA JWT, Argon2id, WebAuthn/Passkeys, HMAC-SHA256 SDK signing. No dependency on third-party IdPs.
- *Data minimization: MINT strips structurally unnecessary tokens; PII redaction at the connector boundary.
- *Full audit trail: every agent decision (input → reasoning → tool call → output) is logged and tied to the originating Exception Card — usable as SOC 2 / ISO 27001 evidence.
- *Compliance posture: GDPR (residency via on-prem, erasure via tenant isolation), India DPDP, SOC 2 Type II on roadmap, architected to stay outside PCI-DSS cardholder-data scope.
- *Model independence: customers can plug in their own LLM endpoint — self-hosted (vLLM, Llama/Qwen/Mistral), Azure OpenAI tenancy, Bedrock, or licensed API. We never force a single inference provider. |
*AI liability frameworks are evolving: liability for autonomous agent actions. Mitigated by per-Card autonomy configuration and full decision audit trails. |
- *Temporal worker pools need horizontal autoscaling + per-tenant queue isolation to prevent noisy-neighbor effects.
- *Foundation model rate limits from any single provider — mitigated by multi-provider routing and on-prem inference for largest tenants.
- *Downstream connector quotas (Shopify, SP-API, etc.) — requires smarter batching and webhook-first patterns. |
Yes we have a lean AI team:
- *Avinash (Co-Founder, Engineering): owns detection, orchestration, resolution. Deep distributed-systems, agent-runtime, and Temporal expertise.
- *Himanshu (Co-Founder, Product): product, agent UX, SOP framework, market positioning. Published Explainable AI research; Neo Fellowship. |
- *Customer operational data: owned by customer, processed under DPA. *Not* used to train shared models.
- *Q1k-owned synthetic + curated data: exception scenarios generated from our Exception Card framework for seeding and stress-testing detection. Owned outright.
- *Open datasets for foundation evaluation (public e-commerce schemas, marketplace API documentation) — used under respective licenses. |
- *Knowledge graph compounding: every resolved exception adds structured signal to that customer's graph — improves accuracy without retraining.
- *Exception Card versioning: SOPs are versioned artifacts; improvements ship as Card updates and propagate across deployments.
- *Eval-first promotion: prompt/model/agent changes must pass golden tests per Card before reaching production.
- *Outcome-based fine-tuning: anonymized, abstracted resolution-outcome data optionally fine-tunes routing and tool-use models (customer opt-in).
- *Tier compression: as smaller models get better, we re-route workloads down the tier to drop cost with no quality regression.
- *Open-source flywheel: external usage and contributions to context-engine, kiban, and MINT stress-test and improve the core. |
Both. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/0Y3GAvEXNjXV-Q1k_Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/0Y3GAvEXNjXV-Q1k_Pitch-Deck.pdf |
https://youtu.be/0G3UWbnsxlw |
Our mission is to Make retail operations autonomous and self-correcting.
We are starting with exception detection and investigation → move toward automated decisioning → eventually enable systems that can resolve a large class of operational issues on their own. |
NA |
Mr Pradeep (IIT Kharagpur) |
checked |
| Apr 28, 2026 @ 3:46 PM |
A. Vaidyanathan |
vaidy@anilyx.com |
|
http://www.anilyx.com |
7349777288 |
A. Vaidyanathan – Co-Founder & Director : Tech entrepreneur with two successful exits and extensive experience in product strategy, business development, and technology commercialization. Advisory Team H. Shreyas – Co-Founder & Director : Electronics Engineer with strong experience in product innovation, embedded systems, and hardware design. Responsible for system architecture, sensor integration, and smart helmet product development. Dr. Raja N Moorthy – Co-Founder & Director : PhD with over 40 years of experience in Research & Development, innovation leadership, and technology commercialization. Dr. Moorthy has led multiple deep-technology initiatives and provides strategic direction for the Anilyx platform. Dr. Manas Gaur – Co-Founder : PhD (AI/ML) and Advisor – University of Maryland, Baltimore County (UMBC), USA. His expertise includes Artificial Intelligence, data analytics, and intelligent systems, supporting the AI architecture and predictive analytics platform for Anilyx. "A. Vaidyanathan – Co-Founder & Director : Tech entrepreneur with two successful exits and extensive experience in product strategy, business development, and technology commercialization. Advisory Team H. Shreyas – Co-Founder & Director : Electronics Engineer with strong experience in product innovation, embedded systems, and hardware design. Responsible for system architecture, sensor integration, and smart helmet product development. Dr. Raja N Moorthy – Co-Founder & Director : PhD with over 40 years of experience in Research & Development, innovation leadership, and technology commercialization. Dr. Moorthy has led multiple deep-technology initiatives and provides strategic direction for the Anilyx platform. Dr. Manas Gaur – Co-Founder : PhD (AI/ML) and Advisor – University of Maryland, Baltimore County (UMBC), USA. His expertise includes Artificial Intelligence, data analytics, and intelligent systems, supporting the AI architecture and predictive analytics platform for Anilyx. " " |
The founding team came together through a shared interest in deep-tech innovation, environmental health, and smart mobility solutions. The founders bring complementary expertise across technology commercialization, electronics engineering, AI/ML, and research & development. Collectively, the team has worked together across product innovation, hardware design, and technology strategy for several years while building Anilyx. |
All |
Our biggest strength lies in the combination of deep domain expertise, technical execution, and commercialization capability. The team brings together experience in AI, embedded systems, R&D, environmental sensing, and startup building. This multidisciplinary background enables us to rapidly move from concept to prototype while maintaining strong product-market alignment. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 26, 2026 @ 2:43 PM |
RAMESHRAO VASANTRAO KULKARNI, SAPNA KULKARNI |
rameshrao.kulkarni@keehoo.ai |
https://www.linkedin.com/in/ramesh-kulkarni/ |
http://NA |
9845233770; 9980097598 |
FOUNDER |
The founders of KeeHoo are a husband-and-wife team who share a strong personal and professional bond. They have been working together since the company’s inception in 2021, combining complementary skills in AI, technology, and business strategy to build and scale the venture. |
2 |
KeeHoo’s biggest strength lies in our multidisciplinary, collaborative team that combines deep expertise in AI, data engineering, and software development with creativity, agility, and a problem-solving mindset. Our ability to integrate technical excellence with strategic thinking allows us to deliver scalable, high-impact solutions while adapting quickly to client needs. |
KEEHOO PRIVATE LIMITED |
www.keehoo.ai |
HUBBALLI, KARNATAKA |
KeeHoo is a 2021-founded AI and data technology startup building intelligent, scalable solutions that simplify complexity and drive business growth. |
KeeHoo solves the problem of complexity in business operations by leveraging AI, data, and automation to deliver actionable insights, intelligent solutions, and scalable technology that drive faster, smarter decision-making. |
KeeHoo provides AI- and data-driven solutions that turn complex business challenges into actionable outcomes. By combining intelligent algorithms, automation, and low-code platforms, our products help organizations make faster decisions, optimize operations, and scale efficiently. |
KeeHoo's solution is unique because it combines AI, data engineering, and low-code automation into an integrated platform that adapts to each client’s needs. Its defensibility comes from our ability to rapidly deliver tailored, scalable solutions, continuously improve through real-time insights, and integrate seamlessly with existing systems - making it more than just a tool, but a strategic enabler. |
Product-Market Fit |
Pilots |
Enterprises AI Solutions (Sugar Industry, Energy and FaithTech) |
The global AI and enterprise automation market is estimated at $400–450B, encompassing AI software, automation tools, and data-driven enterprise solutions. |
Focusing on sectors KeeHoo is targeting - FaithTech, Enterprise Automation & Intelligence, and Digital Psychology - the SAM is roughly $50–60B, representing industries and geographies where our products have direct applicability. |
Considering our current reach, go-to-market strategy, and product readiness, the SOM is approximately $1–2B, representing the near-term revenue opportunity from clients we can realistically acquire in the next 3–5 years. |
KeeHoo generates revenue through a combination of: Product Sales & Subscriptions – Licensing AI- and data-driven products on a subscription or usage-based model. Staff Augmentation / Services – Providing skilled professionals to client projects on a time-and-materials or contract basis. Custom Solutions & Consulting – Building tailored solutions for clients with one-time or milestone-based fees. |
NA |
Acquires customers through a mix of strategic collaborations with key industry players, leveraging existing professional networks, and word-of-mouth referrals. This approach allows us to build trusted relationships, reach high-value clients, and grow sustainably through credibility and proven results. |
Our go-to-market approach focuses on a combination of targeted partnerships, direct engagement with enterprise clients, and leveraging existing networks and referrals. We identify high-value industries and use case opportunities for our AI- and data-driven products, while also offering skilled resources to integrate seamlessly into client projects. Our strategy emphasizes early adoption, quick value delivery, and building long-term, trust-based relationships with clients. |
Our vision is to become a global leader in AI- and data-driven solutions, empowering businesses to operate smarter, faster, and more efficiently. We aim to continuously innovate, expanding our product portfolio and services to transform complex challenges into scalable, high-impact outcomes for enterprises worldwide. |
KeeHoo is a legally registered private limited company, incorporated in 2021, fully compliant with all applicable corporate and regulatory requirements. |
Currently bootstrapped / self-funded, focusing on building multiple products and achieving revenue from market-fit offerings. |
Yes |
Yes |
During the programme, KeeHoo aims to accelerate growth by refining our products, expanding our client base, and leveraging mentorship and industry connections to scale our AI- and data-driven solutions. We also seek strategic partnerships and insights to strengthen our market fit and drive measurable business impact. |
Virtual |
Yes |
Core engine |
KeeHoo’s AI architecture is designed for scalability, flexibility, and real-time insights. It consists of three main layers:
Data Layer – Ingests structured and unstructured data from multiple sources, including databases, APIs, and streaming data, with ETL pipelines for cleaning, transformation, and storage in secure data warehouses or lakes.
AI & Processing Layer – Core layer for model training, inference, and analytics. Includes:
Large Language Models (LLMs) for NLP and conversational AI
Computer vision and predictive models for specialized use cases
Automated pipelines for continuous learning and model updates
Application & Delivery Layer – Exposes AI insights via APIs, dashboards, and low-code platforms. Integrates seamlessly with client systems and supports secure, real-time decision-making.
The architecture emphasizes modularity, security, and scalability, allowing rapid deployment of AI solutions while ensuring compliance and data privacy. |
KeeHoo leverages curated datasets collected and refined through our products and solutions, giving us proprietary insights into business processes, operational patterns, and AI model performance. This unique dataset allows us to train more accurate models, accelerate solution deployment, and deliver insights that are tailored and actionable for our clients - creating a competitive advantage beyond standard public data sources. |
KeeHoo’s AI component is defensible due to a combination of:
Proprietary Data & Insights – Leveraging curated datasets from our deployed solutions that competitors do not have access to.
Custom Model Pipelines – Tailored AI pipelines optimized for specific business use cases, not off-the-shelf models.
Continuous Learning & Improvement – Automated retraining and fine-tuning that adapts to evolving data and client needs.
Seamless Integration – Ability to embed AI into client systems and workflows, making switching costly and complex.
End-to-End Solution – Combining AI with low-code delivery and analytics creates a full-stack, client-specific advantage.
This combination ensures our AI is not just functional, but strategically unique and hard to replicate. |
KeeHoo evaluates AI model performance and reliability using a combination of standard and use-case-specific metrics:
Accuracy & Precision/Recall – For classification and prediction tasks.
F1 Score & ROC-AUC – To balance precision and recall and assess overall model quality.
Perplexity & BLEU/ROUGE – For NLP tasks and language model evaluation.
Mean Absolute Error (MAE) / Root Mean Squared Error (RMSE) – For regression tasks.
Latency & Throughput – Measuring real-time responsiveness and scalability.
Robustness & Drift Monitoring – Tracking model behavior over time to detect performance degradation or data drift.
Regular monitoring, automated retraining, and human-in-the-loop validation ensure models remain accurate, reliable, and aligned with business objectives. |
KeeHoo ensures strict data privacy and security through:
Regulatory Compliance – Adhering to GDPR, HIPAA (where applicable), and client-specific policies.
Data Encryption – Encrypting data at rest and in transit using industry-standard protocols.
Access Control – Role-based access and least-privilege principles to limit data exposure.
Secure Development & Storage – Hosting on compliant cloud platforms with regular security audits.
Anonymization & Masking – Protecting sensitive information in datasets and AI training pipelines.
Continuous Monitoring – Ongoing auditing and vulnerability management to prevent breaches.
This approach ensures client data is handled securely, responsibly, and in full compliance with regulations. |
KeeHoo’s products are designed to comply with applicable data privacy, security, and AI-related regulations. While evolving AI and data laws could introduce new compliance requirements, we proactively monitor regulatory changes and incorporate safeguards - such as data anonymization, access controls, and secure deployment practices - to minimize risk and ensure continued product compliance. |
If KeeHoo’s AI system were scaled 10×, potential challenges could include:
Compute & Storage Load – Increased demand on GPU/CPU resources and data storage, requiring optimized cloud scaling and distributed architectures.
Latency & Throughput – Real-time response times may be impacted without horizontal scaling and load balancing.
Data Pipeline Bottlenecks – ETL and preprocessing pipelines may need parallelization to handle larger volumes.
Model Retraining & Monitoring – Continuous learning pipelines would need automation enhancements to maintain performance.
These are expected engineering challenges rather than failures. Our architecture is designed to scale, and proactive measures—cloud auto-scaling, microservices, and distributed processing - can address these issues while maintaining reliability and performance. |
Yes, KeeHoo has a dedicated in-house AI/ML team with expertise across multiple domains:
Natural Language Processing (NLP) – Conversational AI, summarization, document understanding, and large language models (LLMs).
Computer Vision – Image and video analysis, object detection, and predictive modeling.
Predictive & Prescriptive Analytics – Regression, classification, forecasting, and optimization models.
Model Deployment & MLOps – Scalable pipelines, continuous retraining, monitoring, and integration into client systems.
Data Engineering & Feature Engineering – Building robust data pipelines, feature extraction, and preprocessing for high-quality model inputs.
This team allows KeeHoo to develop, deploy, and maintain AI solutions end-to-end, ensuring high performance, reliability, and customization for client requirements. |
KeeHoo uses a combination of proprietary, client-provided, and licensed datasets depending on the project:
Proprietary Data – Curated and anonymized data collected through our products and internal processes, giving unique insights for AI model training.
Client-Provided Data – Data supplied by clients for project-specific solutions, handled securely under NDA and compliance requirements.
Licensed / Open-Source Data – Trusted third-party datasets for training models where additional domain knowledge or scale is required.
All datasets are used in accordance with licensing agreements, privacy regulations, and security standards, ensuring legal and ethical compliance. |
KeeHoo continuously improves AI model performance through:
Automated Monitoring – Tracking key metrics, detecting drift, and flagging performance degradation in real time.
Regular Retraining – Updating models with new, high-quality data to maintain accuracy and relevance.
Human-in-the-Loop Feedback – Incorporating expert validation and corrections to refine predictions.
Hyperparameter Tuning & Experimentation – Iteratively optimizing model parameters and architectures.
Feature Engineering & Data Enrichment – Continuously enhancing input data to improve predictive power.
This structured approach ensures models remain accurate, reliable, and aligned with evolving business needs. |
KeeHoo is building solutions for both India and global markets. Our products and AI-driven services are designed to address challenges across industries worldwide, while also catering to the specific needs of Indian enterprises. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/8S1PH91wkzji-KeeHoo-Capabilites-One-pager-v2.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/8S1PH91wkzji-KeeHoo-Capabilites-One-pager-v2.pdf |
http://In%20progress |
Yes, KeeHoo is a mission-driven and impact-focused startup. We build scalable product ventures across FaithTech, Enterprise Automation & Intelligence, and Digital Psychology, aiming to create meaningful solutions that improve lives, optimize business operations, and enhance human well-being.
Our mission is to harness AI and data-driven technology to solve real-world challenges while delivering measurable social and business impact, combining innovation with purpose in every product we develop. |
KeeHoo is committed to diversity and inclusion within our team and operations. While the company itself is not formally classified as part of an underrepresented group, we actively promote equitable opportunities and strive to build a diverse workforce across all levels of the organization. |
NA |
checked |
| Apr 26, 2026 @ 9:39 AM |
Saveetha Meganathan and Satyaprakash Pandey |
saveetha.meganathan@gmail.com |
https://www.linkedin.com/in/saveetha-ph-d-226323242/ |
https://sukhapradalifesciences.lovable.app/ |
9741631609 |
Dr. Saveetha Meganathan – Co-founder (Strategy & Health Systems Translation) Holds a PhD in Sociology from the University of Delhi, with deep field experience in maternal and child health, HPV vaccination programs, and women's health across life stages — specifically with vulnerable and underserved communities in India. Her role in AURA encompasses strategic direction, clinical and social contextualisation of the platform, stakeholder and policy engagement, and translating complex technology into real-world, population-relevant use cases. She is the conscience of the venture — ensuring the product remains not just technically sound but ethically grounded and community-appropriate. Dr. Satyaprakash Pandey – Co-founder (Technology & Product R&D) Holds a PhD in Biology from CSIR–IGIB, with postdoctoral experience at IIT Bombay, ERIBA (Netherlands), and the University of Würzburg (Germany). He has industry experience as a Scientist at Tata Medical and Diagnostics. His expertise spans molecular biology, molecular diagnostics, real-time PCR, and CRISPR diagnostics. He leads core technology development, assay and platform architecture, R&D execution, and diagnostics-to-product translation — essentially building the scientific engine that powers AURA. |
Dr. Saveetha and Dr. Satyaprakash met at TIGS — Tata Institute for Genetics and Society, Bengaluru, where both were engaged in translational research. TIGS operates with a strong mandate to bridge laboratory science with societal impact through its translational engine, and it was precisely this shared philosophy — that innovation must reach and benefit real communities — that brought the two together. Working within TIGS's ecosystem exposed both founders to the structural gaps in women's diagnostics in India: brilliant science that never reached the women who needed it most.
Their collaboration grew organically from that shared frustration and shared conviction. Having worked together within this environment, they bring not just complementary skills but a tested working relationship built on mutual trust, scientific rigour, and a deeply aligned social mission. AURA is a direct product of the translational philosophy: take cutting-edge biology, embed it in social context, and design it to actually work in the field — not just the lab. |
2 |
The team's greatest strength is its rare and deliberate complementarity across the two axes that most health startups fail on: scientific credibility and social embeddedness.
Most femtech ventures are either technically sophisticated but socially tone-deaf (urban-centric, stigma-blind), or socially aware but lacking the deep diagnostics expertise to build something clinically meaningful. Sukhaprada bridges this gap by design.
Dr. Pandey brings the hard science — indigenous diagnostic development, molecular assay architecture, and a track record of building cost-effective, deployable tools — ensuring AURA is not just an app but a genuine diagnostic intelligence system grounded in validated biology.
Dr. Meganathan brings the social architecture — an understanding of why women in India don't seek care, how stigma operates across life stages, and how public health systems can be engaged as partners rather than bypassed. She ensures the platform is trusted, used, and impactful at the community level.
Together, they embody the principle that technology without context is just a product, but technology with context becomes infrastructure. Their shared origin in TIGS's translational mission means they have been building toward real-world impact from day one — not pivoting toward it after the fact. |
Sukhaprada Life Sciences |
https://sukhapradalifesciences.lovable.app/ |
Bengaluru |
Sukhaprada Life Sciences is a Life-Course AI Health Intelligence Platform that converts women's body signals into personalized, actionable diagnostic pathways — from adolescence through menopause. |
Women's healthcare in India is fundamentally fragmented, reactive, and late. Symptoms like fatigue, menstrual irregularities, pelvic pain, and pregnancy complications are routinely under-diagnosed because no integrated system exists to monitor, interpret, or act on them longitudinally. The scale of failure is stark: India accounts for 27% of global maternal deaths despite representing only 18% of global births. PCOS affects nearly 1 in 5 Indian women. Anemia affects over 50% of women of reproductive age. Yet diagnostics, health apps, and clinical systems all operate in silos — offering episodic, reactive care with no memory of a woman's health history across life stages. Existing solutions compound the problem rather than solve it. Symptom-tracking apps (Flo clones) offer transactional logging with no biological validation. Deep-tech diagnostics are high precision but narrow in focus and urban-only. Public clinics are overcrowded, lack privacy, and are limited to basic antenatal care. The result: symptom normalisation, social stigma, and missed early diagnosis — year after year, across every stage of a woman's life. The opportunity lies not in another point solution, but in continuous, proactive, life-course monitoring — something no platform currently delivers at scale for Indian women. |
AURA — Adaptive, Unified, Risk-Responsive Analytics — is a life-course women's health intelligence platform that integrates body signals, clinical diagnostics, and AI-driven predictive analytics across every major life stage: adolescence (10–19), reproductive years (20–35), pregnancy, and post-partum & midlife.
At its core, AURA operates as a Health Navigation Engine with three layers:
Input Layer (Raw Signals): Women log menstrual flow, pain levels, mood patterns, sleep data, energy, and blood pressure — passively and actively, on mobile.
Intelligence Engine (AURA AI): Rather than static symptom tracking, AURA's signal interpretation engine maps patterns across five core diagnostic axes — Hormonal-Metabolic (PCOS, LH/FSH/AMH), Nutritional-Haematologic (iron deficiency, anaemia via Ferritin+CRP), Inflammatory-Pain (endometriosis via menstrual fluid diagnostics), Placental-Vascular (preeclampsia risk via sFlt-1/PlGF ratios), and Infectious-Oncologic (self-collected HPV DNA screening).
Output Layer (Actionable Pathways): AURA generates a personalised 6-month health vision board with pattern insights, preventive guidance, and diagnostic referral pathways — helping women navigate care proactively, not reactively.
The platform is built as an app for mobile, and integrates self-sampling diagnostic kits with partner lab networks — creating a unified physical-digital preventive health ecosystem accessible to both urban and peri-urban women. |
AURA's defensibility rests on five compounding moats that no existing competitor combines: 1. Life-Course Longitudinal Intelligence No femtech platform in India currently tracks and interprets a woman's health across all life stages in a single continuous data thread. AURA's "Sukhaprada Data Thread" creates a longitudinal health record from menarche to menopause — a dataset that becomes more predictive and more valuable with every year of use. This 30+ year user retention window creates exceptional Lifetime Value (LTV) that transactional apps cannot match. 2. Multi-Modal Signal-to-Diagnostic Engine Unlike apps that log symptoms and stop there, AURA's AI engine maps signals across five validated biological axes to suggest specific diagnostic pathways. This moves AURA from a "tracker" to a genuine "health intelligence system" — a fundamentally different and harder-to-replicate product category. 3. Integrated Diagnostics Ecosystem AURA combines the digital layer (app, AI) with the physical layer (self-sampling HPV kits, portable Ferritin+CRP sensors, non-invasive Hb sensors) — closing the loop from signal detection to diagnostic confirmation. Competitors are either purely digital or purely diagnostic; AURA is both. 4. Decentralised, Rural-Ready Architecture Positioned uniquely on the competitive map as the only platform offering both an Integrated Ecosystem and Decentralised Access, AURA is designed to work beyond urban clinics — reaching the 40% of rural women who live more than 1km from a healthcare facility. This rural whitespace ($1B potential) is currently uncontested. 5. Scientifically Credentialed Founding Team The team's deep expertise — molecular diagnostics (CRISPR, real-time PCR, point-of-care panels) combined with 18+ years of public health field experience across vulnerable communities — means the platform is built on biologically validated signal interpretation, not algorithmic guesswork. This scientific credibility is not easily replicated by app-first competitors. Defensibility compounds over time: every woman who joins AURA contributes longitudinal health data that trains more accurate predictive models, creates stronger population health insights for institutional partners, and deepens the platform's clinical evidence base — creating a data moat that strengthens with scale. |
MVP |
Pilots |
AURA serves two distinct customer segments:Primary (B2B Institutional) Private hospitals and gynaecology clinics seeking AI-powered preventive monitoring tools to enhance patient outcomes; preventive health and diagnostics networks looking to expand women's health offerings; and corporate wellness programs responding to growing D&I mandates around women's health at the workplace.Secondary (B2G & B2B2C) Government health systems operating under NHM and RMNCH+A programs, where AURA's longitudinal monitoring can plug directly into existing maternal health infrastructure; CSR and public health initiatives targeting underserved women; and insurance and population health analytics providers who need longitudinal women's health data for risk modeling.End User (B2C) Women aged 10–45 across urban, peri-urban, and eventually rural India — specifically those managing PCOS, endometriosis, anaemia, pregnancy risk, or navigating menopause — who currently have no integrated platform to make sense of their symptoms across life stages. |
$10 Billion 620 million women in India represent the full addressable population for women's health products and services. India's femtech and women's diagnostics market is valued at $10B, growing at a 15.6% CAGR driven by rising awareness, digital adoption, and increasing prevalence of conditions like PCOS and anaemia. |
$2.5 Billion Urban empowered women with digital access and financial independence (54% mobile penetration, 57% with Prime+ credit scores, 78.6% with bank accounts) combined with the rural whitespace opportunity — $1.5B urban spend + $1B rural potential — form our near-term reachable segment. |
~$2.2 Billion Specific high-value segment opportunities we can capture in the near term: IVF services ($1.2B) and fertility apps ($1B), where AURA's hormonal-metabolic and longitudinal tracking capabilities are directly applicable and where willingness to pay is already demonstrated. |
AURA operates a diversified, high-LTV revenue model across four streams, designed so that a user acquired at adolescence generates revenue across 30+ years: 1. B2C Subscriptions (Freemium → Premium) The AURA app launches free with core signal logging. Premium tiers unlock AI-driven pattern insights, personalised diagnostic pathway recommendations, and the 6-month health vision board. This is the primary user acquisition and retention engine. 2. B2B Device & Platform Sales Modular maternal monitoring hardware and diagnostic panel access sold to private clinics, gynaecology practices, and diagnostic networks. Clinics pay for the admin dashboard, patient monitoring infrastructure, and integrated reporting tools. 3. Direct Product Sales (D2C Diagnostics) At-home HPV self-sampling kits, non-invasive haemoglobin sensors, and portable Ferritin+CRP test kits sold directly to consumers or through clinic partnerships — converting digital risk flags into physical diagnostic action. 4. Financing, Risk & Institutional Analytics IVF "Protection Fee" models and risk underwriting for insurance providers using AURA's longitudinal health data. Population health analytics packages sold to insurers, public health bodies, and research institutions. Key metric: High Lifetime Value via 30+ year retention. A woman onboarded at 15 through adolescent anaemia screening remains a user through reproductive years, pregnancy, and menopause — across every revenue stream. |
AURA competes across multiple segments, but no single competitor spans all of them: SegmentGlobalIndianDigital ClinicsMaven, Advantia Health, MidiProactive for Her, Savemom, Elda Health, GytreeApps & TrackingFlo (Unicorn), Clue, Natural CyclesHealofy, Pinky Promise, iMumzDeep Tech / DevicesVayyar, Elvie, WillowNiramai (breast cancer), Janitri, InitoHygiene / Products—Pee Safe, Sirona, NuaSupplements—Gynoveda, The Good Bug, Hera Sukhaprada's competitive edge: Every competitor listed above is siloed — focused on just one condition (Niramai: breast cancer only), one life stage (Clue: cycle tracking), or one modality (Inito: fertility hormones). None offers life-course longitudinal intelligence combined with decentralised diagnostic access. On the competitive positioning map, Sukhaprada occupies the upper-right quadrant alone — the only platform combining an Integrated Ecosystem with Decentralised Access. Flo/Clue sit urban-integrated but lack diagnostics. Public health systems sit decentralised but offer no integrated ecosystem. That white space is AURA's territory. |
AURA uses a hub-and-spoke acquisition model that builds trust before asking for behaviour change: B2B First, B2C Pull Institutional partners — clinics, diagnostic labs, corporate HR teams — are the primary acquisition channel. When a gynaecologist recommends AURA or a corporate wellness program onboards it, individual women follow. This dramatically lowers B2C CAC and builds clinical credibility simultaneously. Community Health Worker (CHW) Networks For peri-urban and rural reach, AURA works through ASHA workers, ANMs, and NGO field staff who already have trusted relationships with women in underserved communities. This channel is government-aligned and low-cost. Stigma-Free Digital Content Organic content addressing normalised but undiagnosed conditions — PCOS fatigue, endometriosis pain, pregnancy risk — builds awareness among women who don't yet know they have a diagnosable condition. This is particularly powerful for the 20–35 reproductive years segment where symptom normalisation is the primary barrier. Government Program Integration Alignment with Janani Suraksha Yojana (JSY) and RMNCH+A creates pathways for AURA to be embedded within existing maternal health touchpoints — reaching women at scale with zero incremental CAC. Corporate Wellness Rising D&I mandates mean corporates are actively seeking credible women's health solutions. AURA's employer dashboard and anonymised population health reports make it a compelling corporate benefit — acquired through HR and wellness procurement channels. |
AURA follows a phased, clinically validated GTM that de-risks adoption at each stage: Phase 1 — 0 to 12 Months: Validate & Anchor (Urban & Peri-Urban) Deploy pilot-ready MVP across 2–5 clinical and diagnostic partners in Bengaluru. Target 500–1,000 women users. Focus use cases: anaemia detection and maternal risk monitoring, where clinical evidence is strongest and institutional buy-in is most accessible. Generate real-world outcome data to build the evidence base for scale. Phase 2 — 12 to 18 Months: Expand & Monetise (Metros + Tier 2) Expand to other Indian metros and Tier 2 cities through hospital networks and corporate wellness programs. Activate first B2B revenue streams. Launch D2C diagnostic kit sales alongside app premium subscriptions. Onboard dedicated engineering lead to support platform scaling. Phase 3 — Post-Validation: Integrate & Scale (Public Health + Rural) Integrate with NHM and RMNCH+A programs to reach underserved populations through government health infrastructure. Partner with NGOs and CSR initiatives. Introduce population health analytics and insurance risk modeling products for institutional revenue at scale. GTM principle: India-first, evidence-led, institution-anchored. We do not go wide before we go deep. Every phase is gated by clinical validation data from the previous phase, ensuring we scale what works rather than what sounds good.URA's long-term vision is to become India's foundational health intelligence infrastructure for women — the equivalent of a lifelong health record, diagnostic navigator, and preventive care companion rolled into one platform. In 10 years, we envision a world where: Every woman in India — urban or rural, teenage or menopausal — has access to a continuously learning health intelligence system that knows her biological history, anticipates her risks, and guides her to the right care before a crisis develops. AURA's longitudinal dataset — built across 30+ year user relationships spanning adolescence through menopause — becomes the largest and most clinically validated women's health dataset in India, enabling breakthroughs in population health research, insurance risk modeling, and public health policy that are not currently possible. The platform's diagnostic ecosystem evolves from point-of-care panels and self-sampling kits today into a fully integrated physical-digital health network — where AURA's AI flags a risk, a community health worker delivers a test kit, a partner lab confirms the result, and a clinician acts — all within a seamless, privacy-first workflow. Internationally, once the India model is validated, AURA scales to other high-burden, underserved women's health markets across South and Southeast Asia and Sub-Saharan Africa — geographies with comparable diagnostic gaps, digital adoption curves, and unmet need. The ultimate measure of success is not revenue or users — it is a measurable reduction in India's maternal mortality rate, a decline in late-stage PCOS and endometriosis diagnoses, and millions of productive years of life restored to women who currently spend 25% of their lives in poor health because the system failed to see them early enough. |
URA's long-term vision is to become India's foundational health intelligence infrastructure for women — the equivalent of a lifelong health record, diagnostic navigator, and preventive care companion rolled into one platform.
In 10 years, we envision a world where:
Every woman in India — urban or rural, teenage or menopausal — has access to a continuously learning health intelligence system that knows her biological history, anticipates her risks, and guides her to the right care before a crisis develops.
AURA's longitudinal dataset — built across 30+ year user relationships spanning adolescence through menopause — becomes the largest and most clinically validated women's health dataset in India, enabling breakthroughs in population health research, insurance risk modeling, and public health policy that are not currently possible.
The platform's diagnostic ecosystem evolves from point-of-care panels and self-sampling kits today into a fully integrated physical-digital health network — where AURA's AI flags a risk, a community health worker delivers a test kit, a partner lab confirms the result, and a clinician acts — all within a seamless, privacy-first workflow.
Internationally, once the India model is validated, AURA scales to other high-burden, underserved women's health markets across South and Southeast Asia and Sub-Saharan Africa — geographies with comparable diagnostic gaps, digital adoption curves, and unmet need.
The ultimate measure of success is not revenue or users — it is a measurable reduction in India's maternal mortality rate, a decline in late-stage PCOS and endometriosis diagnoses, and millions of productive years of life restored to women who currently spend 25% of their lives in poor health because the system failed to see them early enough. |
Private Limited |
NA |
Yes |
We are applying to Avinya because it is the only accelerator in Karnataka's ecosystem that treats AI not as a feature to bolt on, but as the foundational layer of a venture's architecture — and that is precisely how we have built AURA. AURA is not a health app that uses AI for recommendations. It is an AI-driven signal interpretation engine that maps women's biological patterns across five diagnostic axes to generate clinically meaningful, actionable health intelligence. The distinction matters, and Avinya's selection criteria reflect an understanding of that distinction. Programmes that evaluate "AI startups" by surface-level metrics would miss what we are building. Avinya's emphasis on problem significance, real-world impact, and responsible innovation means we are evaluated on the dimensions that actually define AURA's value. Three specific aspects of Avinya drew us in: The Karnataka ecosystem fit is structural, not incidental. Karnataka is home to some of India's most advanced biotech, diagnostics, and digital health infrastructure. Our pilots are designed to run in Bengaluru first. Our institutional partnerships — clinics, diagnostic labs, corporate wellness programs — are being built here. Being embedded within the IIT alumni network, government IT/BT support, and ACSEL's industry connections means Avinya doesn't just offer us mentorship in the abstract — it offers us proximity to the exact stakeholders we need to convert from conversations into signed pilots. The founder-first philosophy addresses a real gap we face. Building a women's health platform in India is not just a technology challenge. It is a sustained exercise in navigating stigma, institutional resistance, regulatory complexity, and the emotional weight of working in a space where the stakes are lives, not metrics. The fact that Avinya integrates leadership coaching and founder well-being into its structure signals that it understands what it actually takes to build in this space — and that it invests in founders as people, not just their cap tables. The ethical innovation mandate aligns with our non-negotiables. AURA handles deeply sensitive health data — menstrual patterns, pregnancy status, HPV risk, hormonal profiles — from some of the most vulnerable women in India. We have built privacy-first by design, not as a compliance checkbox. Avinya's emphasis on ESG integration, diversity and inclusion, and long-term societal impact means we will be in a cohort that holds us to the same standards we hold ourselves to, and that will push us to embed these values more deeply as we scale. We are also applying because we are at exactly the stage where Avinya is most valuable. We have a validated problem, a scientifically grounded solution, a functional prototype, and a founding team with the domain depth to execute. What we need now is structured support to convert that foundation into a pilot-ready product with clinical partners, institutional credibility, and a clear path to first revenue. That is precisely the transition Avinya is designed to accelerate. |
We enter Avinya with four specific, measurable goals — not aspirations, but milestones we will hold ourselves accountable to across the 16 weeks: 1. Secure at least two confirmed clinical or diagnostic pilot partners Our most urgent need is converting warm institutional conversations into signed pilot agreements. We are targeting 2–5 partners — gynaecology clinics, diagnostic networks, or corporate wellness programs in Bengaluru — to deploy AURA with 100–300 women users during or immediately after the programme. Avinya's ecosystem access and mentor introductions are the most direct path to compressing what would otherwise be a 6–12 month business development cycle into weeks. 2. Achieve a pilot-ready, clinically validated MVP Our Flutter app and signal interpretation architecture exist. What we need to validate before pilot deployment is the AI engine's diagnostic pathway mapping against real clinical feedback — ensuring that AURA's outputs align with what gynaecologists and diagnosticians would actually act on. We want to exit Avinya with a product that has been reviewed, stress-tested, and refined by clinical advisors, not just built by us. 3. Define and stress-test our regulatory and IP strategy AURA sits at the intersection of a health app, a diagnostic support tool, and a data platform — a combination that creates genuine regulatory complexity under India's Digital Personal Data Protection Act, CDSCO medical device guidelines, and emerging AI-in-healthcare frameworks. We want Avinya's mentors and legal resources to help us map the regulatory landscape clearly, identify what we need to comply with before we scale, and protect our core innovation — particularly the signal interpretation engine and the menstrual fluid diagnostics approach — through indigenous IP. 4. Build investor readiness for a seed round The ₹10L seed from Avinya marks the start of our external funding journey, not the end. We want to exit the programme with a data room, a refined financial model, a clear articulation of our unit economics, and warm introductions to investors aligned with HealthTech and deep-tech at the seed stage. Avinya's investor matching in Phase 2 is a key reason we believe the programme can genuinely accelerate our fundraising timeline rather than just prepare us for it. Beyond these four milestones, there is something harder to quantify but equally important: we want to emerge from Avinya as sharper founders. We are two scientists and public health professionals who have chosen to build a company. The structured mentorship, peer learning from the cohort, and leadership coaching that Avinya offers will help us develop the founder muscle — in sales, in storytelling, in managing uncertainty — that no PhD programme or research career prepares you for. We believe the best outcome of Avinya is not just a better product, but two founders who are genuinely ready to lead AURA through its next chapter. |
Yes |
Yes |
Core engine |
AURA's AI architecture is designed as a multi-layer health signal interpretation system built around three functional components:
Layer 1 — Signal Ingestion & Structuring
Women log health signals through the mobile app — menstrual flow patterns, pain intensity and location, mood states, sleep quality, energy levels, and blood pressure. These are captured as structured time-series data, stored in a compliant PostgreSQL database, and normalised across life-stage context (adolescent vs. reproductive vs. pregnant vs. post-partum) before any interpretation occurs. Life-stage context is critical because the same signal — fatigue, for instance — carries entirely different diagnostic weight depending on whether the user is 16, 28, or 34 weeks pregnant.
Layer 2 — Signal Interpretation Engine (The Core AI)
This is AURA's primary innovation. Rather than rule-based symptom matching, the engine uses a pattern-to-pathway mapping model that correlates longitudinal signal combinations against five diagnostic axes:
Hormonal-Metabolic (PCOS, anovulation, thyroid disruption)
Nutritional-Haematologic (iron deficiency anaemia, B12 deficiency)
Inflammatory-Pain (endometriosis, pelvic inflammatory disease)
Placental-Vascular (preeclampsia, gestational hypertension)
Infectious-Oncologic (HPV risk, STI probability)
At current stage, the engine operates on a clinically validated rule-and-weight system — built from published clinical guidelines (FOGSI, WHO, ACOG) and co-designed with medical advisors — that scores signal combinations against each diagnostic axis and generates a ranked probability output. This is the honest description of where we are: the foundation is clinical logic codified into an interpretable model, not a black-box neural network. This is deliberate. At TRL-3 with limited real-world data, interpretability and clinical trust matter more than model complexity. As pilot data accumulates, the engine is designed to transition toward supervised machine learning models — initially logistic regression and gradient boosting for tabular health signal data, eventually toward sequence models (LSTMs or Transformers) for longitudinal pattern detection across menstrual cycles and life-stage transitions.
Layer 3 — Output & Personalisation Engine
Interpreted signals are converted into three output types: pattern insights (what your body is telling you), diagnostic pathway suggestions (what to test for and why), and the 6-month personalised health vision board (what to work toward). The vision board uses a goal-setting logic layer that weights outputs by user-stated priorities — energy, hormonal balance, fertility, stress — creating a personalised experience that improves with continued engagement.
Infrastructure: Android and ioS App (mobile), PostgreSQL (compliant data storage), cloud-hosted APIs, with an admin dashboard for clinical partners. Designed to be modular — each diagnostic axis can be updated independently as clinical evidence or data improves. |
We will be direct: at this stage, our data advantage is structural and prospective, not yet accumulated.
What we have today is a clinically grounded data architecture — designed from the ground up to capture the signals that matter diagnostically, not just the signals that are easy to log. Most femtech apps collect symptom logs as engagement features. AURA collects signals as diagnostic inputs — the schema, the signal taxonomy, and the life-stage context layer are all designed around clinical utility. This means that when data accumulates, it will be the right kind of data.
What we are building toward is a proprietary longitudinal dataset that will be genuinely rare and hard to replicate:
Depth: Multi-signal data (menstrual, pain, mood, sleep, BP) combined with diagnostic confirmation data (lab results from partner diagnostic networks) creates labelled training data — signal input mapped to confirmed clinical outcome — that cannot be scraped from the internet or purchased from a data broker.
Continuity: 30+ year user relationships mean AURA will eventually hold the only longitudinal women's health dataset in India that tracks the same individuals across adolescence, reproductive years, pregnancy, and menopause. This is structurally impossible for competitors who acquire users at a single life stage.
India-specificity: Indian women's health patterns differ from Western populations in clinically significant ways — higher anaemia prevalence, distinct PCOS phenotypes, different nutritional profiles, different stigma and care-seeking behaviours. A model trained on Indian data will outperform one trained on global data for Indian women. No global competitor has this data. Most Indian competitors are not collecting it with diagnostic intent.
The honest near-term position: Our first pilot with 500–1,000 women will be the beginning of this data moat, not its realisation. We are transparent about this with partners and advisors. The advantage compounds with time and scale — which is why the 30+ year LTV model and early institutional pilot strategy are both data strategy decisions, not just commercial ones. |
Defensibility in AI comes from four sources: data, architecture, clinical validation, and switching costs. AURA builds moats across all four:
Data moat (long-term, compounding): As described above, labelled longitudinal Indian women's health data with diagnostic confirmation is not replicable quickly. Every year of operation deepens it. Every pilot partner who contributes diagnostic confirmation data strengthens it.
Architectural moat (multi-axis, not single-condition): Every deep-tech competitor in women's health — Niramai, Inito, Janitri — is optimised for one condition or one life stage. Their AI is narrow by design. AURA's five-axis diagnostic architecture means the system sees signal interactions that single-condition models structurally cannot — for example, the co-occurrence of anaemia signals and menstrual irregularity that may indicate both iron deficiency and a hormonal disorder simultaneously. This cross-axis pattern recognition is AURA's core IP and requires both the data architecture and the clinical knowledge to build correctly.
Clinical validation moat: As pilot data generates outcome evidence — signal patterns that predicted confirmed diagnoses — that evidence becomes a regulatory and trust asset. Health systems and clinicians do not switch platforms once a tool has demonstrated clinical reliability in their setting. First-mover advantage in institutional clinical validation is durable.
Switching cost moat: A woman who has logged 18 months of health signals on AURA, built a diagnostic history, and received clinical referrals through the platform has a longitudinal health record she cannot take elsewhere. This is not data lock-in by design — it is the natural consequence of building something genuinely useful over time.
IP protection: We are pursuing indigenous IP protection for the signal interpretation engine and the menstrual fluid diagnostics methodology for endometriosis — a novel diagnostic approach with no direct equivalent in current Indian IP landscape. |
Our evaluation framework operates at two levels — clinical and technical — because a women's health AI that is technically accurate but clinically misleading is worse than no AI at all.
Clinical Performance Metrics
Sensitivity and Specificity per diagnostic axis — particularly for high-stakes pathways (preeclampsia risk, HPV, anaemia) where false negatives carry serious clinical consequences. We prioritise sensitivity in early iterations, accepting lower specificity to avoid missed diagnoses.
Positive Predictive Value (PPV) for diagnostic pathway suggestions — what proportion of women flagged for a pathway actually receive a confirming diagnosis through partner labs.
Clinical concordance rate — agreement between AURA's signal interpretation outputs and the assessment of gynaecologist or clinical advisors reviewing the same case. This is our primary validation metric during pilot phase.
Technical Performance Metrics
Signal completeness rate — proportion of users logging sufficient signal data for meaningful interpretation (a model is only as good as its inputs).
Pathway activation rate — proportion of users for whom the engine generates a specific diagnostic pathway recommendation vs. a generic insight, tracking whether the signal collection design is working.
Longitudinal consistency — whether the engine's outputs for the same user remain coherent and directionally stable across sessions, or show unexplained variance.
Reliability & Safety Guardrails
All diagnostic pathway outputs are framed as navigational suggestions, not clinical diagnoses — explicitly communicated in the UI and the terms of service.
A confidence threshold system ensures the engine only surfaces a pathway recommendation when signal evidence meets a minimum threshold, rather than generating spurious outputs from sparse data.
Clinical advisor review loop during pilot phase — a sample of flagged pathway outputs reviewed monthly by medical advisors to catch systematic errors before they scale.
As we move toward ML-based models, we will add standard model evaluation metrics: AUC-ROC per diagnostic class, F1 scores, calibration curves, and regular re-evaluation as new training data is added. |
Women's health data is among the most sensitive categories of personal data — and for our users, who may be tracking pregnancy, HPV risk, or reproductive conditions they have not disclosed to family members, a privacy breach is not just a regulatory event. It is a personal harm. We treat this accordingly.
Regulatory Compliance Framework
Digital Personal Data Protection (DPDP) Act 2023: AURA is being built with DPDP compliance as a baseline, not an afterthought. This includes explicit, granular consent for each data category collected, the right to erasure, and clear data principal rights embedded in the user journey.
CDSCO Medical Device Rules 2017: We are actively mapping AURA's diagnostic pathway suggestion feature against the Software as a Medical Device (SaMD) classification framework to determine whether any features require regulatory approval before commercial deployment.
IT Act 2000 and SPDI Rules: Sensitive Personal Data and Information rules apply to health data and govern our storage, processing, and third-party sharing obligations.
Technical Security Architecture
All health signal data encrypted at rest and in transit (AES-256 / TLS 1.3).
PostgreSQL databases deployed on compliant cloud infrastructure with role-based access controls — clinical partners see only their patient cohort, never the broader dataset.
No third-party data sharing without explicit user consent and data processing agreements.
Anonymisation and pseudonymisation protocols for any data used in model training or shared with research partners.
Privacy-by-Design Principles
Minimal data collection — we collect only signals with diagnostic utility, not engagement signals that serve the platform more than the user.
On-device processing for sensitive inputs where technically feasible, reducing cloud exposure.
Transparent data use policy written in plain language, not legal boilerplate — critical for building trust with users who are already navigating stigma around their health. |
Are There Any Regulatory Risks That Could Impact Your Product?
We believe in naming risks honestly rather than burying them. There are three genuine regulatory uncertainties we are actively managing:
1. SaMD Classification Risk
AURA's signal interpretation engine generates health pathway suggestions that could be interpreted as diagnostic decision support — a category that may require CDSCO approval as a Software as a Medical Device. The boundary between a "wellness app" and a "medical device" in Indian regulation is not yet sharply defined for AI-driven health platforms. Our mitigation: we are engaging regulatory consultants during Avinya to map this boundary precisely and ensure our language, UI framing, and clinical claims stay on the right side of it. If approval is required, we will pursue it — it is a moat, not just a burden.
2. DPDP Act Implementation Uncertainty
The DPDP Act 2023 is enacted but key rules and sectoral guidelines — particularly for health data — are still being notified. This creates compliance uncertainty for any health data platform. Our mitigation: we are building to the most stringent plausible interpretation of the Act, so that as rules are clarified, we are ahead of requirements rather than catching up.
3. AI-in-Healthcare Regulatory Evolution
India does not yet have a comprehensive AI-in-healthcare regulatory framework. NITI Aayog's principles exist; binding rules do not. As AURA's AI component becomes more clinically significant, it may attract regulatory attention that does not yet have a clear framework. Our mitigation: proactive engagement with regulatory bodies and alignment with international frameworks (EU AI Act, FDA SaMD guidance) so that we are prepared for whatever framework India adopts.
What we are not worried about: Basic health app operation, data collection with consent, and diagnostic referral pathways are well within existing legal frameworks. The risks above are medium-term scaling questions, not immediate blockers. |
This is the question we most appreciate being asked, because it forces honesty about where the brittleness actually lives.
Signal data quality will degrade before model performance does. At 1,000 users, we can monitor signal completeness and coach users toward better logging. At 10,000 users, logging drop-off, inconsistent inputs, and population diversity (new geographies, languages, literacy levels, health contexts) will create noise that the current interpretation engine — calibrated on a relatively homogeneous early-adopter population — will not handle well. Mitigation: adaptive onboarding that adjusts signal collection to user capability, and model retraining pipelines that account for population shift.
The clinical validation loop will break if not systematised. Currently, we plan for clinical advisors to review a sample of pathway outputs monthly. At 10x users, this manual loop becomes a bottleneck. We need to build automated flagging — identifying outputs that fall outside confidence thresholds and routing them for human review — before we hit that scale.
The PostgreSQL architecture will hit performance limits. Our current database design is appropriate for a pilot of hundreds to low thousands of users. At 10x, time-series health data volume, concurrent API calls, and longitudinal query complexity will require migration to a purpose-built time-series database and a properly sharded architecture. This is a known and solvable engineering problem — but it requires the dedicated engineering lead we are actively planning to hire.
Diagnostic axis coverage gaps will become visible. At small scale, users who fall outside our five diagnostic axes simply receive general wellness guidance. At 10x, the volume of such users will be large enough to matter clinically and commercially. We need expansion of the axis framework and escalation pathways for signals the engine cannot interpret.
The honest summary: Data quality, clinical review scalability, and infrastructure are the three failure modes. All are anticipated and addressable with the right hires and architecture decisions in the next 18 months. |
Currently, AURA's AI architecture and clinical logic have been designed by the founding team — specifically Dr. Satyaprakash Pandey, whose background in molecular diagnostics and biological signal interpretation provides the scientific grounding for the diagnostic pathway model. The clinical logic encoding was done in close consultation with gynaecology and public health advisors.
We do not yet have a dedicated ML engineer — and we are transparent about this. The current engine is a clinically validated rule-and-weight system that does not yet require deep ML infrastructure to operate. This was a deliberate decision: building interpretable clinical logic first, and transitioning to ML-based models once we have the labelled training data to justify it, is more responsible than building a complex model on insufficient data.
Our hiring roadmap is specific: We are using Avinya and the seed funding to onboard one dedicated engineering lead with experience in health data systems and ML model development. The profile we are targeting is someone with experience in tabular health data modelling (not computer vision or NLP — the wrong ML backgrounds for what we are building) and familiarity with India's health data landscape.
The founding team's strength is that the clinical and biological intelligence required to build a meaningful women's health AI is not something an ML engineer brings — it has to come from domain experts. We have that. What we are adding is the technical execution capacity to operationalise it at scale. |
Currently used for model design and clinical logic calibration:
Published clinical guidelines from FOGSI (Federation of Obstetric and Gynaecological Societies of India), WHO, and ACOG — publicly available, used to establish diagnostic thresholds and signal weightings.
Peer-reviewed literature on Indian women's health epidemiology — PCOS prevalence, anaemia burden, preeclampsia risk factors, HPV incidence — used to calibrate population-level priors in the interpretation engine.
De-identified aggregate data from public health surveys (NFHS-5, DLHS) — publicly available government datasets used to understand baseline health signal distributions in Indian women.
None of our current datasets are proprietary. This is the honest position at TRL-3, and it is why the pilot data strategy is so critical — our first 500–1,000 users, consented to research use of their data, will be the foundation of our first proprietary training dataset.
Pilot data ownership: All user data collected through AURA is owned by Sukhaprada Life Sciences under the terms of our privacy policy and DPDP-compliant consent framework. Users consent to anonymised use of their data for model improvement. Diagnostic confirmation data from partner labs will be acquired under data processing agreements that explicitly vest training rights in Sukhaprada.
No third-party licensed datasets are currently in use. We have evaluated commercial health data providers but have not found India-specific women's health datasets of sufficient quality or relevance to justify licensing costs at this stage. |
Our model improvement strategy follows a validate-then-automate approach — manual and high-touch in early stages, progressively systematised as data volume justifies it:
Phase 1 — Pilot (0–12 months): Clinical Feedback Loop
Every diagnostic pathway recommendation generated during pilots is tracked against clinical outcome — did the user seek care, what did the clinician find, did the lab result confirm the pathway flag? This outcome linkage is the labelled dataset that trains the next model iteration. Clinical advisors review a monthly sample of outputs. Model parameters are updated quarterly based on accumulated feedback.
Phase 2 — Early Scale (12–24 months): Semi-Automated Retraining
As labelled data volume grows past statistically meaningful thresholds per diagnostic axis, we transition from rule-and-weight updates to supervised ML model training — logistic regression and gradient boosting first, evaluated against the clinical performance metrics described above. Retraining pipelines run on a defined cadence with human sign-off before any model update goes to production.
Phase 3 — Scale (24+ months): Continuous Learning Infrastructure
Active learning pipelines that identify the highest-uncertainty cases for clinical review and prioritised labelling; A/B testing of model variants against user health outcomes; and federated learning exploration for privacy-preserving model improvement across institutional partner deployments.
Structural improvement mechanisms that are always on:
User feedback signals (did this insight feel relevant?) incorporated as soft labels for relevance scoring.
Signal completeness monitoring to improve data collection design — if a particular signal is consistently missing or inconsistently logged, the onboarding or logging UX is redesigned.
Cohort analysis to identify population segments where model performance diverges — rural vs. urban, different age cohorts, different health literacy levels — and targeted data collection to address gaps. |
India first. Emphatically and strategically.
This is not a limitation — it is a deliberate choice grounded in both opportunity and competitive logic.
India represents a $10B women's health market growing at 15.6% CAGR, with 620 million women, 350 million of reproductive age, and some of the world's highest burdens of PCOS, anaemia, and maternal mortality. The problem we are solving is most acute here, most urgent here, and most underserved here. No global competitor has built for this population with the depth, cultural specificity, and diagnostic grounding that the problem requires. That is our home ground advantage, and we intend to own it before expanding.
Building for India also means building AI that works for India — models trained on Indian health signal distributions, calibrated against Indian clinical guidelines, tested in Indian healthcare contexts, and designed for Indian infrastructure realities (variable connectivity, multilingual users, community health worker delivery channels). A model trained in San Francisco and deployed in Bengaluru will underperform. We are building the model that outperforms it in Bengaluru first.
Global markets are a phase-three ambition, not a phase-one distraction. The most relevant international expansion markets are those with comparable profiles: high women's health burden, growing digital adoption, and underserved diagnostic infrastructure — South and Southeast Asia (Bangladesh, Indonesia, Vietnam) and Sub-Saharan Africa (Nigeria, Kenya, Ethiopia). These markets share enough epidemiological and infrastructural characteristics with India that a validated Indian model provides a genuine head start, not just a name.
We will know we are ready to go global when we have demonstrated measurable health outcomes in India at scale — reduced late-stage diagnoses, improved maternal risk detection rates, documented reductions in out-of-pocket spend. That evidence, built in India, is what makes AURA credible everywhere else. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/O6RNttb1WXzV-Avinya-ASCEL-30th-April-2026_Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/O6RNttb1WXzV-Avinya-ASCEL-30th-April-2026_Pitch-Deck.pdf |
https://drive.google.com/file/d/1YNZl7MdwO9rKvkSwPymzAvQKg3gciDNo/view?usp=sharing |
Yes — and we want to be precise about what that means, because "mission-driven" has been diluted by overuse. So here is what our mission actually costs us and why we would not build any other way.
Dr. Saveetha has spent 18 years working directly with women in underserved communities — sitting with women who normalised debilitating pain for years because every system around them said it was just part of being a woman, and documenting maternal deaths that were preventable if a single risk signal had been caught weeks earlier. Dr. Satyaprakash built diagnostic technologies and watched scientifically robust tools fail to reach the populations who needed them most because the translation infrastructure between lab and community did not exist.
AURA was not conceived in a boardroom. It was built from the accumulated weight of those experiences. We met at TIGS — an institution whose core mandate is translational science that reaches society — and that philosophy is embedded in everything we build.
The mission places non-negotiable constraints on how we build:
We build for the hardest-to-reach women, not just the easiest to acquire. Our architecture — decentralised access, community health worker integration, government program alignment — is designed from day one for women beyond the reach of a gynaecologist. If AURA only ever serves urban, digitally fluent women, we have built a good product. We will not have fulfilled the mission.
We hold privacy as a dignity issue, not a compliance checkbox. In India's social context, a data breach for a woman tracking her menstrual patterns or HPV risk can have consequences for relationships and safety. We build privacy-first because our users deserve it.
We measure success in health outcomes, not engagement metrics. The metric that matters is whether a woman received an earlier diagnosis, whether a maternal death did not happen because preeclampsia was detected at 24 weeks instead of 34.
We could build AURA exclusively for urban, high-willingness-to-pay users and show cleaner early metrics. We are not doing that. Building for rural and peri-urban women means harder infrastructure, messier early financials, and a slower clinical validation cycle. We have made that choice with full awareness of what it costs, because a women's health AI that skips clinical validation is not mission-driven — it is mission-branded.
There is a cleaner, narrower version of this business. We are not building it — not because we are indifferent to commercial success, but because we started from the problem, not the market opportunity.
Women in India spend 25% of their lives in poor health due to compounding systemic neglect. 75 million years of life are lost annually to the women's health gap. That is what we woke up to every morning.
AURA is our answer. The mission is not what we added to make the business meaningful. The business is what we built to make the mission possible. |
Yes — in more than one dimension, and we think it matters to name them specifically. Dr. Saveetha Meganathan is a woman founder in deep-tech and health innovation — a space where women-led ventures receive a disproportionately small share of venture funding globally, and even less so in India. She holds a PhD in Sociology, not engineering or medicine — which means she brings a discipline that is almost entirely absent from the founding teams of technology health startups, yet is arguably the most important one when building for underserved women in India. Her expertise in how communities actually behave, why stigma persists, and how trust is built in low-resource settings is not a soft complement to the technology — it is load-bearing. That kind of founder is genuinely rare in Indian deeptech. Dr. Satyaprakash Pandey comes from a non-IIT, non-metro academic background — CSIR-IGIB and postdoctoral stints built through scientific merit rather than elite institutional networks. In an ecosystem where founder credibility is often shorthand for pedigree, researchers who have built in government science institutions and translated that work into deployable products represent an underrepresented and undervalued founder profile. Together, we are a two-person founding team building at the intersection of women's health, molecular diagnostics, and AI — a combination that does not fit neatly into any single investor category, which has meant navigating a funding landscape not designed with us in mind. The community we are building for is itself the most underrepresented group in health innovation. Women in rural and peri-urban India — who bear the highest burden of anaemia, maternal mortality, undiagnosed PCOS, and cervical cancer — are almost entirely absent from the user research, clinical trials, and product design processes of the health technology industry. Building with them, not just for them, is both our mission and our method. Avinya's explicit commitment to diversity, inclusion, and founders from underrepresented backgrounds is one of the reasons this programme felt like the right fit. We are not looking for accommodation. We are looking for an ecosystem that recognises that the most important problems are often being solved by the founders least likely to be in the room — and that actively works to change that. |
We came across Avinya through our own research into Karnataka's AI startup ecosystem — specifically while exploring accelerators aligned with responsible AI, health innovation, and founder-first support structures. As a Bengaluru-based venture embedded in Karnataka's science and technology community, ACSEL and its connection to the IIT Alumni Centre Bengaluru were already on our radar. When Avinya was announced, the programme's specific emphasis on real-world impact, ethical AI, and support for underrepresented founders made it immediately relevant to where we are and what we are building. |
checked |
| Apr 25, 2026 @ 6:08 PM |
Ashwin Satheesh; Cyriac T Meppuram |
ash566301@gmail.com |
https://www.linkedin.com/in/ashwin-satheesh-a5a087228 |
https://assistrend.com/ |
9495279233 |
CEO, a passionate community architect with a 3years of experience in building and managing communities. |
How we connected is a small but meaningful story.
I'm a bit of a LinkedIn addict — I personally reach out to almost all my connections. Around four months into my startup journey, I sent Cyriac a connection request. I followed up with a message introducing myself, and to my surprise, he responded quickly and even joined my early-stage community.
A few months later, we had our first call. I tried to share my vision — but it was still raw and evolving. What was supposed to be a short elevator call but turned into a one-hour deep conversation.
At the end of it, Cyriac patiently said,
“Okay, Ashwin ithrem neram paranjath njan churukki parayaam…”
(“Let me sum up everything you’ve said…”)
And then he gave me the one keyword that captured the heart of what I was trying to build:
Mental Wellness.
And that's how this partnership began — grounded in clarity, compassion, and a shared vision. |
two |
The founders are Ashwin from India, with 5 years of experience in building and managing student communities across kerala, and Mr. Cyriac, a co-founder and venture growth expert with 20 years of experience in the wellness ecosystem.
Along with the founders, other key players include:
Isaac, founder of global suicide prevention squad and impact thousands of life within 20 years, who collaborate to lead our wellness layer.
Arun Luiz, a respected professor with a decade of experience in institutional ecosystems, who leads our vendor relations.
Ajith S, with 25 years in software architecture, mentors our dev pipeline and ensures we build for scale
Febin S, with 4 years of experience as an ethical penetrator and manages our digital security systems.
We've already shipped in community building, ML workflows, wellness mapping, app development and security. Which will make us confident in executing our vision.
Partnership:
ASSISTREND has initiated strategic partnerships with local academic institutions across Kerala and Chennai, wellness ventures like Wellness Junxion, and student influencers such as KTU Special to strengthen its outreach and impact. These partnerships have successfully helped us gain and manage a growing community of users. They also support the integration of wellness experts into the platform's intervention loop, ensuring that users receive credible and empathetic guidance. With ongoing dialogues with university wellness cells, ASSISTREND is actively building a trusted ecosystem focused on proactive emotional well-being for Gen Z. |
Assistrend Ai Private Limited |
https://assistrend.com/ |
Kerala |
To address youth well-being imbalance-where countless wellness apps fail due to low retention, data scarcity—introducing an adaptive companion application which acts like a purpose-driven wellness companion, and proposes first ever rapid approach-switching capability, that proactively supports youth mental well-being, tackles loneliness and emotional disconnect before it escalates |
Millions of students today are struggling to cope with the pressures of daily life challenges, creating a silent epidemic around the globe, the Mental Well-being Imbalance. Leading to loneliness, hopelessness, and, at its worst, contributing to the rising numbers of suicides and crimes — a tragic symptom of a system that fails to address human pain. In India, 1–1.5 million engineering graduates are produced every year across 7,000+ colleges. One in three students shows symptoms of anxiety or depression, yet only 5–15% seek professional help due to stigma and access gaps. This creates a large, underserved need right at the campus level. Globally, there are 200 million+ higher-education students, and student mental health has become a priority worldwide. The global wellness market is already $20–28 billion, growing at 28% CAGR, and projected to cross $80 billion by 2032. Between 2018–2022, India saw a 44% rise in mental-health–related suicides (21 per 100,000), and 16% of high-risk youth account for 54% of violent-crime arrests. Studies show that young people use mental-health apps mainly during crisis moments, not as part of daily life. Today, most students cope in silence, lean on a few close friends but still face personal stigma, or turn to mental health apps that are focused primarily on reactive like therapy, counselling and semi proactive like meditation, sound frequency based tools, have low potential in user mapping cause of the absence of personalization, led by lack of indepth user context, without clear user context, any digital solution can't be generate highly personalised intervention. This leads to one-size-fits-all suggestions, low retention and high churn — leaving students without consistent, meaningful support, especially in moments when they need it most. |
ASSISTREND addresses this by introducing a purpose driven wellness companion.
The frontend serves as a purpose-driven community app — much like an everyday social app where users can share their content and engage with posts from the people they follow, while the backend powers a wellness-focused intelligent agent system that provides meaningful support, even when users may not be fully aware of their challenges. Together, they create a unified, purpose-driven wellness companion platform.
In which the Purpose Layer — preparing every youth to discover, pursue, and achieve their dreams. On that journey, many will inevitably face depression, anxiety, or emotional imbalance from various factors in there daily life. That's where our Wellness Layer comes in — proactively addressing these challenges, not as a separate add-on, but as an integrated part of their personal growth.
Why it works:
Our eudaimonic well-being framework aligns with PYD's "5 Cs" model. http://exploresel.gse.harvard.edu/frameworks/52 |
Our application brings the first-ever rapid approach-switching ability, ensuring that every day, users receive the right kind of support—whether it's reactive care in moments of crisis, proactive guidance to prevent struggles, or other approaches that help them navigate life's journey with resilience and hope. How it works: Its architecture is powered by five intelligent modes—Reactive, Proactive, Semi-Proactive, General, and Context. Here, an agent called the Determiner analyzes the user's query to identify the most appropriate approach at that moment. The companion then acts accordingly to fulfill the user's needs. UVP: The Proactive companionship mode! It deeply maps users, forecasts their journey, identifies risks early, and suggests personalized steps to mitigate them. |
two |
one |
Genz (Engineering Students) |
₹4,672.8 crores |
₹3,273.6 crores |
₹70 lakh |
Freemium, Target Advertising |
Wysa, Headspace, Calm |
Existing Community onboarding. |
We scale through community-led growth: activate 10K users, then 90k organic, then expand across Kerala colleges, then penetrate South India with campaigns and roll out nationally via influencers. |
Milestones:
Month 1: Onboard our existing community users in whatsapp on our product.
Month 2: Auto post generator after our podcast feature.
Month 3: 1,200 paying users, 30%+ month-2 retention validated, ₹1.2L MRR
Month 4: Utilize organic outreach of over 1Lakh engineering students from our past influencer page.
Month 5: Introducing fully fledged companion based on memorag and fine tuning pre trained open source model that enables forecasting, predicting and mitigation steps suggestion ability.
Month 6: 6 institutional partnerships signed, 3500 users, ₹1.8L monthly revenue
Our vision is to create a future where mental-health intervention is not delayed or stigmatized but seamlessly woven into students daily digital lives — helping them thrive academically, socially, and emotionally.
Here, we aim to a future where every young person can discover, pursue, and achieve their dreams—supported by timely, personalized guidance that integrates mental wellness as a natural part of that journey. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 25, 2026 @ 4:58 PM |
Rakesh Adepu |
RAKESHTGR2@GMAIL.COM |
https://www.linkedin.com/in/rakesh-adepu-437866371/ |
https://ekithab.com/ |
07660065073 |
co-founder venu and ui & ux rajukumar |
all are college friends |
2 |
Our biggest strength is the combination of a builder's mindset and technical depth operating simultaneously.
Rakesh, our founder, built eKithab AI from zero — no funding, no coding background — by learning every tool needed and shipping fast. He brings relentless product execution, user empathy, and go-to-market instinct born from selling digital products since 2022.
Venu, our co-founder, brings rare technical credibility — a Samsung ISWDP semiconductor fellowship, three concurrent VLSI internships, and a 9.83 GPA — proving he executes at the highest level under pressure.
Together we move fast without breaking things. One founder who has already proven he can build a market from nothing. One engineer who has proven he operates at the level of the world's best chip companies. That combination — street-smart execution plus deep technical rigour — is what makes us different from every other student startup team. |
ekithab ai |
https://www.linkedin.com/in/rakesh-adepu-437866371/ |
Hyderabad |
eKithab AI turns any idea into a professional, sell-ready eBook in 10 minutes — built for India's 700 million vernacular creators |
Creating and selling a professional eBook in India today requires 4-6 different tools, 3-5 days of work, and design skills most creators don't have. A fitness coach with 10 years of expertise spends 3 days switching between ChatGPT for content, Canva for design, Grammarly for editing, and Gumroad for selling — before earning a single rupee. Most give up halfway. The problem is not lack of knowledge. It is the broken, expensive, and time-consuming production process that stops India's creators from monetising what they know. India has 700 million vernacular internet users, 40 million+ knowledge creators, and 1.5 lakh coaching centres — yet zero AI publishing tool built for them in their language, at their price point, for their market. eKithab AI solves this by replacing the entire workflow — content generation, quality scoring, cover design, regional language support, and business insights — in one platform, in 10 minutes |
eKithab AI is a complete AI publishing platform that takes a creator from idea to published, sell-ready eBook in 10 minutes — without design skills, coding, or multiple tools.
Here is how it works:
A creator enters any topic in any language. Our AI instantly generates a fully structured eBook with professionally written chapters, optimised for their chosen category — whether self-help, finance, fitness, cooking, or 11 other categories. Every chapter is automatically scored from 0 to 100 by our AI quality engine, which detects weak content, flags vague sentences, and suggests specific fixes before the creator publishes.
The platform then generates a professional AI cover using DALL-E 3, converts the eBook into an audiobook using ElevenLabs voices in Hindi, Telugu, and Tamil, and produces an AI Business Insights report telling the creator exactly who to sell to, what price to charge, which platforms to use, and which Instagram hooks will drive sales.
Everything — creation, quality control, cover design, audiobook, business strategy — happens inside one platform. No Canva. No ChatGPT switching. No 3-day production cycle.
Creators download their finished eBook as PDF or DOCX with full commercial rights and sell it immediately on Gumroad, Instagram, WhatsApp, or Amazon KDP.
Starting at ₹99 for 50 credits. 15 free credits with no card required.
Current traction: 1,300+ registered users, ₹50,000+ bootstrapped revenue, 127 eBooks generated in a single 24-hour peak — zero external funding |
Here's your answer! 🎯 What makes your solution unique or defensible? eKithab AI is defensible on five levels that no competitor can replicate quickly or cheaply. First — AI Quality Scoring Engine. We are the only AI publishing platform in the world with a built-in quality scoring system that evaluates every chapter from 0 to 100, detects weak content, and suggests specific fixes before the creator publishes. Jasper, Automateed, Designrr, and ChatGPT all generate content and stop there. We generate and then guarantee quality. This single feature changes the product category from a content generator to a publishing intelligence platform. Second — Indian Language Moat. We generate natively structured content in Telugu, Hindi, Tamil, Kannada, and Malayalam — not just translation, but culturally appropriate content with Indian examples, references, and audience context. Building this properly requires months of prompt engineering and deep understanding of each language's audience. No Silicon Valley competitor will invest engineering resources here. No Indian competitor has done it yet. This moat grows stronger every month we operate. Third — Complete Workflow in One Platform. Every competitor solves one piece of the publishing problem. ChatGPT writes. Canva designs. Grammarly edits. Gumroad sells. We do all of it — generation, quality scoring, cover design, audiobook narration, business insights, and selling — without the creator leaving the platform. The switching cost of rebuilding a 5-tool workflow is enormous. Once a creator builds their publishing process on eKithab AI they have no reason to leave. Fourth — India-Specific Business Intelligence. Our AI Business Insights feature tells creators exactly who to sell their eBook to, what price to charge in INR, which Indian platforms to use, and which Instagram hooks will drive sales in the Indian market. No global tool understands the Indian creator monetisation landscape well enough to give this advice. We do. Fifth — Creator Behaviour Data Flywheel. Every eBook generated, every quality score issued, every edit made, and every sale completed on our platform teaches us what works for Indian creators. This behavioural dataset — which we own entirely — becomes the foundation for fine-tuning our own publishing-specific AI models. The longer we operate, the smarter our platform becomes, and the harder it is for any new entrant to catch up without the same years of Indian creator data. |
Revenue |
Revenue |
eKithab AI serves three distinct customer segments across B2C and B2B markets. Primary — Individual Creators (B2C) Coaches, educators, fitness trainers, finance influencers, and content creators who have knowledge and want to monetise it through digital products. Typically aged 18-35, active on Instagram and YouTube, earning less than ₹50,000/month and looking for passive income streams. They know their subject deeply but lack design skills, time, and tools to package it professionally. Secondary — Students and Educators (B2C) College students creating study guides, engineering notes, and exam preparation material to sell to classmates. Teachers and coaching centre faculty who need structured digital study material quickly and affordably. India has 250 million students and 1.5 lakh coaching centres — all underserved by existing tools. Tertiary — EdTech Platforms and Institutions (B2B) EdTech startups, coaching institutes, and universities needing bulk AI-generated regional language content at scale. These clients pay ₹20,000–50,000/month for API access and white-label solutions. One B2B client generates more revenue than 100 individual subscribers. Geographic Focus: India-first — specifically Tier 1 and Tier 2 cities with high creator economy activity — Hyderabad, Bangalore, Chennai, Mumbai, Delhi, Lucknow, Jaipur. |
Global AI Publishing Market: $41.2B by 2033 (CAGR 30.8%) + Global Self-Publishing Market: $52.3B by 2032 (CAGR 17%) + India Creator Economy: $61.87B by 2033 (CAGR 22.4%) + India EdTech Content Market: $6.5B by 2030. Combined TAM: $100B+ by 2033. |
AI-powered content creation and self-publishing tools targeting India, Southeast Asia, Middle East, and Africa. Focus on vernacular internet users, knowledge creators, coaching centres, and EdTech platforms. SAM estimate: $8–12B by 2033. |
Realistic capture of 1–3% of SAM over 8–10 years = $80M–$360M annual revenue. At 10–15x revenue multiple = $800M–$5.4B valuation. Near-term SOM: 2,000 paying creators + 20 B2B EdTech clients = ₹5 Lakh MRR within 12 months. |
Five revenue streams. B2C credit packs — ₹99 (50 credits), ₹199 (150 credits), ₹349 (400 credits) — credits never expire. B2C monthly subscriptions — ₹299/month, ₹799/month, ₹1,999/month. B2B EdTech API — ₹20,000–50,000/month per enterprise client for bulk regional language content generation. Creator marketplace commission — 15% on every eBook sold through eKithab Store. AI Business Insights reports — 5 credits per premium report. Current: 1,300 users, ₹50,000+ bootstrapped revenue, zero external funding. |
Jasper AI ($49–69/month) — generates marketing copy but no book formatting, no cover, no quality scoring, no Indian languages. Designrr ($29/month) — formats existing content but has zero AI generation capability. Automateed ($149 one-time) — most complete global competitor but no Indian language support, no quality scoring, no B2B API. ChatGPT Plus ($20/month) — generates text but requires Canva, Grammarly, and 3 days of manual work after. eKithab AI is the only platform combining creation, quality intelligence, regional language support, and business insights in one place for the Indian market. |
Organic Instagram content — before/after quality score demonstrations and creator success stories drive signups at near-zero CAC. Currently 13,100+ followers with zero paid marketing spend. Direct founder outreach — personal DMs to creators in niche communities. WhatsApp creator groups and regional language communities. Referral program — give 5 credits, get 5 credits. College ambassador network across Hyderabad institutions. B2B — direct outreach to coaching centres and EdTech platforms with live product demos. |
Phase 1 (Now–Month 6): Product-led growth. Free 15 credits on signup. Viral loop via "Made with eKithab AI" watermark on free downloads. Instagram Reels targeting Indian coaches, educators, and creators. Reach 2,000 paying creators. Phase 2 (Month 6–18): Channel expansion. Product Hunt launch targeting 5,000+ signups. Partnership with Notion Press (30,000–40,000 books/year). B2B pilot with 10 Hyderabad coaching centres. PR in Inc42, YourStory, Economic Times. Phase 3 (Month 18–36): Enterprise and international. EdTech API for Unacademy, PhysicsWallah tier platforms. Expansion to Southeast Asia, Middle East, and Africa replicating India playbook with local language and payment localisation. |
eKithab AI's vision is to become India's knowledge marketplace — the Canva of publishing for Bharat. In 10 years, any person in India with knowledge and ambition — a yoga teacher in Vijayawada, a finance coach in Lucknow, a school teacher in Madurai — can turn what they know into a professional product, sell it globally, and earn passive income in their own language. We are building the infrastructure for India's 700 million vernacular internet users to participate in the global creator economy on equal terms. Long-term we target 5 million creators across 20 countries, $100M ARR, and a category-defining position as Asia's leading AI publishing platform — the company that proved world-class software can be built from Hyderabad for Bharat and the world |
NA |
NA |
Yes |
We are applying to Avinya because it is one of the few accelerators in India that understands what it actually takes to build an AI product — not just fund ideas, but develop founders who can execute responsibly and at scale. Three specific reasons bring us here. First, the IIT alumni network and structured mentorship. eKithab AI is built by a non-technical founder using AI tools. Our biggest gap is not the product — it is access to senior technical and product minds who can help us architect the next version properly. Avinya's IIT alumni expertise and customised learning tracks are exactly what we need at this stage. Second, the seed funding of ₹10 lakhs unlocks our single most critical next step — hiring our first engineer to move off no-code infrastructure onto our own scalable architecture. This one hire changes everything. We have proven product-market fit with 1,300 users and ₹50,000 bootstrapped revenue. We just need the infrastructure to scale what we already validated. Third, Avinya's emphasis on responsible entrepreneurship, ESG integration, and mental health support for founders resonates deeply with how we want to build. We are not chasing valuation. We are building India's knowledge infrastructure for 700 million vernacular internet users — a mission that requires the kind of long-term, values-driven support Avinya offers. We are an AI-first startup at MVP stage with real traction, two full-time founders, and a clear path to ₹5 Lakh MRR. Avinya is the right program at the right moment. |
We want to leave Avinya in 16 weeks as a fundamentally stronger company — not just better funded, but better built. Specifically we have four goals for the programme. First — Technical architecture. We want to work with Avinya's IIT alumni mentors to design and begin executing our migration from Emergent AI to our own scalable AWS-hosted infrastructure. Right now our platform breaks on complex features because no-code tools have hard limits. We need senior technical guidance to architect the right stack before we hire our first engineer. By week 16 we want a clear technical roadmap and deployment-ready architecture. Second — First 100 paying subscribers. We have 1,300 registered users and only a handful paying. Avinya's structured mentorship on product-market fit, conversion optimisation, and growth strategy should help us unlock what is blocking paid conversion. Our goal is to exit the programme with 100+ active paying subscribers and a clear repeatable acquisition playbook. Third — First B2B client. We want to close our first coaching centre or EdTech platform as a paying B2B client during the programme. Avinya's network of industry experts and investors can open doors that cold outreach cannot. One signed B2B client at ₹20,000/month changes our entire fundraising story. Fourth — Seed round readiness. We want to use the 16 weeks to build a fundable company — proper Pvt Ltd registration, DPIIT recognition, clean cap table, investor-grade metrics dashboard, and a pitch that gets us to pre-seed conversations immediately after the programme ends. By the end of Avinya we want to be a company that any early-stage investor looks at and says — this team knows exactly what they are building, w |
BOTH |
Yes |
Supporting feature |
eKithab AI uses a multi-layer AI architecture built on third-party API integrations orchestrated through Emergent AI's managed runtime. The content generation layer uses OpenAI GPT-4 for structured eBook generation — we pass category-specific system prompts that constrain output into chapter-by-chapter formatted content with consistent tone, reading level, and structure per genre. The image generation layer uses DALL-E 3 for AI cover design, with style prompts calibrated to 15 eBook categories. The audio layer uses ElevenLabs Text-to-Speech API for audiobook narration with Indian voice personas in Hindi, Telugu, and Tamil. The quality scoring engine runs a secondary GPT-4 call that evaluates each chapter against five parameters — content depth, sentence clarity, structural flow, specificity of examples, and reading level appropriateness — returning a 0-100 score with actionable fix suggestions. The payment layer uses Razorpay API supporting UPI, PhonePe, Google Pay, and Net Banking. Our roadmap includes migrating from Emergent AI to a self-hosted stack on AWS with fine-tuned open-source LLMs (Llama or Mistral) for cost efficiency and quality improvement at scale. |
Our current proprietary data advantage is behavioural — we have accumulated real usage patterns from 1,300 Indian creators generating eBooks across 15 categories and 22 languages. This gives us unique insight into which prompts produce the highest quality output per category, which chapter structures resonate with Indian audiences, which quality score thresholds correlate with successful sales, and which language combinations are most requested by vernacular creators. As our marketplace grows, we will accumulate sales data showing which eBook topics, price points, and categories perform best in the Indian market — data no global competitor is collecting. Long-term this creator behaviour dataset becomes the foundation for fine-tuning our own publishing-specific language models that outperform generic LLMs on structured Indian-language content generation. |
Three layers of defensibility. First — category-specific prompt architecture. Our system prompts are not generic. Each of our 15 eBook categories has a custom prompt stack tuned for that genre's structure, tone, and Indian audience expectations. This took months of iteration and produces output quality that generic ChatGPT prompting cannot replicate without the same tuning investment. Second — quality scoring moat. Our AI quality scoring engine is a feature no competitor has built for the Indian publishing market. It creates a feedback loop where creators improve their content before publishing, generating better outcomes and higher platform loyalty. Third — Indian language depth. We do not just translate — we generate natively structured content in Telugu, Hindi, and Tamil with culturally appropriate examples and references. Global players like Jasper and Automateed will not invest engineering resources to build this for India's vernacular market, giving us a durable first-mover moat. |
We currently evaluate model performance across five metrics. Output quality score — every generated eBook is scored by our own quality engine immediately after generation, giving us an internal benchmark of average output quality per category per day. User edit rate — we track how frequently users use Regenerate, Expand, and Simplify buttons after generation, which signals dissatisfaction with initial output. Generation success rate — we monitor API call completion rates and failure rates to detect reliability issues. User retention after first eBook — if a user creates a second eBook within 7 days it signals the first output met expectations. Credit consumption per eBook — unusually high credit usage indicates users are regenerating repeatedly due to poor initial output. Our roadmap includes A/B testing different prompt architectures per category and building a formal human evaluation pipeline where sample outputs are rated by domain experts before prompt changes go live. |
User eBook content is processed through OpenAI and ElevenLabs APIs under their standard data processing agreements, which include commitments not to use API-submitted content for model training by default. We do not store raw user content on our servers beyond what is necessary for the active session. User account data including email and payment information is managed through Emergent AI's secure managed runtime with standard encryption at rest and in transit. Payment data is handled entirely by Razorpay, a PCI-DSS compliant payment processor — we never store card or UPI credentials. We are currently in the process of registering as a Pvt Ltd company to formalise our compliance framework, after which we will apply for DPIIT recognition and implement a formal privacy policy, data retention policy, and terms of service aligned with India's Digital Personal Data Protection Act 2023 |
Three regulatory risks exist. First — India's Digital Personal Data Protection Act 2023. As DPDPA implementation progresses, we will need to ensure explicit consent collection, data localisation compliance, and a formal grievance officer appointment. We are building this into our Pvt Ltd registration process. Second — AI-generated content regulation. India and global markets are moving toward disclosure requirements for AI-generated content. We are proactively adding "Created with eKithab AI" attribution to all generated eBooks, positioning this as a feature rather than a compliance burden. Third — copyright and intellectual property risk. AI-generated content ownership remains legally ambiguous in India. We address this by granting users full commercial rights to their generated content and maintaining clear terms of service that establish user ownership of eBook output. We monitor evolving IP law developments and will update our legal framework accordingly. |
Honest answer — several things. First — API cost structure. At 10x users our OpenAI GPT-4 API costs will scale linearly while our credit pricing may not cover margins. We need to implement a dual-model strategy using GPT-3.5 Turbo or fine-tuned open-source models for drafting and GPT-4 only for final quality passes. Second — generation queue management. Currently eBook generation is synchronous. At 10x concurrent users we will hit timeout failures and race conditions without a proper async job queue system such as Redis or AWS SQS. Third — Emergent AI infrastructure limits. Our current no-code runtime was not designed for high concurrency. A 10x scale event would likely cause platform-level failures. This is the primary reason AWS migration is our top post-funding priority. Fourth — quality consistency. At scale, prompt drift and model version changes from OpenAI can silently degrade output quality. We need automated regression testing on sample outputs to catch quality degradation before users experience it. |
We do not currently have a dedicated AI/ML engineer. Our AI capabilities are delivered through API integrations architected and prompt-engineered by our founding team. Rakesh, our founder and AI Product Manager, has hands-on experience designing prompt systems, evaluating output quality, and building feature workflows using GPT-4, DALL-E 3, and ElevenLabs APIs. Venu, our co-founder, brings deep technical foundation from three concurrent semiconductor engineering internships including DakshinSilicon, MasterVLSI, and Codec Technologies, and a Samsung ISWDP fellowship — giving him the systems-level thinking needed to architect our infrastructure migration. Our first hire post-funding will be a full-stack engineer with ML integration experience. Avinya's mentorship network is one of the reasons we applied — access to IIT alumni with AI and ML expertise would directly accelerate our in-house capability building. |
We use three categories of data. First — OpenAI and ElevenLabs training data via API. We consume their models through commercial API agreements and do not own or control the underlying training datasets. These are licensed through our API subscription agreements. Second — user-generated content. Every eBook created on eKithab AI is owned by the user per our terms of service. We do not claim ownership of user content and do not use it for model training without explicit consent. Third — our own behavioural dataset. We own and control our platform usage data — generation patterns, quality scores, user edit behaviour, credit consumption, and category preferences across 1,300 users. This proprietary behavioural dataset is not licensed from any third party. It is our first-party owned asset and forms the foundation of our future fine-tuning strategy. As we grow our marketplace, eBook sales performance data will become a second owned dataset giving us unique market intelligence on Indian creator monetisation patterns. |
Five-part improvement roadmap. First — prompt iteration based on quality scores. We analyse average quality scores per category weekly and refine system prompts for underperforming categories. Second — user edit signal analysis. High Regenerate or Expand button usage on specific chapter types signals prompt weaknesses we can fix. Third — A/B testing prompt architectures. We will systematically test prompt variations on sample users and measure quality score improvement before rolling out platform-wide. Fourth — fine-tuning open-source models. Post-funding we will fine-tune Llama or Mistral on high-quality Indian publishing content — public domain books, educational material, and user-consented eKithab outputs — creating a proprietary model that outperforms generic LLMs on structured Indian content generation at lower API cost. Fifth — human evaluation pipeline. We will recruit 10 domain expert evaluators across key categories — fitness, finance, education, cooking — to rate sample outputs monthly and provide structured feedback that feeds directly into prompt improvement cycles. |
India first. Global next. Deliberately in that order. India is not our fallback — it is our strategic moat. We are building the deepest, most defensible AI publishing platform for India's 700 million vernacular internet users, 40 million knowledge creators, 45,000 colleges, and 1.5 lakh coaching centres. No global competitor — Jasper, Automateed, or Designrr — will invest in Telugu, Hindi, and Tamil native generation, INR pricing, UPI payments, and India-specific content categories. We dominate this market first. By Year 3 we expand to Southeast Asia — Indonesia, Philippines, Vietnam — replicating the India playbook with local language and payment localisation. By Year 4 we enter the Middle East with Arabic content generation. By Year 5 we launch in the US and UK with premium pricing targeting the 10 million English-language self-publishers globally. India gives us the traction, the moat, and the credibility. The world gives us the scale. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/S7M2qgHRonEl-eKithab_AI_UPGRADED_2026-1.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/S7M2qgHRonEl-eKithab_AI_UPGRADED_2026-1.pptx |
https://ekithab.com/ |
Yes. Deeply and deliberately.
eKithab AI's mission is to democratise knowledge publishing for India's vernacular creators — the yoga teacher in Vijayawada, the finance coach in Lucknow, the school teacher in Madurai, the home chef in Coimbatore — people who have spent years building real expertise but have never had the tools, the design skills, or the technical resources to package and monetise what they know.
The impact we are creating operates on three levels.
Economic empowerment. India has 40 million knowledge creators. The vast majority earn nothing from their expertise because the barrier to creating a professional digital product is too high. A fitness coach who charges ₹500 per session can create one eBook on eKithab AI in 10 minutes and sell it to 1,000 people for ₹299 each — generating ₹2.99 Lakhs from knowledge they already have. We are building passive income infrastructure for people who have never had access to it.Language equity. The global creator economy is dominated by English. Indian creators who think and teach in Telugu, Hindi, or Tamil are systematically excluded from monetisation tools designed for Western markets. Every feature we build — native language generation, regional voice audiobooks, INR pricing, UPI payments — is a deliberate choice to include the 700 million vernacular internet users that the global tech industry ignores. We believe your language should never limit your income.
Education access. When a coaching centre teacher in a Tier 2 city uses eKithab AI to create structured study material in Telugu for their students, they are solving an access problem that the Indian education system has struggled with for decades. Regional language educational content created at scale and low cost changes what is possible for students who cannot afford premium English-medium resources.
This mission is personal for our founder Rakesh. He grew up in a government BC hostel, earned ₹300 a day from catering work to pay hostel fees, and built eKithab AI because he was the exact person the platform is built for. He was not building a product for a market he had researched. He was solving his own problem and then making it available to everyone like him.
We measure our impact not just in revenue but in creators empowered, regional languages supported, and knowledge products published by people who never thought they could be authors.
That is the company we are building. Not just a billion dollar SaaS. A platform that proves where you come from and what language you speak should never limit what you can create |
NA |
NA |
checked |
| Apr 25, 2026 @ 2:45 PM |
Yadla shiva Prasad |
shivayadla66@gmail.com |
|
http://NA |
7075769166 |
Founder |
|
one |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 25, 2026 @ 12:22 PM |
Shahnaz Shaikh |
shahnaz.shaikh@ai-genix.com |
https://www.linkedin.com/in/shahnaz-shaikh-856a86a0/ |
https://ai-genix.net/ |
7021528976 |
Shahnaz Shaikh Founder & CEO, MSc Microbiology Worked as a junior scientist in a bio-fertilizer company for 4 years. 15 years of reach experience in Business management, sales & marketing, etc. Worked as a sales manager in the telecommunications industry. Working on research and development of AI-Enabled crop protection technology since 2014. Khalil Shaikh Co-Founder & CTO, Computer & Ind. Electronics Engineering Held several positions in the telecom, electronics and computer engineering field as a techno-commercial business development professional Worked for Indian and international companies. Working on research and development of AI-Enabled crop protection technology since 2014 |
Our family is involved in the farming. While cultivating various crops on our farm, we faced acute insect-pest infestation and synthetic pesticide toxicity issues with our farm, farm workers, and dairy animals.
These issues made us think over how to solve insect-pest management issues without use of toxic pesticides. We started working on finding alternative solutions to toxic pesticides available in the market.
We realized that there is no viable and efficient solution available in the market for insect-pest management.
We two started working on an environment-friendly, nontoxic, agricultural pest management solution to serve the most deprived global farming community by reducing their input cost of crop protection and improving their living condition by minimizing their exposure to highly toxic chemicals. |
2 |
Our biggest strength is turning deep scientific insight into practical, scalable agricultural solutions—combining R&D, engineering, and real farming experience into systems that work in the field. |
AI-GENIX INTERNATIONAL PVT LTD |
https://ai-genix.net |
Mumbai |
Replacement of chemical pesticides with AI |
Problem Statement (Concise) Agriculture today is heavily dependent on chemical pesticides, leading to: Environmental damage (soil degradation, biodiversity loss, residue buildup) Health risks to farmers and consumers Rising pest resistance, making chemicals progressively less effective At the same time, farmers lack scalable, effective, and non-toxic alternatives for pest and pathogen control. |
AI-GENIX provides a non-chemical, physics-based crop protection platform that replaces pesticides with intelligent electronic systems.
Our solution uses a combination of:
AI-driven pest behavior modulation (light, frequency, and signal-based disruption)
Targeted insect attraction and elimination systems
Tremendously High-Frequency Electromagnetic Waves (THF-EMW) to neutralize pathogens and soft-bodied pests at the DNA/RNA level
This enables broad-spectrum pest and disease control without toxic residues while preserving beneficial insects and improving farm productivity. |
AI-GENIX is defensible through proprietary physics-based pest control, a deep IP stack (frequencies + hardware + AI), zero resistance risk, and a full-stack non-chemical platform that competitors cannot easily replicate. |
Revenue |
Revenue, Pilots, Testimonials |
AI-GENIX targets corporate farms, greenhouse growers, FPOs, government programs, and export-oriented supply chains—starting with high-value, early adopters and scaling to mass agriculture through partnerships. |
9.3 T USD |
1.13 T USD |
86.4 B USD |
AI-GENIX generates revenue through high-value equipment sales, recurring consumables, AMC contracts, and subscription-based crop protection services, supported by distribution and enterprise partnerships. |
Our competitors include chemical giants like Bayer Crop Science and UPL Limited, biological control companies like Koppert Biological Systems, and agtech monitoring platforms like Trapview—however, none offer a scalable, non-chemical, physics-based pest control solution like AI-GENIX. |
AI-GENIX acquires customers through direct enterprise sales and pilot deployments, scales via distributor and government partnerships, and expands revenue through repeat deployments and long-term contracts. |
AI-GENIX enters via pilot deployments with corporate farms, builds credibility through proven ROI, scales through distributor and government channels, and expands revenue through full-farm ecosystem adoption. |
AI-GENIX aims to replace chemical pesticides globally by building an intelligent, non-chemical crop protection infrastructure that enables autonomous, residue-free, and sustainable farming at scale. |
Incorporated in 23.05.2019 |
Bootstrapped |
Yes |
We are applying to Avinya to accelerate the scale of our validated non-chemical crop protection technology through mentorship, strategic partnerships, and institutional support, enabling large-scale impact in sustainable agriculture. |
During the programme, we aim to achieve large-scale validation, secure anchor customers, build a scalable go-to-market engine, and become investment-ready for rapid expansion. |
Virtual Participation |
Yes |
Core engine |
AI-GENIX uses an edge-first, closed-loop AI architecture that combines real-time sensing, on-device intelligence, and adaptive actuation to optimize pest and pathogen control continuously.” |
AI-GENIX has a proprietary dataset combining insect frequency-response behavior, real-time field performance data, and electromagnetic interaction models, creating a compounding data moat that improves system efficacy and is difficult to replicate without large-scale deployments. |
AI-GENIX’s AI is defensible due to its proprietary frequency-response datasets, tight integration with specialized hardware, and closed-loop learning from real-world deployments—enabling adaptive pest control that competitors cannot replicate with software-only approaches.” |
AI-GENIX evaluates performance using field-level pest reduction, yield improvement, and ROI as primary metrics, supported by model accuracy, control effectiveness, and system reliability—ensuring real-world impact rather than just predictive accuracy. |
AI-GENIX ensures data privacy and security through edge-first processing, minimal data collection, end-to-end encryption, strict access controls, and compliance with regulations like GDPR and India’s DPDP Act—while anonymizing data used for AI training. |
AI-GENIX faces moderate regulatory risks mainly around product classification and electromagnetic compliance but avoids heavy pesticide regulations due to its non-chemical nature, with overall regulatory trends favoring sustainable solutions. |
At 10× scale, risks include data drift, poor signal generalization, edge compute limits, fleet management complexity, and closed-loop instability. We mitigate these through region-specific models, edge-first architecture, staged OTA updates, constrained control policies, and robust MLOps with continuous feedback. |
Yes, AI-GENIX has an in-house AI/ML team with expertise in machine learning, signal processing, embedded systems, and entomology, enabling real-time, adaptive pest control through tightly integrated hardware and AI systems. |
AI-GENIX primarily uses proprietary datasets generated from field deployments, including pest behavior, frequency-response, and electromagnetic interaction data, all of which are owned in-house. Public scientific data is used only for reference, and the system is not dependent on licensed datasets. |
AI-GENIX continuously improves model performance through closed-loop learning from field outcomes, reinforcement learning for control optimization, active data selection, and staged deployment—creating a self-improving system that gets better with every deployment. |
AI-GENIX is building for both India and global markets, with India as the primary launch and scale base, and parallel expansion into regions like the Middle East, Africa, and Southeast Asia. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/xuvmHMsrI4Mx-Updated-Investor-Pitch-deck-AIGENIX-INTERNATIONAL-PVT-LTD.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/xuvmHMsrI4Mx-Updated-Investor-Pitch-deck-AIGENIX-INTERNATIONAL-PVT-LTD.pdf |
https://youtu.be/hAmXvUdYLVw?si=u8D55W9rf8lbHnVd |
Yes, AI-GENIX is a mission-driven startup focused on eliminating chemical pesticides and enabling sustainable, residue-free agriculture while improving farmer income and environmental outcomes. |
Yes, we come from a farming background, which is underrepresented in deep-tech startups. Our firsthand experience with pesticide-related challenges shaped the foundation of AI-GENIX and gives us a unique, ground-level perspective on solving real agricultural problems. |
ACSEL Technology Forum |
checked |
| Apr 25, 2026 @ 2:16 AM |
Founder - Pratik Raghuwanshi | Co-Founder - Akhilesh Raibhog |
pratik@kshaminnovation.in |
https://www.linkedin.com/in/pratik-raghuwanshi/%20%20 |
https://kshaminnovation.in |
+91 9373124725 |
As CEO, Pratik oversees product design, development, fund raising and strategic outreach, while Akhilesh, as COO, manages software, firmware, and technical operations. We have clearly defined roles but all strategic decisions are made through mutual discussion and consultation with our core team. This collaborative approach ensures that every decision is vetted from both a technical and operational perspective, keeping the entire team aligned and motivated toward our objective of making quality hearing care accessible. |
We first met during our Diploma in Electronics, where we collaborated on numerous technical projects throughout the course. We have worked together for eight years. This early experience of working together laid the foundation for our professional synergy, allowing us to combine our individual technical strengths toward a shared vision. |
All |
Our biggest strength as a team lies in our strong combination of deep technical expertise epically in electronic hardware development and real-world domain understanding of the hearing care ecosystem. Our team has built strong capabilities in circuit and PCB design for advanced audio systems, embedded firmware development, and AI/ML integration. Our in-house expertise spans across DSP-based speech processing, mobile application development and mechanical design through 3D modeling and rapid prototyping, enabling us to build and iterate complex hardware-software systems end-to-end.
Through continuous interactions with 50+ audiologists and real users, we have developed a deep understanding of audiological challenges, user behavior, and gaps in existing solutions. Our team is further strengthened by guidance from an experienced ENT specialist & Audiologist combined having over four decades of clinical experience, along with audiology advisors who provide critical inputs for validation and product development. This combination of engineering depth and clinical insight allows us to build solutions that are both technologically robust and clinically relevant. |
Ksham Innovation Private Limited |
https://kshaminnovation.in |
Amravati, Maharashtra |
We are a deeptech assistive technology startup building the Able Glasses, India's first non-surgical bone conduction smart aid glasses for people with hearing loss and Able Assistant App to make hearing care affordable, stigma-free and accessible for all. |
India faces a silent epidemic with over 63 million individuals suffering from hearing loss, yet More than 90% of hearing impaired individuals do not use hearing aids due to stigma, high device costs & lack of awareness. This gap is most severe for individuals with Conductive and Mixed Hearing Loss, where standard air-conduction hearing aids are often ineffective due to chronic ear infections, anatomical malformations or blocked ear canals. For this demographic, the current solutions are expensive and inadequate. On one end lie surgical implants (like BAHA), which are invasive and prohibitively expensive (costing >INR 8 Lakhs), restricting access to hearing impaired individuals who need it the most. On the other end are non-surgical alternatives, typically bulky, archaic headbands or adhesive devices. These workarounds fail in the real-world. India’s hot and humid weather conditions lead to sweat and this causes adhesive pads to fail and wear out frequently, they are visibly stigmatizing, uncomfortable for long-term wear and do not support any essential smart features like Bluetooth & Rechargeable Battery. A lack of awareness and accessible hearing screening infrastructure leads to millions going undiagnosed for hearing loss for years, leading to severe cognitive decline, social isolation and lost economic potential because of this delayed intervention. The critical problem is the complete absence of a comprehensive ecosystem that offers an affordable, non-surgical and socially acceptable solution while simultaneously addressing the grassroots challenge of early detection. |
Able Glasses are India’s first non-surgical smart aid glasses designed to rehabilitate individuals suffering from Conductive, Mixed and Sensorineural hearing loss. Able Glasses utilise our indigenous Multi-SoC Speech Processor that enables a dual-mode output system to address different types of hearing pathologies. Our Multi-SoC Speech Processor uses our developed embedded DSP audio processing firmware algorithm for speech selection and selective amplification. For patients with Conductive and Mixed hearing loss, the system utilizes a patented Solid-State dome bone conduction Transducer which is 98% lighter and 95% more power-efficient, used to deliver high-fidelity sound vibrations to the mastoid bone to bypass defective outer & middle-ear. For Sensorineural loss, the device features a swappable design that allows the bone conduction actuator to be replaced with a standard Balanced Armature receiver for air conduction for individuals with sensorineural hearing loss. The hardware is integrated into a universal rigid-flex frame that supports prescription lens fitting and it also provides true smart & essential features like wireless Bluetooth connectivity and over 12 hours of rechargeable battery life. To ensure high-quality hearing experience Able Glasses supports 16 to 20-channel digital Programmability and fine-tuning using a companion Able Connect app, enabling true customization based on each hearing impaired individual's specific hearing rehabilitation needs.
Our Able Assistant App supports 12+ Indian native languages to increase awareness and accessibility to quality hearing care, enabling a complete rehabilitation loop through its three key features. The Audiometry Screening Test allows users to conduct a hearing screening from the comfort of their home and receive a simple WHO-grade hearing report. The fine tuned LLM model ‘Able Assistant’ acts as a virtual audiologist answering all hearing-related queries in the user's preferred language and it also has a directory of 6,000 RCI-certified audiologists across India enabling users to easily connect with a certified hearing care professional near them. Thus with help our Able ecosystems hearing care solutions like Able Glasses & Able Assistant app to quality hearing care accessible and affordable for all. |
Our innovation is anchored in a proprietary hardware-firmware integration that miniaturizes and destigmatizes hearing rehabilitation. At the core is our Patented Solid-State Dome Bone Conduction Transducer, which is 98% lighter and 95% more power-efficient, enabling a discreet eyewear form factor for all day long use. This hardware is driven by our indigenous Multi-SoC Speech Processor featuring an embedded DSP audio processing firmware algorithm, which enables precise speech selection and selective amplification. Uniquely, this architecture supports Dual Output Driving, allowing the device to adapt to different pathologies by driving either the bone conduction actuator for conductive and mixed hearing loss or a Balanced Armature receiver for sensorineural hearing loss. To ensure long-term wearability and optimal transducer placement at the Mastoid bone, we utilize a two-material design frame manufactured via custom 3D head scan-based fabrication. Furthermore, the system offers wireless digital programming, allowing 16-20 channel fine-tuning via a companion Able connect app, at 1/10th the cost of existing alternatives. |
MVP |
Pilots, Signups, Testimonials |
Our target market includes individuals with conductive or mixed hearing loss for whom traditional hearing aids are ineffective. This clinically-defined group serves as our primary entry point. Our key growth segment includes students and working professionals who seek a discreet, fashionable and smart hearing solution with smart features like Digital Programmability, Bluetooth connectivity and a rechargeable battery that seamlessly integrates into their active personal and professional lives. We will then expand to serve the large and growing population of elderly and middle-aged individuals who experience both hearing and vision challenges, as Able Glasses can be fitted with prescription lenses for an integrated solution for both vision and hearing loss. While these groups are our end-users, our paying customers are a network of audiologists, ENTs, partner NGOs, corporate CSR programs and government bodies, creating a sustainable B2B2C business model designed for last-mile hearing healthcare delivery. |
USD 10 Billion |
USD 1.5 Billion |
USD 136 Million |
Our revenue model follows a scalable, ethical B2B2C approach aligned with RCI guidelines, where Able Glasses are prescribed and fitted by professional audiologists. Our primary revenue stream is the sale of Able Glasses, priced at INR 24,000 per unit. We partner with audiologists, ENT clinics, NGOs and optical chains who serve as our point-of-sale and fitting experts. This ensures clinical validity while leveraging existing healthcare infrastructure for rapid distribution without high customer acquisition costs. A secondary, recurring revenue stream is Brand Integration. We charge platform fees to service providers like Spotify or Audible, allowing them to utilize our smart glasses to reach the untapped market of hearing-impaired users and deliver their services. |
Our primary competitors include Cochlear (BAHA) and MED-EL (Adhear), which cater to individuals with conductive and mixed hearing loss. Cochlear BAHA is a surgical implant costing upwards of INR 8 lakhs, making it inaccessible for most users due to its invasive nature and high cost. MED-EL Adhear is a non-surgical alternative but relies on adhesive patches, costs over INR 2.4 lakhs, and involves recurring maintenance expenses of adhesive pads. Both solutions are limited in terms of comfort, long-term usability in Indian conditions and lack integrated smart features, resulting in a passive and often stigmatizing user experience. Able Glasses differentiates itself as India’s first non-surgical, non-adhesive smart aid integrated into a spectacle form factor, making it more discreet and suitable for all. Our core innovation lies in a patented solid-state dome bone conduction transducer that is 98% lighter and 95% more power-efficient than traditional electromagnetic systems. Combined with our indigenous Multi-SoC speech processor, the device performs on-device speech processing with 16–20 channel digital programmability, enabling personalized and high-clarity sound output. With Bluetooth connectivity, 12+ hours of rechargeable battery life, and pricing at nearly one-tenth the cost of existing solutions making Able Glasses an ideal solution for people with hearing loss. |
Our primary customer acquisition engine is a scalable B2B2C clinical partnership model. Traditional audiologists and ENT clinics currently lose a massive percentage of patients who simply cannot afford imported bone-conduction devices costing upwards of INR 1.5 Lakhs. By offering the Able Ecosystem at a disruptive B2B2C price of INR 24,000, we empower these medical professionals to convert those previously "lost" patients. This pricing leaves room for healthy channel margins, directly incentivizing clinicians to prescribe and dispense our device while maintaining the critical medical trust required for hearing rehabilitation. To build a strong digital top-of-funnel, we leverage our Able Assistant companion app as a powerful, pre-purchase lead generation tool. The app offers highly accessible, at-home hearing screening and counseling via our AI Virtual Audiologist in 12+ regional Indian languages. Once a user identifies a hearing issue through this free digital interface, the ecosystem naturally funnels them toward purchasing the Able Glasses online or booking a consultation with one of our partnered clinicians, a strategy that has already successfully screened over 2,000 individuals. Finally, we drive bulk acquisition through strategic partnerships with medical institutions, NGOs, and corporate CSR initiatives focused on rural healthcare and accessibility. By co-hosting and participating in localized hearing health awareness camps, we bypass expensive traditional D2C marketing and acquire users at scale within trusted community settings. This combined approach of clinical authority, digital reach, and institutional backing drastically lowers our CAC and has already allowed us to secure 1,500+ paid pre-orders prior to our official commercial launch. |
Our go-to-market strategy follows a structured B2B2C approach, where audiologists and hearing care centers act as our primary distribution and fitting partners, ensuring clinical trust and correct user adoption. On the demand side, we drive awareness and early detection through region-based hearing health awareness campaigns. In our hearing health awareness campaign, we conduct hearing screenings for free using our Able Assistant app, identify individuals with potential hearing loss and guide them toward appropriate diagnosis and rehabilitation. Through this approach, we have already conducted 2000+ hearing screenings and facilitated 280 early interventions, creating a strong pipeline of identified users. Our Able Assistant app strengthens this funnel by acting as a scalable digital entry point, enabling users to conduct hearing screening, access guidance in 12+ Indian languages, and connect with certified audiologists, thereby generating high-intent leads at low cost. To improve accessibility and scale, we will also work with government-backed schemes like ADIP and partner with NGOs to provide subsidized access making hearing care accessible and affordable for all. |
Our ultimate vision is to build the world's most accessible ecosystem of smart devices for the 680 million+ hearing, speech, and visually impaired individuals across the globe.
We will achieve this by first dominating the accessible hearing care market with the Able Ecosystem, utilizing our Multimodal AI and disruptive pricing to eradicate current barriers of cost and clinical scarcity. Once established, we plan to scale our core technology seamlessly bridging affordable deeptech hardware with native language AI support to develop a broader suite of assistive smart devices. Ultimately, Ksham Innovation will transition from a hearing-focused startup into a ubiquitous global healthcare platform, ensuring that comprehensive sensory rehabilitation is a fundamental, easily accessible right for all. |
Private Limited |
INR 1.42 Cr. |
Yes |
Mentorship and regulatory guidance: We are applying to Avinya to receive structured mentorship from the ACSEL ecosystem, especially to refine our go-to-market strategy and navigate regulatory approvals like CDSCO and clinical validation effectively. Support for validation and commercialization funding: The program can help us access follow-on funding opportunities and investor networks, which will support us in raising INR 3.2 crore required for clinical validation and commercialization of our Able Glasses. Access to strong networks and ecosystem: Connect with industry experts and ecosystem partners, which will help us accelerate partnerships with audiologists, institutions and scale our solution across India. |
During the programme, we aim to complete MVP development of Able Glasses and prepare for clinical validation. We seek mentorship to refine our go-to-market strategy, navigate regulatory approvals like CDSCO, and strengthen our fundraising approach to support validation and early commercialization of our Able Glasses. |
Yes. We are open for hybrid participation. |
Yes |
Automation layer |
Our AI architecture is designed as a modular system that connects user inputs, clinical data and device outputs into a single pipeline. When a user uploads an audiogram or asks a query, the system first converts visual or text inputs into structured hearing profiles using image processing and data parsing techniques. This information is then matched with relevant clinical knowledge from our internal database. The processed data is passed to a language model that generates clear, medically grounded responses and personalised recommendations for hearing impaired user.
When a user uploads an audiogram, the system uses image processing to read the report and extract hearing thresholds across frequencies, converting it into a structured hearing profile. This data is then mapped to acoustic parameters through our AI-based One Tap Programming system. The model translates the user’s hearing loss pattern into precise multi band gain settings, which are wirelessly synced to Able Glasses. This allows the device to be automatically configured based on the user’s specific hearing needs..
This architecture allows us to bridge clinical data, user interaction, and hardware execution, enabling personalized hearing care while maintaining accuracy and consistency across the system. |
Our core proprietary advantage lies in our Multimodal "Cyber-Physical" Data Loop, which captures the entire patient journey within the unique context of the Indian demographic. Unlike publicly available US and European audiology datasets, our proprietary dataset is built upon continuous, interacting data streams that generic AI systems cannot replicate. We are amassing a vast library of annotated clinical audiograms paired directly with the subsequent 20-band acoustic tuning parameters used in our hardware, allowing our AI to learn from exact, real-world calibrations rather than theoretical models.
Furthermore, through the Able Assistant, we are building a linguistic and behavioral dataset of hearing care interactions, troubleshooting queries, and rehabilitation responses in 12+ native Indian languages, while our hardware simultaneously captures the unique, high-noise acoustic environments specific to Indian use-cases. Because we own both the digital interface and the physical endpoint, we create a closed-loop data flywheel. Every new user interaction feeds directly back into our Multimodal AI, meaning our system continuously learns from actual patient outcomes over time, resulting in a highly defensible, compounding data moat that exponentially improves our personalized care. |
Our defensibility comes from the combination of domain-specific data and continuous clinical validation. While many systems can access general-purpose AI models, our system is built on a structured hearing care pipeline that integrates diagnostic data, device configuration, and user feedback.
A key part of this is our expert feedback loop, where audiologists validate outputs and guide improvements in the system. This ensures that recommendations remain clinically relevant and safe.
Over time, this creates a tightly integrated system where data, model behavior, and real-world outcomes reinforce each other. This depth in a specialized domain like hearing care makes it difficult for competitors to replicate without similar data and clinical integration. |
We evaluate model performance through a combination of rigorous internal testing, user validation and clinical feedback. Outputs generated by the system are reviewed against known audiogram inputs and expected hearing profiles to check accuracy of parameter mapping and response quality. We conduct structured user testing to assess clarity, relevance, and usability of recommendations in real scenarios. In parallel, audiologists and clinicians validate responses and device configurations to ensure medical consistency. Feedback is systematically incorporated to refine retrieval logic and output generation. We also track performance metrics such as response accuracy, consistency across similar inputs and user satisfaction to continuously improve reliability of the system. |
We follow a strict data privacy and security framework aligned with India’s Digital Personal Data Protection Act (DPDPA). User data is collected only with consent and used solely to improve hearing care services within our ecosystem. We do not share personal or medical data with external parties unless required for clinical collaboration and only with explicit user permission. All sensitive information is securely stored with access controls and encryption to prevent unauthorized use. We do not publicly release any identifiable user data. Our systems are designed to ensure confidentiality, integrity, and secure handling of all user information at every stage. |
There are no direct regulatory risks that block our product, but there are clearly defined steps we must follow to obtain approvals. Able Glasses fall under CDSCO’s Class B category medical device regulations and we are progressing through CDSCO Test License, followed by clinical validation and full certification. In parallel, we are implementing ISO 13485 quality systems and aligning with IEC safety standards & CDSCO approval to ensure compliance.
Potential challenges may include delay timelines for clinical validation and conducting multicenter cohort studies to validate the efficacy of Able Glasses. From a product standpoint, we have already built Able Glasses from scratch using biocompatible materials suitable for prolonged skin contact, which reduces regulatory uncertainty. However, our in-house expertise across hardware, firmware and design allows us to quickly iterate and resolve such issues. This ensures that while regulatory processes may take time, they do not pose a fundamental risk to our ability to successfully bring the product to market. |
At a higher scale, the main challenges will be maintaining response speed and managing system load. As the number of users increases, retrieving relevant clinical data and generating responses in real time can introduce latency.
We will need to improve how data is indexed and retrieved to ensure fast and accurate outputs. In addition, handling a higher volume of requests will require better management of model usage and infrastructure to maintain consistent performance. These are expected challenges in scaling AI systems and our architecture is designed to evolve with improved retrieval methods and optimized processing to ensure reliability as usage grows. |
Yes, we have an in-house AI team with strong cross-functional expertise. Our team includes an AI and firmware engineer who works on embedded signal processing, device-level intelligence, and integration of AI-driven audio optimization within Able Glasses. We also have an app development and AI engineer who focuses on building the Able Assistant platform, including hearing screening, data processing, and user interaction systems. Together, they work closely with our hardware and clinical teams to ensure that our AI systems are practical, accurate, and seamlessly integrated into both the device and the user-facing applications, enabling a reliable and scalable hearing care solution. |
We utilize a hybrid data strategy that seamlessly integrates strictly proprietary, internally owned datasets with legally compliant, licensed external clinical knowledge. To train our LLM-powered Virtual Audiologist, we leverage specialized external datasets encompassing academic literature, clinical case studies and research papers sourced directly from Audiology and Speech-Language Pathology (ASLP) curricula. Our internally owned dataset forms the core of our LLM real-world Indian context, consisting of patient-consented clinical audiograms, case notes and hardware tuning logs that directly power our cross-modal AI programming.
These external clinical guidelines and medical texts are ingested via fully licensed research databases to form the foundation of our Retrieval-Augmented Generation (RAG) architecture. This dual approach ensures that our foundational internal data remains a highly defensible, proprietary moat, while guaranteeing that the medical knowledge our AI dispenses is accurate, culturally contextual and strictly legally compliant. |
We plan to continuously improve our Multimodal AI performance through a Human-in-the-Loop architecture that leverages both expert clinical validation and real-world user behavior. For our visual-to-acoustic calibration model, the Able App allows partnered audiologists to manually fine-tune the baseline 20-band gain settings generated by our "AI One-Tap Programming." The delta between our AI’s initial prediction and the audiologist’s final clinical adjustment acts as direct training data, continuously refining the accuracy of our cross-modal mapping algorithms. Simultaneously, we use Reinforcement Learning from Human Feedback to optimize our LLM-powered Able Assistant; by tracking user resolution rates and follow-up queries across 12+ regional languages, we train the conversational AI to deliver increasingly accurate and culturally contextual medical troubleshooting. Finally, as our hardware captures diverse, high-noise Indian acoustic environments, we will train our embedded DSP algorithms on these localized audio sets, deploying enhanced speech-isolation and noise-suppression models directly to the Able Glasses via seamless Over-The-Air updates. This ensures our cyber-physical ecosystem becomes exponentially smarter, more personalized, and clinically precise with every new user. |
We are building a global solution, but intentionally launching with India as our primary beachhead market. India presents a massive, deeply underserved demographic of 63 million individuals with hearing loss, where nearly 90% remain untreated due to extreme price sensitivity, clinical scarcity, and a lack of regional language support. In India, the need is driven by low awareness, high cost and limited access to audiologists, especially in Tier 2 and Tier 3 regions. Our Able Ecosystem addresses this through affordable Able Glasses for rehabilitation and the Able Assistant app for early detection, awareness, and connection to nearby audiologists, making hearing care accessible at scale. By engineering the Able Ecosystem to thrive in this complex environment, specifically through disruptive ₹24,000 B2B2C pricing, 12+ native language AI models, and embedded DSP algorithms trained on unique, high-noise Indian acoustics.
We are building a hyper-resilient cyber-physical system. Once validated and scaled in this rigorous market, our architecture becomes highly exportable to other developing nations across Southeast Asia, Africa, and Latin America facing identical clinical barriers. Ultimately, our Multimodal AI and accessible hardware are designed for the global stage, allowing us to disrupt expensive Western incumbents and democratize medical-grade hearing care for the 430 million+ individuals suffering worldwide. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/hMfXVieG2tqP-Ksham-Innovation-Pitch-Deck-Application.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/hMfXVieG2tqP-Ksham-Innovation-Pitch-Deck-Application.pdf |
https://drive.google.com/file/d/1MYm4vrsFEjLM20eK14m4FC6X2GSldb7N/view?usp=drive_link |
Yes, we are building a mission-driven startup focused on making hearing care accessible, affordable and acceptable for all. Our Able Ecosystem directly enables early detection and effective rehabilitation for individuals with hearing loss, helping them lead independent and dignified lives.
Our work aligns with SDG 3 (Good Health) by improving access to hearing care, SDG 4 (Quality Education) by improving learning outcomes for children with hearing loss, SDG 8 (Decent Work) by enabling individuals to participate confidently in the workforce, SDG 9 (Industry & Innovation) through indigenous assistive technology solutions and SDG 10 (Reduced Inequalities) by reaching underserved communities for hearing rehabilitation.
We conducted a hearing health awareness campaign in Tier 2 regions such as Amravati and Akola, where we carried out 2000+ free hearing screenings through our Able Assistant app. With support of our clinical partners, we facilitated 280 early interventions by connecting individuals to audiologists for further care. We also identified 10 individuals who were medically suitable for Able Glasses fitting, helping us validate both the need and real-world applicability of our solution.
We are aligned with the Ministry of Social Justice and Welfare’s vision for empowering persons with disabilities and aim to partner with central and state government initiatives such as ADIP and NPPCD to scale access to hearing care, ensuring that even the last mile population receives timely and effective hearing care support. |
Assistive Technology |
T R Anand |
checked |
| Apr 24, 2026 @ 4:13 PM |
Dr. PREMANAND MUKUND HONAVAR |
premanand18honavar@gmail.com |
http://www.linkedin.com/in/dr-premanand-mukund-honavar |
http://N/A |
9448153586 |
CEO, RESEARCH BACKGROUND |
RTD, Principal and working in the Dr. M. P. KARKI INSTITUTE OF EXCELLENCE AND RESEARCH as filtrate consultant |
|
Research skill |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 24, 2026 @ 3:58 PM |
Madhava Madanapalli and Rajesh Muppalla |
madhava@mactai.com |
https://www.linkedin.com/in/madhavam/ |
http://www.mactai.com |
9980999535 |
Madhava comes with nearly 37 years of insurance domain experience in India and outside. The last 25 years has been in the tech world providing technology solutions for the insurance industry. Rajesh is an experienced AI/ML entrepreneur and engineering leader. He co-founded Indix, which was acquired by Avalara in 2019, where he led the data platform for large-scale product intelligence. Post-acquisition, he served as Senior Director of Engineering at Avalara, focusing on AI-driven tax automation. Previously, he was a tech lead on GoCD at ThoughtWorks. Rajesh earned a Computer Science gold medal from Savitribai Phule Pune University. |
We met through a common friend in May / June 2024 and have been working together. We built an initial product, took it to customers, pivoted based on feedback and we now have signed customer for a three year contract. A large general insurance company in India. |
two |
Industry knowledge and technology depth. |
ANVIUM AI LABS PRIVATE LIMITED |
WWW.MACTAI.COM |
BENGALURU |
AI SOLUTIONS FOR THE INSURANCE INDUSTRY AND EXPAND TO OTHER DOMAINS |
Extracting information from varying types of documents. They can be structured, unstructured, in any language, hand written as well as accident sketches. Our solution goes beyond OCR/ ICR and infers data from the context of the documents |
The Motor Third Party claims in India present multiple, unique challenges. One of them is that each claim requires multiple documents to be submitted to qualify for compensation as well as quantum of compensation. Almost all documents are state specific or sometimes even district specific. The documents are in varying formats, varying structures, in Indian languages. The documents can be hand written statements or even haphazardly written doctors' notes. They can even include sketches describing how the accident occurred. They include photos as well. IN addition, data is required to be validated with both internal databases of insurance companies as well as external data sources like E-courts and Sarathi and Vaahan portals of Government of India.
Our solution extracts the required information (usually between 120 - 150 fields) that Insurers need to process the claim. We extract in about 2-4 minutes (against the current time of hours and sometimes days) that insurance companies spend on extracting data from multiple documents.
We publish the data in any format that the Insurers require.
We do not hold any data in our system as our solution is head-less and hence comply with DPDP provisions. Our solution can be hosted within the cloud infrastructure of the client organisation and there are models which ensure that the data does not go out of the client's cloud infrastructure. |
In our experience we have not seen any such solution in the insurance space. Customer validation in the form of a PO is a strong indicator of the ubiquity of our solution. |
Revenue |
two, Pilots, Signups |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 24, 2026 @ 12:33 PM |
|
|
|
|
|
|
|
All |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 23, 2026 @ 7:24 PM |
Madhavi Shapeti ; Vijaya Srisha |
bossmastudioworks@gmail.com |
https://www.linkedin.com/in/madhavishapeti/ |
http://bossma.in/ |
+919630034495 |
Founder-led venture. Madhavi Shapeti leads full-time, with support from Venkatesh Hariharan (India Stack) (Advisor & Mentor) and a network of mentors and consultants across product, technology, and ecosystem partnerships. |
Venky is an advisor and mentor from the founder’s professional network. Additional mentors and consultants are engaged through prior ecosystem work and collaborations. |
one |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 23, 2026 @ 6:32 PM |
Dr Kishore Kumar and Samarth C S |
kk@landright.in |
https://www.linkedin.com/in/samarthcs?utm_source=share&utm_campaign=share_via&utm_content=profile&utm_medium=ios_app |
https://landright.in |
9686262712 |
Founder and CEO, Co-founder and CPTO |
We were classmates during 11th and 12th Standards, known each other since then, have been working on this start-up for about 10months. |
All |
I have the domain knowledge and experience for over a decade while Samarth has the strong technical background with experience of building Start-Ups. |
Landright Technologies Private Limited |
https://landright.in |
Bangalore |
Landright combines the power of AI with Legal expertise to bring a new era of transparency in real estate. |
High risk of property related litigation |
1. AI-powered document parsing extracts ownership, encumbrances, and anomalies from land records within hours instead of weeks.
2. Litigation risk engine cross-checks property data with court records to flag hidden disputes and legal exposure early.
3. Automated verification of registry, mutation, and compliance with state land laws ensures standardised legal due diligence.
4. Structured legal opinion reports validated byexpert lawyers provide clear, actionable risk assessment for stakeholders |
Over robust system performs analysis of property documents and highlights risks for review by Legal Professionals expediting the whole process, cutting down human hours required for same work with better accuracy. |
MVP |
Pilots |
Banks, NBFCs, Builders and Developers, Individual Property Buyers |
25,000crores |
8,000crores |
800crores |
B2C Orders with one time fee of 5000(PSU Banks) and 1200(NBFCs) per legal consultation |
Lending Katalyst, Prava |
Strategic Partnership with Banks, NBFCs, Developers, Proptech Firms |
Sales to Banks, NBFCs, etc |
To be a reliable market leader in the Property Due Diligence space with prominent presence across all major states in India. |
Incorporated as Private Limited Company on 15th July 2025 |
NA |
Yes |
For Mentorship and Funding |
Improve the product, have access to market and funding |
Yes |
Yes |
Core engine |
Gemini Flash 2.5 API, etc |
A custom built Rule-Engine |
Best in class providing extensive analyses with Human validation layers. |
We are using Open Telemetry to track the model outputs and metrics, so that we can utilise it for our improvements and fine-tuning. |
We are using most of the services from GCP and AWS which are managed solutions so that the security and data privacy are pre-handled.
We are using Secrets Manager to manage all the secrets so that we do not have to rely on environment variables.
We are using Service Accounts wherever possible, avoiding API Keys.
We are using Strictest typing and linting practices. |
None |
Vertex AI organisation rate limit could be a bottle-neck, which can be resolved. |
No, Co-founder has Applied AI Expertise. |
We are creating our own data-sets with each run. |
1. Prompt Engineering
2. Generating data for Fine-tuning
3. Going to a better model
4. Harness Engineering |
India |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/tjb3iU8T9sgt-LANDRIGHT-TECHNOLOGIES-PRIVATE-LIMITED-PITCH-DECK-GENERAL.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/tjb3iU8T9sgt-LANDRIGHT-TECHNOLOGIES-PRIVATE-LIMITED-PITCH-DECK-GENERAL.pdf |
http://NA |
Impact focused start-up, as 66% of court cases in India are related to property disputes, where a huge amount of capital gets tied-up and people's life-time earnings are stuck in litigation for generations. |
NA |
Founders' Whatsapp Group |
checked |
| Apr 23, 2026 @ 5:27 PM |
Shiv Shankar; RathaRamanan Natarajan |
shiv@easeworkai.com |
https://www.linkedin.com/in/shiv-shankar-mktg/ |
https://easeworkai.com |
09901931337 |
Shiv Shankar (CEO & Co-founder) MBA, Mechanical Engineer with 16+ years in global manufacturing. Led sales, marketing, and business development roles, working closely with enterprise customers across regions. Brings deep understanding of procurement workflows, customer pain points, and how buying decisions are made in large organizations. At EaseworkAI, leads GTM, customer acquisition, and product direction with a strong focus on real-world usability and outcomes. RathaRamanan Natarajan (CPTO & Co-founder) Mechanical Engineer with 20+ years of experience in procurement and supply chain transformation. Has led large-scale enterprise initiatives, delivering next-gen procurement solutions and driving digital transformation programs. Deep expertise in building AI/ML-driven systems for real-world enterprise use cases. At EaseworkAI, leads product architecture, AI development, and platform scalability, ensuring the solution is both technically robust and aligned with customer workflows. |
We met at a startup event in Bangalore and have known each other for over 2.5 years now. Both shared the same passion to build our own startup and given our complimentary skillsets and domain understanding, we started Easework AI together. |
All |
Our biggest strength is the combination of deep domain expertise and the ability to translate it into working AI products that deliver measurable outcomes.
We’ve spent decades inside procurement and manufacturing environments, so we understand the real complexity, not just the surface-level workflows. That allows us to identify high-impact problems that are often overlooked or poorly solved by generic tools.
At the same time, we have the technical capability to build and deploy AI systems that actually work in enterprise settings. This is reflected in our live deployments with companies like major enterprises, where we’re already reducing manual effort by ~65% and improving sourcing efficiency.
What this really means is we’re not experimenting in isolation, we’re building with constant customer feedback, validating quickly, and iterating toward real business outcomes.
That combination of insight, execution speed, and customer proximity is what gives us an edge. |
Easework AI |
https://easeworkai.com |
BANGALORE |
Category-intelligent Agentic AI for procurement and supply chain in manufacturing |
Procurement in manufacturing is still highly manual, fragmented, and reactive, especially in project-driven environments where every sourcing need is different. Teams rely on a mix of ERP systems, spreadsheets, emails, and supplier interactions that don’t talk to each other. As a result, sourcing cycles are slow, decisions depend heavily on individual experience, and opportunities for cost savings or risk mitigation are often missed. Most existing tools focus on visibility or basic automation, but they don’t adapt well to changing specifications, supplier dynamics, or negotiation scenarios. This makes them ineffective for real-world procurement, where variability is the norm. What this leads to is wasted time, inconsistent outcomes, and significant untapped value in enterprise spend. We’re solving this by addressing the core issue, not just the symptoms, enabling procurement teams to move from manual coordination to intelligent, adaptive execution. |
EaseworkAI is an AI-native platform that deploys specialized agents to automate and optimize end-to-end procurement workflows.
Instead of relying on static tools or dashboards, our system uses AI agents that understand context, whether it’s a new project requirement, supplier constraints, or category-specific nuances, and take action across the sourcing lifecycle. This includes supplier discovery, RFQ/RFP creation, bid evaluation, negotiation support, and post-PO tracking.
The platform integrates with existing ERP and SaaS systems, along with external data sources, to create a unified intelligence layer. This allows the agents to continuously learn from past decisions, supplier performance, and market signals, improving outcomes over time.
What this really means for customers is a shift from manual, fragmented processes to intelligent, adaptive execution. Teams move faster, make better decisions, and unlock measurable value in cost, efficiency, and risk reduction.
We’re already seeing this in production, with enterprise customers achieving ~65% reduction in manual effort and significantly faster sourcing cycles. |
We combine deep procurement expertise with AI agents that execute workflows, powered by strong category intelligence and our own middleware layer. At the core is our category intelligence. Our platform understands category-specific cost drivers, supplier dynamics, and negotiation levers, allowing our agents to make context-aware decisions across sourcing, evaluation, and negotiations. This is critical in project-driven environments where variability is high. We built our own middleware, with no dependency on third-party integration layers. This allows us to integrate faster, maintain higher reliability, and have full control over data flow and system behavior across ERP, SaaS tools, and external data sources. It also gives us flexibility to adapt quickly to complex enterprise environments without being constrained by external platforms. Unlike traditional tools, we don’t stop at insights. Our AI agents execute across the procurement lifecycle, embedding directly into day-to-day operations and creating strong workflow ownership. Our defensibility compounds through usage. As customers use the platform, we continuously build structured intelligence across categories, suppliers, pricing patterns, and outcomes, making the system smarter and harder to replace over time. |
Revenue |
Revenue |
Procurement teams in manufacturing companies, specifically automotive |
USD75Bn |
USD15Bn |
USD155Mn |
We operate on a subscription-based SaaS model with modular pricing. Customers pay an annual subscription based on the modules they use, such as sourcing, supplier management, negotiation intelligence, and post-PO automation. This allows enterprises to start with a specific use case and expand over time. |
Lio, Didero, Procure.ai |
Events, LinkedIn, System integrator partnerships |
We target manufacturing enterprises with a land-and-expand approach, starting with high-impact use cases to prove value quickly. Customer acquisition is driven by founder-led sales, targeted outreach, and problem-focused demos. Once deployed, we expand across categories and workflows within the same account. Partnerships with consultants and system integrators help scale access, while our proprietary middleware enables faster integration and shorter sales cycles. |
Our vision is to become the default AI operating layer for procurement and supply chain in manufacturing.
We believe procurement will shift from manual, process-heavy coordination to autonomous, intelligence-driven execution. In this future, AI agents will not just support decisions but actively run sourcing, negotiations, and supplier management with minimal human intervention.
EaseworkAI is building toward that by combining category intelligence, agentic workflows, and a proprietary middleware layer that connects fragmented enterprise systems into a unified execution engine.
Over time, as we scale across customers and geographies, our platform will develop a deep, continuously improving intelligence layer across categories, suppliers, pricing, and market dynamics. This positions us to not only optimize individual enterprise decisions but also redefine how procurement operates at an industry level. |
Incorporated in Canada. India to be initiated. |
USD 65000 |
Yes |
We’re applying to Avinya because it’s one of the few accelerators that is deeply aligned with what we’re building, AI-native, enterprise-grade solutions solving real-world problems. Avinya’s founder-first approach, combined with structured mentorship, industry access, and curated datasets, is particularly relevant for us at this stage. As we scale EaseworkAI, access to domain experts, IIT alumni networks, and enterprise ecosystems will help us refine our product, strengthen our GTM, and accelerate enterprise adoption. We also see strong value in the program’s focus on real-world pilots and industry engagement. Our platform benefits significantly from continuous feedback and deployment in live procurement environments, and Avinya’s ecosystem can help us expand these opportunities faster. |
We want to use the program to accelerate three things: product maturity, enterprise traction, and GTM readiness for global markets. First, we aim to deepen our product by strengthening category intelligence and expanding agent capabilities across more procurement workflows, backed by real-world feedback from pilot deployments. Second, we want to secure additional enterprise customers and run high-quality pilots that demonstrate measurable outcomes. This will help us build stronger case studies and refine our value proposition. Third, we want to sharpen our go-to-market strategy for scaling, particularly in the US, by leveraging mentorship, partnerships, and investor access within the ecosystem. |
Hybrid |
Yes |
Core engine |
Our architecture is built as a modular, AI-native system combining LLMs, domain-specific intelligence, and a proprietary middleware layer to enable end-to-end procurement execution.
At the core, we use a combination of leading large language models including Claude, OpenAI, and DeepSeek, alongside smaller task-specific models for structured prediction and optimization. The LLMs handle unstructured workflows such as RFQ generation, supplier communication, and document understanding, while specialized models support bid evaluation, scoring, and recommendation logic.
We follow an agent-based architecture where multiple AI agents are orchestrated across the procurement lifecycle. These agents are context-aware, leveraging retrieval-augmented generation (RAG) over enterprise data, historical transactions, supplier information, and category-specific knowledge bases.
Our platform runs on a scalable, cloud-native stack with Python-based services for AI orchestration, microservices for modularity, and API-first design for interoperability. We use vector databases for semantic search and knowledge retrieval, along with structured pipelines to ingest data from ERP systems and external sources.
A key differentiator is our proprietary middleware, built without third-party dependency. It enables faster and more reliable integration with ERP, SaaS tools, and external systems, while giving us full control over data flow, performance, and security.
We also incorporate human-in-the-loop controls, allowing procurement teams to review and guide agent decisions, ensuring trust, compliance, and practical usability in enterprise environments.
Overall, the system is designed to be adaptive, scalable, and continuously improving as more data and usage flow through it. |
Our data advantage comes from how we structure and learn from real procurement workflows, not just access to raw data.
As our platform is used, we generate proprietary, structured datasets across categories, suppliers, pricing patterns, negotiation behaviors, and sourcing outcomes. This data is captured in the context of actual decision-making and execution, which makes it far more valuable than static or third-party datasets.
A key differentiator is our category intelligence layer. We continuously build normalized, category-specific knowledge around cost drivers, supplier dynamics, and negotiation levers, which improves the performance of our agents over time.
Because we own the middleware layer, we also control how data is ingested, standardized, and connected across ERP systems, external sources, and workflows. This allows us to create a unified and cleaner data foundation compared to fragmented enterprise environments.
Importantly, this advantage compounds with usage. As we onboard more customers and transactions, our models improve, our benchmarks get stronger, and our recommendations become more accurate and context-aware. |
Our defensibility comes from the combination of category intelligence, workflow ownership, and continuously compounding data tied to real execution.
First, our AI is not generic. It is trained and tuned on category-specific procurement intelligence, including cost drivers, supplier behavior, and negotiation patterns. This context allows our agents to make better decisions in real-world, variable environments where horizontal AI tools struggle.
Second, we are embedded in execution, not just analysis. Our agents operate directly within procurement workflows, generating RFQs, evaluating bids, and supporting negotiations. This gives us access to high-quality, decision-level data that competitors who sit at the dashboard layer don’t capture.
Third, our proprietary middleware ensures we control how data flows across systems. This allows us to standardize fragmented enterprise data and build a cleaner, more reliable foundation for our models, without dependency on third-party integration layers.
Finally, our advantage compounds over time. As more transactions run through the platform, we continuously build structured intelligence across categories, suppliers, pricing, and outcomes. This creates a feedback loop that improves performance and increases switching costs. |
We evaluate performance at three levels: model quality, workflow outcomes, and system reliability.
At the model level, we track task-specific metrics depending on the use case. For structured tasks like bid evaluation or classification, we use accuracy, precision/recall, and consistency against historical decisions. For LLM-driven tasks such as RFQ generation or supplier communication, we evaluate response quality, factual correctness, and adherence to defined templates and constraints.
At the workflow level, we focus on business impact. Key metrics include reduction in manual effort, cycle time improvements in sourcing, quality of supplier shortlisting, and alignment of recommendations with final procurement decisions. This ensures the models are not just technically sound but actually useful.
For reliability, we track system-level metrics such as response latency, failure rates, and consistency of outputs across repeated runs. We also monitor hallucination rates and enforce guardrails using retrieval (RAG), structured prompts, and validation layers.
We incorporate human-in-the-loop evaluation as well, where procurement users review and rate outputs, providing continuous feedback for improvement.
Finally, we run offline benchmarks using historical procurement data and continuously test models in controlled environments before deploying updates. |
We treat data privacy and security as foundational, especially given the sensitivity of procurement and supplier data.
At the core, we follow a strict data isolation approach. Each customer’s data is logically segregated, ensuring no cross-tenant access or leakage. Customer data is never used to train shared models without explicit consent.
All data is encrypted in transit and at rest using industry-standard protocols. We enforce role-based access control (RBAC), maintain detailed audit logs, and ensure full traceability of user actions and system decisions.
Our proprietary middleware gives us complete control over data flow across ERP systems, SaaS tools, and external sources. This allows us to enforce security policies, validate inputs and outputs, and minimize exposure without relying on third-party integration layers.
For AI/LLM interactions, we implement safeguards such as controlled prompting, data masking where required, and secure API configurations to prevent sensitive data leakage.
For large enterprises with stricter requirements, we also support on-premise or private cloud deployments. This ensures that all data, model interactions, and workflows remain within the customer’s controlled environment, addressing data residency, compliance, and internal security policies.
From a compliance standpoint, we align with enterprise-grade security practices and are working toward certifications such as ISO 27001 and SOC 2. |
There are evolving regulatory considerations around data privacy, AI usage, and cross-border data flows, especially as we scale into global markets.
In procurement, the primary risks relate to handling sensitive enterprise data, supplier information, and ensuring compliance with data protection regulations such as GDPR and similar frameworks in other regions. There is also increasing scrutiny around the use of AI in decision-making, particularly in terms of transparency, auditability, and bias.
However, our architecture is designed to mitigate these risks. We ensure strict data isolation, do not use customer data to train shared models without consent, and maintain full auditability of AI-assisted decisions. Our human-in-the-loop approach also ensures that critical decisions remain reviewable and compliant with internal policies.
We also support flexible deployment models, including on-premise and private cloud, which helps address data residency and regulatory requirements for large enterprises.
As regulations evolve, particularly around AI governance, we see this as a tailwind rather than a constraint. Enterprises will increasingly prefer solutions that are transparent, controllable, and compliant by design, which aligns with how we are building our platform. |
At 10x scale, the main challenges are around data variability, orchestration complexity, and maintaining consistent performance across diverse enterprise environments.
First, variability increases significantly. Different customers have different data structures, procurement processes, and category nuances. If not handled well, this can impact model accuracy and consistency. We address this through our middleware layer and standardized data models, but scaling this across many enterprises is a non-trivial challenge.
Second, agent orchestration becomes more complex. As we expand across more workflows and categories, coordinating multiple AI agents reliably and in real time requires strong control systems, monitoring, and fallback mechanisms. Without this, latency and failure rates can increase.
Third, LLM reliability and cost at scale become important. Ensuring consistent output quality, managing hallucinations, and optimizing cost-performance trade-offs across models like Claude, OpenAI, and DeepSeek is an ongoing challenge.
Fourth, maintaining category intelligence at scale requires continuous learning and validation. As we expand into new categories and geographies, ensuring our intelligence layer remains accurate and relevant is critical.
We’re actively building for this by investing in modular architecture, strong evaluation frameworks, human-in-the-loop systems, and our proprietary middleware to standardize and control data flows. |
Yes, we have an in-house AI/ML team led by our co-founder and CPTO, with deep expertise in building enterprise-grade AI systems for procurement and supply chain.
The team’s strength lies in applied AI, not just model development. We focus on designing and deploying real-world systems that combine LLMs (Claude, OpenAI, DeepSeek), retrieval-augmented generation (RAG), and task-specific models to solve complex, unstructured workflows.
Our expertise spans:
- Agent-based architecture and orchestration across multi-step enterprise workflows
- Building and fine-tuning domain-specific intelligence layers, especially category intelligence in procurement
- Data engineering and pipeline design for integrating fragmented ERP and external data
- Evaluation frameworks for measuring model accuracy, reliability, and business impact
- Designing human-in-the-loop systems to ensure control, auditability, and trust
We also have strong backend and systems engineering capabilities, which are critical for building our proprietary middleware and ensuring scalable, reliable integrations without third-party dependencies. |
We primarily work with a combination of customer-owned data, internally structured datasets, and selective external data sources.
The core of our system is built on customer data, including historical procurement transactions, supplier information, RFQs/RFPs, contracts, and process workflows. This data remains owned by the customer, and we use it in a tenant-isolated manner to power context-aware automation and decision-making.
On top of this, we generate and maintain internally structured datasets derived from real usage. This includes normalized category intelligence, supplier performance patterns, pricing benchmarks, and negotiation insights. These are built from aggregated learnings at a structural level, without exposing any customer-specific sensitive data.
We also incorporate external or third-party data sources where relevant, such as market signals or supplier information, which are either publicly available or appropriately licensed depending on the use case.
Importantly, we do not rely on large proprietary pre-trained datasets as our primary advantage. Our strength comes from continuously learning from live workflows and turning that into structured, reusable intelligence. |
We improve performance through a closed-loop system that combines real-world usage, structured feedback, and continuous evaluation.
First, every interaction on the platform generates learning signals. As users review, modify, or approve outputs, we capture that feedback to refine prompts, agent logic, and decision frameworks.
Second, we continuously build and update our category intelligence layer. As more transactions flow through the system, we improve our understanding of cost drivers, supplier behavior, and negotiation patterns, which directly enhances model outputs.
Third, we run regular offline evaluations using historical procurement data, along with A/B testing in controlled environments before deploying updates to production. This ensures improvements are measurable and do not degrade performance.
Fourth, we optimize model selection and orchestration. We dynamically use models like Claude, OpenAI, and DeepSeek based on task requirements, balancing accuracy, latency, and cost.
We also maintain human-in-the-loop controls, allowing domain experts to guide and validate outputs, especially in critical workflows.
Finally, our proprietary middleware ensures clean, structured, and consistent data pipelines, which is critical for maintaining and improving model quality at scale. |
We are building for India and North America (NA). NA is attractive because they adopt technologies early. India is a focus because of our base. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qC15sUj4hPn0-Easework_Investor_Pitch-Apr-26.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qC15sUj4hPn0-Easework_Investor_Pitch-Apr-26.pdf |
https://drive.google.com/file/d/1P0lLg3a-PuJj_xeRMBbTP4W9-ExUPpXq/view?usp=sharing |
Yes, we are building a mission-driven company, with a focus on making procurement smarter, more efficient, and more responsible through AI.
Our mission is to enable businesses to procure faster, make better decisions, and operate with greater transparency. Today, procurement is often manual, opaque, and reactive, which leads to inefficiencies, missed savings, and inconsistent supplier decisions.
By bringing intelligence and automation into the core of procurement workflows, we help organizations reduce waste, optimize spend, and make more informed, data-driven choices. This has a direct business impact, but also a broader effect in terms of responsible sourcing and better supplier engagement.
As our platform evolves, we see an opportunity to drive more meaningful impact, such as enabling ethical sourcing decisions, improving supplier diversity, and reducing inefficiencies across global supply chains. |
NA |
NSRCEL |
checked |
| Apr 22, 2026 @ 10:15 PM |
Rituraj Borah, Nabajyoti Borah |
kazirangaai@gmail.com |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 22, 2026 @ 9:57 PM |
Lakshmeesha N |
lakshmeeshan1234@gmail.com |
https://www.linkedin.com/in/lakshmeesha--n/ |
https://opandz-deploy.onrender.com/ |
6363682051 |
Founder & Technical Architect B.E. in Artificial Intelligence & Data Science |
We met as classmates in the same branch—Artificial Intelligence and Data Science—during our undergraduate program. We have been working together on academic projects |
1 |
Our biggest strength as a team is the combination of technical depth and market reach. I focus on building and engineering AI-driven solutions, while my co-founder brings expertise in content creation, marketing, and influencer engagement. Together, we bridge innovation and audience trust—developing robust technology and ensuring it connects with the right users. |
Opandzai |
https://opandz-deploy.onrender.com/ |
Tumkur |
Opandz AI is the Blueprint Governance Engine—an autopilot that eliminates repetitive formatting by extracting the structural DNA of documents and generating error‑free outputs instantly. |
Professionals in law, finance, healthcare, and academia waste up to 30% of their time on repetitive “document janitorial work”—manual formatting, structural corrections, and compliance checks. Current AI tools generate text but are blind to strict layout standards, forcing experts to fix outputs manually. This productivity leak undermines efficiency and credibility, and even small formatting errors can render legal or academic documents invalid. |
Opandz AI solves this gap. We are building the Blueprint Governance Engine—a system that transforms one perfect document into a permanent automation asset. By extracting the “Structural DNA” (coordinates, font hierarchies, compliance rules) from a gold-standard file, Opandz AI stores it in a secure vault. Professionals can then instantiate new documents with one click, automatically born into their correct professional form.
Our solution goes beyond formatting:
Living Blueprint Vaults memorialize standards across industries.
Domain-Aware Plugins ensure content is written with the tone, vocabulary, and logic of law, medicine, finance, or academia.
Data-Agnostic Automation keeps sensitive information private while reusing structural excellence.
Collaborative Micro-Communities turn one person’s effort into shared institutional assets, compounding efficiency across teams, classes, and firms.
The impact is exponential. In a class of 60 students, one person’s 30-minute effort eliminates 120+ hours of redundant work. In a law firm, the first-to-finish petition becomes a reusable blueprint for the entire team. Opandz AI transforms paperwork from a daily frustration into a shared autopilot for professionals everywhere—preserving trust, saving time, and scaling seamlessly across industries.
We are not just building another AI writer. We are building the institutional memory of professional paperwork, where every perfected document becomes a reusable automation asset.
Opandz AI is the future of paperwork: from manual formatting to collective autopilot. |
Opandz AI is unique because it turns perfected documents into reusable automation assets. By extracting their “Structural DNA” and storing it in secure vaults, we enable one‑click, compliant outputs across industries—saving time, preserving trust, and scaling efficiency. |
MVP |
Users |
Our target customers are professionals and institutions whose credibility depends on flawless documents. This includes students and researchers (academic theses, journals), lawyers and legal firms (court filings, briefs), and finance/healthcare professionals (compliance reports, audits). |
$32.1B |
$4.8B |
$150M, |
We operate on a freemium SaaS model. Basic document automation is free for short reports, while professional subscriptions are required for longer documents and domain‑aware plugins (legal, medical, finance, academia). This ensures recurring ARR from high‑value users while keeping entry accessible. |
Adobe Acrobat (manual formatting), and general AI writers like ChatGPT/Claude (structurally blind). Unlike them, Opandz AI uniquely captures and reuses document “Structural DNA” as a secure automation asset. |
We acquire customers through academic cohorts (students, researchers), pilot programs with legal firms, and influencer‑driven content marketing. Early traction comes from surveys and beta testing with students and professionals facing “formatting fatigue. |
Phase 1: Academic Alpha—launch with students and researchers for theses/journals. Phase 2: Legal & Finance—deploy domain‑aware plugins for briefs and compliance reports. Phase 3: Enterprise SaaS—scale via APIs and collaborative blueprint libraries. |
To become the universal Blueprint Governance Engine—the autopilot for professional paperwork across law, finance, healthcare, and academia. Our vision is to build the institutional memory of documents, where every perfected file becomes a reusable automation asset, preserving trust and scaling globally.We are not selling AI features. We are elevating the professionals who safeguard society’s trust through paperwork—because every document is more than text, it is proof, integrity, and credibility made visible |
NA |
NA |
Yes |
We are applying to Avinya to gain structured mentorship, access to a strong founder network, and guidance on scaling our Blueprint Governance Engine beyond academia into legal and financial domains. Avinya’s focus on deep‑tech and impact‑driven startups aligns perfectly with our mission to preserve professional trust while automating paperwork. |
During the programme, we aim to refine our MVP, validate product‑market fit with pilot cohorts, and strengthen our go‑to‑market strategy. We also want to build investor readiness by sharpening our financial model and positioning Opandz AI as a defensible, trust‑first platform |
Yes |
Yes |
Core engine |
Dual‑agent system: Extraction Agent captures document “Structural DNA” (coordinates, hierarchies, compliance rules), while Editor Agent rewrites content autonomously. Stack: FastAPI backend, Python (LangChain/LiLT/YOLO), Supabase for vaulting, Docker for deployment, GPU clusters for training, opan ai APIs for LLM inference. |
it treats document structure as a reusable automation asset. By extracting the “Structural DNA” of a gold‑standard file—its coordinates, hierarchies, and compliance rules—we store it in a secure vault that professionals can reuse instantly. Our Blueprint Governance Engine goes beyond formatting with Living Vaults, Domain‑Aware Plugins, and Data‑Agnostic Automation, ensuring privacy and accuracy. This creates exponential impact: one perfected document eliminates hours of redundant work across teams, turning paperwork into a collective autopilot |
Our defensibility comes from proprietary “Blueprint Vaults” that store perfected document structures (Structural DNA). This institutional memory is hard to replicate, and combined with domain‑aware plugins, it ensures compliance across law, finance, healthcare, and academia. Beyond technology, we create a community effect—each perfected document strengthens the shared vault, making the system smarter and more valuable as adoption grows |
We measure structural fidelity (1:1 layout match), compliance accuracy, latency benchmarks, and user satisfaction. Reliability is validated across diverse domains—academic, legal, and financial—to ensure robustness beyond a single use case. |
Opandz AI is defensible because it captures and reuses the “Structural DNA” of documents across law, finance, healthcare, and academia. Our proprietary Blueprint Vault and large curated datasets are difficult to replicate, and domain‑aware plugins ensure compliance and professional tone across industries. |
Yes—jurisdiction‑specific formatting rules, evolving AI/data regulations, and compliance requirements in legal and healthcare sectors. We mitigate these risks with domain‑aware plugins and local‑first vaulting. |
Latency for large files and GPU bottlenecks. Scaling requires optimized orchestration, distributed vaulting, and higher‑tier GPU clusters. |
Yes. Our team has expertise in AI & Data Science, document rendering pipelines, ML integration, backend systems, and has won innovation challenges for scalable AI systems. |
We are currently working with curated academic datasets that we own, focusing on labeling and structuring them for MVP development. As we expand, we will incorporate legal and financial document datasets, |
We plan to train separate models for each domain (law, finance, healthcare, academia) while also maintaining generalist models. Using a mixture‑of‑experts approach, we combine domain‑specific precision with generalist scalability. Continuous improvement comes from fine‑tuning on new datasets, pilot feedback loops, and expanding blueprint libraries to evolve with professional standards. |
Both—starting with IndiWe plan to train separate models for each domain (law, finance, healthcare, academia) to capture domain‑specific standards, while also maintaining generalist models for cross‑domain tasks. Using a mixture‑of‑experts approach, we combine specialized models with generalists to balance precision and scalability. Performance will be iteratively improved through fine‑tuning on new datasets, pilot feedback loops, and expansion of blueprint libraries to evolve with professional standards.a for early pilots, then expanding globally across legal, financial, healthcare, and academic sectors |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/2zvwya1KllUC-Opandz-AI.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/2zvwya1KllUC-Opandz-AI.pptx |
https://youtu.be/R3mXgXiVv0k |
Opandz AI is mission‑driven: we exist to preserve professional trust while automating paperwork. Our impact focus is twofold—reducing “formatting fatigue” for students, lawyers, doctors, and researchers, and ensuring that every perfected document becomes a reusable automation asset. By building the Blueprint Vault, we empower communities of professionals to share structural knowledge, creating a collective network effect. The mission is not just efficiency—it’s credibility at scale, ensuring that automation strengthens institutions rather than eroding them. |
NA |
NA |
checked |
| Apr 22, 2026 @ 9:17 PM |
Amit Yadav |
amit.fortax@gmail.com |
https://in.linkedin.com/in/amit-fortax |
https://fortax.in |
9559272225 |
Founder & CEO |
We met at a startup event in 2023. |
two |
10 |
fortax |
https://fortax.in |
Lucknow |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 22, 2026 @ 3:30 PM |
Vipul Pandey, Kunal Pandey |
vipulpandey001@gmail.com |
https://www.linkedin.com/in/vipulpandey001/ |
https://unmaster.in/ |
8825675766 |
Vipul Pandey — CEO & Co-Founder 4 years as Assistant Manager at State Bank of India — directly experienced the competitive exam ecosystem and aspirant pain points Research Scholar at IIT Bombay — where the idea for Unmaster originated Leads product vision, growth strategy, and business development Strong domain insight in banking exams, user behavior, and monetization strategy Kunal Pandey — CTO & Co-Founder PhD Candidate (Final Year) at Jamia Millia Islamia with specialization in AI and computational research Published researcher in Physical Review D (Dec 2025), demonstrating deep technical and analytical expertise Leads AI architecture, quiz generation engine, and personalization systems Responsible for building scalable ML-driven learning infrastructure Team Strength Summary Complementary founder skillset: domain + deep tech Built, launched, and monetized product pre-funding Combines firsthand problem understanding with research-grade AI execution |
Vipul Pandey and Kunal Pandey have a long-standing association as siblings, which evolved into a professional collaboration driven by complementary strengths.
The idea for Unmaster originated during Vipul’s time as a Research Scholar at IIT Bombay, where he began exploring inefficiencies in competitive exam preparation based on his prior experience at the State Bank of India. Recognizing the need for a strong AI-driven approach, he partnered with Kunal, who was pursuing a PhD with a focus on advanced computational and AI systems.
They have been actively working together on Unmaster since its inception, taking the product from concept to a live, revenue-generating platform with 4,000+ users and 600+ paid subscribers — built entirely bootstrapped.
Their collaboration combines deep domain insight with research-grade technical execution, enabling fast product iteration and disciplined growth. |
1 |
Complementary execution across domain + deep tech — with proven ability to build and monetize fast
Lived problem + technical depth
Vipul brings firsthand insight from the banking exam ecosystem (ex-SBI), understanding exactly where aspirants struggle
Kunal brings research-grade AI capability (PhD, published in Physical Review D), enabling us to build—not outsource—the intelligence layer
0 → 1 execution without capital
Built, launched, and monetized Unmaster before raising any funding
Achieved 600+ paid users, 2,300 MAU, and ₹8K+ MRR purely through product-led growth
Speed with discipline
Rapid product iteration (Filter → Quiz live; Adapt in progress)
At the same time, strong unit economics (5%+ conversion, 1.9x LTV:CAC) — not growth at any cost
Clear strategic alignment
Shared long-term vision: building the learning decision engine layer, not another content platform
No misalignment on product direction, monetization, or scaling approach
In short: We combine insider understanding of the problem + in-house AI capability + disciplined execution, which allows us to move faster and more efficiently than typical edtech teams. |
UNMASTER |
https://unmaster.in/ |
LUCKNOW |
Unmaster is an AI-powered learning decision engine that tells competitive exam aspirants exactly what to study next, reducing time spent deciding and improving preparation efficiency. |
Competitive exam aspirants face overwhelming content but lack clear guidance on what to study next—resulting in 40–60% of their study time being wasted on decision-making instead of actual learning. |
Unmaster is an AI-powered learning decision engine that helps competitive exam aspirants decide what to study, how to practice, and what to revise next—instead of overwhelming them with content.
Filter: Students select exam-relevant current affairs using date-range and topic filters, eliminating irrelevant content instantly
Quiz: One-click AI-generated quizzes transform filtered content into practice questions
Adapt (in progress): The system analyzes performance (accuracy, response time, topic gaps) to recommend personalized revision and next-best actions
Built for low-bandwidth environments and mobile-first usage, Unmaster reduces content search time by 40–60% and enables focused, decision-driven preparation.
Unlike traditional platforms that deliver content, Unmaster acts as an intelligence layer that guides students on what to focus on—based on their performance and exam goals. |
Unmaster’s defensibility comes from being an intelligence layer—not a content platform—and from data that compounds with usage. Learning decision engine (not content delivery) While platforms like Testbook or Unacademy deliver content, Unmaster decides what a student should study next based on performance. This shifts the value from content volume to decision accuracy. Data moat from user behavior Every quiz attempt, topic selection, accuracy, and response time feeds into the system. Over time, this creates a proprietary dataset on learning patterns—making recommendations smarter and harder to replicate. AI tightly integrated into the workflow AI is not a feature layer—it powers the core loop (Filter → Quiz → Adapt). This ensures continuous improvement in personalization rather than static content consumption. Product-led growth with validated unit economics 5%+ free-to-paid conversion and ₹8K+ MRR achieved organically indicate real user value—not marketing-driven growth. This reduces dependency on paid acquisition. Scalable across exams without linear cost The same AI engine can extend to SSC, UPSC, Railways, etc., without proportional content expansion—unlike traditional edtech models. In essence: the moat is not content—it is a compounding intelligence system built on real user data and behavior. |
Revenue |
Revenue |
Students preparing for government exams (Banking, SSC, etc.) who face content overload and need clear guidance on what to study next. |
150000000 |
350000000 |
18000000 |
Subscription based |
Testbook, Adda247, Affairs Cloud |
**Primarily product-led, organic acquisition with minimal paid spend** * **Organic search (SEO):** High-intent users discovering current affairs via Google * **Direct traffic (46.7%):** Returning users driven by daily-use utility and habit formation * **YouTube (9.2%):** Educational content funneling users into the product * **Telegram communities & word-of-mouth:** Peer sharing among aspirants **Why it works:** * Clear utility (saves 40–60% study time) drives **self-discovery and sharing** * 5%+ free-to-paid conversion validates strong product-market fit without heavy marketing spend **Going forward:** * Scale SEO + YouTube + partnerships while maintaining a **product-led growth engine** rather than ad-heavy acquisition. |
**Product-led, exam-cycle aligned go-to-market** * **Start with a wedge (Current Affairs):** High-frequency, daily-use need → drives habit formation and repeat usage * **Organic acquisition first (validated):** SEO (high-intent searches), YouTube, and Telegram communities → low CAC, scalable channels * **Convert through utility, not push:** Free users experience immediate value (Filter + Quiz) → **5%+ conversion to paid** * **Exam-cycle amplification:** Increase acquisition and engagement during peak exam periods when intent is highest * **Retention via daily workflow:** Product becomes part of the student’s routine (3–4 sessions/week), improving LTV * **Expand horizontally with same engine:** After Banking (validated), extend to SSC, UPSC, Railways without linear content cost **In essence:** Acquire organically → convert through product value → retain via daily habit → expand across exams using the same AI engine. |
**To become the intelligence layer for exam preparation in India.**
Unmaster aims to evolve from a current affairs tool into a full-stack **AI learning decision engine** that guides every aspirant on **what to study, when to study, and what to skip**—across Banking, SSC, UPSC, Railways, and beyond.
* **Short term (0–2 years):**
Build deep personalization — adaptive quizzes, revision planning, and performance tracking
* **Mid term (3–5 years):**
Expand across all major exam categories using the same AI engine, becoming a daily-use platform for millions of aspirants
* **Long term:**
Become the **default decision layer** in education — where content is abundant, but intelligent guidance is the real value
**End goal:** Shift edtech from content consumption to **data-driven learning decisions at scale.** |
LLP |
NA |
Yes |
We are applying to Avinya (ACSEL Accelerator) to accelerate the **AI and personalization layer of Unmaster** with the right ecosystem support. * **DeepTech-aligned mentorship:** As a learning decision engine, our core challenge is building adaptive intelligence at scale. ACSEL’s focus on AI-driven startups aligns directly with our roadmap (adaptive difficulty, personalization, multilingual NLP). * **Structured scale-up support:** We have already validated demand (**600+ paid users, ₹8K+ MRR**) in a bootstrapped setup. We now need guidance to transition from early traction to a scalable, high-retention system. * **Access to networks & partnerships:** Strategic connects (academia, industry, and distribution) will help accelerate expansion into SSC/UPSC and institutional channels. * **Execution-focused ecosystem:** We are not at an idea stage—we have a live, revenue-generating product. ACSEL can help us **refine, scale, and operationalize** what is already working. **In short:** We are applying to move from a validated product to a **scaled AI-driven learning system**, with the right mentorship and ecosystem support. |
**Clear objective: transition from early traction to a scalable AI-driven learning system** During the programme, we aim to: * **Build and deploy the core AI layer (v1):** Adaptive quiz engine + personalized revision system based on accuracy, response time, and topic-level performance * **Scale paid users from 600 → 1,500+:** Strengthen product-led growth and improve conversion beyond the current 5%+ baseline * **Improve retention and engagement:** Move from a utility tool to a **daily habit product** with stronger session frequency and learning loops * **Launch next exam vertical (SSC):** Expand beyond Banking using the same AI engine without linear content cost * **Refine monetization:** Transition from ₹59 validation pricing to a sustainable ₹99–129 plan with better ARPU **Outcome:** A **repeatable, scalable growth engine** combining AI-driven personalization, strong retention, and predictable revenue growth. |
Yes |
Yes |
Core engine |
**Modular, data-driven AI architecture designed around a closed learning loop (Filter → Quiz → Adapt)**
**1. Input & Structuring Layer (Content Intelligence)**
* Ingests current affairs data from multiple sources
* NLP pipeline for **topic classification, tagging, and normalization** (exam-relevant taxonomy)
* Outputs structured, queryable content mapped to exam syllabus
**2. Quiz Generation Engine (Live AI Layer)**
* Transforms structured content into MCQs using controlled LLM pipelines
* Ensures **fact-grounded, exam-oriented question generation**
* Difficulty tagging based on content complexity and question structure
**3. User Interaction & Data Layer**
* Captures granular signals: **accuracy, response time, topic selection, frequency of attempts**
* Builds a continuously evolving **user performance profile** at topic and subtopic level
**4. Personalization Engine (In Progress)**
* Recommendation system that determines:
* What to revise
* What to skip
* What difficulty level to attempt next
* Combines rule-based logic (initial cold start) with ML models (as data scales)
**5. Feedback Loop (Core Moat)**
* Every interaction feeds back into the system
* Improves **question quality, difficulty calibration, and recommendation accuracy** over time
**6. Infrastructure & Design Principles**
* Modular microservices architecture (content, quiz, user modeling, recommendation)
* Optimized for **low latency and low-bandwidth environments**
* Scalable across exams without linear increase in content cost
**In essence:**
A **closed-loop AI system** where content → questions → user behavior → recommendations → improved learning outcomes—continuously compounding with data. |
**A growing, behavior-driven learning dataset that compounds with every user interaction**
* **Granular learning behavior data (core asset):**
We capture topic-level accuracy, response time, quiz attempts, frequency, and revision patterns—mapped to specific exam categories (Banking currently).
* **Content ↔ performance mapping:**
Each piece of current affairs content is linked to how users perform on it (questions, difficulty, retention), creating a **feedback loop between content and learning outcomes**.
* **Decision-layer data (not just consumption data):**
Unlike content platforms that track views, we track **what users chose to study, what they skipped, and what actually improved performance**—this is far more valuable for personalization.
* **Cold-start advantage already solved for our niche:**
With **4,000+ users and 600+ paid subscribers**, we already have meaningful early-stage data to train and refine recommendations.
* **Compounding moat:**
As usage grows, recommendations improve → better outcomes → higher retention → more data. This flywheel makes the system increasingly difficult to replicate over time.
**In essence:**
Our proprietary advantage is not static content—it is a **continuously improving dataset of how aspirants learn, perform, and make study decisions.** |
**Defensibility comes from tightly coupled data + workflow + continuous learning—not just AI models**
* **AI embedded in the core workflow (not a feature):**
Our system operates as a closed loop (**Filter → Quiz → Adapt**). Competitors can add AI features, but replicating an end-to-end decision engine integrated into daily usage is significantly harder.
* **Proprietary behavioral dataset (decision-layer data):**
We capture **what users study, skip, get wrong, and how fast they respond**—mapped to topics and difficulty. This goes beyond content consumption data and directly powers personalization.
* **Content → question → performance feedback loop:**
Every interaction improves **question quality, difficulty calibration, and recommendations**. This creates a compounding system where accuracy improves with scale.
* **Cold-start advantage already in motion:**
With **600+ paid users and consistent engagement**, we are already generating structured learning data—new entrants start with zero signal.
* **System-level moat, not model-level:**
LLMs are accessible to everyone; our edge is in **how data is structured, how feedback loops are designed, and how decisions are generated**—which is harder to copy than the model itself.
* **Scalable without content dependency:**
The same AI engine extends across exams without linear content cost, making the system more efficient and harder to compete with using traditional content-heavy approaches.
**In essence:**
The defensibility is not the AI model—it is the **compounding intelligence system built on proprietary user behavior and tightly integrated learning loops.** |
**We evaluate performance at three levels: output quality, user learning outcomes, and system reliability**
**1. Model / Output Quality (Quiz Generation & NLP)**
* **Question accuracy rate:** % of factually correct, error-free questions (manual + sampled audits)
* **Relevance score:** Alignment with exam syllabus and topic tagging
* **Difficulty calibration:** Distribution of easy/medium/hard vs expected levels
* **Regeneration rate:** % of questions discarded/regenerated (proxy for quality issues)
**2. Learning Effectiveness (Core Metric Layer)**
* **Accuracy improvement over time (per user, per topic)**
* **Retention signals:** Repeat attempts on weak topics, revision completion rates
* **Response time trends:** Faster correct answers = improved mastery
* **Session frequency:** 3–4 sessions/week during exam cycles (engagement proxy)
**3. Recommendation Engine Performance (Adapt Layer)**
* **Recommendation acceptance rate:** % of suggested quizzes/revisions users follow
* **Post-recommendation improvement:** Accuracy lift after following suggested actions
* **Topic mastery progression:** Movement from weak → strong categories
**4. System Reliability & Consistency**
* **Latency:** Time to generate quizzes (must remain near-instant)
* **Failure rate:** % of broken/irrelevant outputs
* **Consistency checks:** Same input → stable, non-contradictory outputs
**5. Business Validation (Ground Truth Check)**
* **Free-to-paid conversion (5%+)** and **retention** act as real-world validation of model usefulness
* Revenue and engagement ensure the system is not just technically correct—but actually valuable
**In essence:**
We don’t evaluate AI in isolation—we measure whether it **improves learning decisions, user outcomes, and retention in a live system.** |
**Privacy-first, minimal-data architecture with secure, compliant infrastructure**
* **Data minimization by design**
We only collect what is essential for personalization (quiz responses, performance metrics). No unnecessary personal or sensitive data is stored.
* **Secure storage & transmission**
* All data is encrypted in transit (HTTPS) and at rest
* Role-based access control (RBAC) ensures only authorized access to systems
* Regular backups and secure cloud infrastructure (AWS/GCP-standard practices)
* **User data isolation & protection**
* Logical separation of user data to prevent cross-access
* Strict internal access policies for founders and systems handling production data
* **Compliance-ready approach**
* Aligned with India’s **Digital Personal Data Protection (DPDP) Act** principles
* Clear user consent for data collection and usage
* Ability to delete/export user data on request (data rights readiness)
* **AI safety & data handling**
* No exposure of personally identifiable data to external AI systems unnecessarily
* Controlled pipelines for content-to-question generation (no sensitive data dependency)
* **Scalable security mindset**
As we grow, we plan to implement **audit logs, automated monitoring, and periodic security reviews** to meet enterprise-grade standards
**In essence:**
We treat user data as a **core asset that must be protected**, ensuring privacy, security, and compliance are built into the system—not added later. |
**Low direct regulatory risk, with manageable compliance considerations**
* **No licensing dependency (core advantage)**
Unmaster operates as a learning tool, not a certified education provider or degree-granting institution—so there are **no strict regulatory approvals required** to operate.
* **Data privacy compliance (primary area)**
Must align with India’s Digital Personal Data Protection Act, 2023
* User consent, data minimization, and storage safeguards
* Manageable with current architecture and policies
* **Content & misinformation risk**
As we generate quizzes from current affairs, ensuring **factual accuracy** is critical
* Mitigated via structured pipelines and validation layers
* **AI regulation (emerging, not restrictive yet)**
Potential future guidelines around AI transparency or accountability
* Our system is assistive (not high-risk), so exposure is limited
* **Platform dependency risks (low)**
Limited reliance on third-party platforms reduces exposure to policy changes
**In essence:**
There are **no structural regulatory barriers**, only standard compliance requirements (primarily data privacy and content accuracy), which are already manageable and built into our system design. |
At 10x scale, the system won’t fail at the model level—it will be stressed at the **data, personalization, and infrastructure layers**. We are already designing for these constraints.
**1. Personalization quality dilution (signal vs noise)**
* With a larger, more diverse user base, weak or inconsistent signals can reduce recommendation accuracy
* **Mitigation:** stronger user segmentation, topic-level embeddings, and hybrid models (rule-based + ML) to maintain precision
**2. Cold-start problem at scale**
* New users across multiple exams (SSC, UPSC, etc.) → limited initial data for personalization
* **Mitigation:** default learning paths + cohort-based initialization until sufficient user data is generated
**3. Difficulty calibration drift**
* As question volume increases, maintaining consistent difficulty levels becomes harder
* **Mitigation:** continuous calibration using aggregate performance data and feedback loops
**4. Infrastructure & latency bottlenecks**
* Real-time quiz generation and recommendations may face latency spikes under heavy load
* **Mitigation:** caching, pre-generation of quizzes, and scalable cloud architecture (auto-scaling, load balancing)
**5. Data pipeline complexity**
* Increased ingestion → higher risk of noisy or misclassified content affecting downstream outputs
* **Mitigation:** stricter validation layers and automated quality checks in the NLP pipeline
**6. Model evaluation & monitoring gaps**
* At scale, silent failures (e.g., subtle drops in recommendation quality) become harder to detect
* **Mitigation:** robust metric tracking (accuracy lift, recommendation acceptance rate) and automated alerts
**7. Cost vs performance trade-offs**
* AI inference and compute costs can rise significantly with scale
* **Mitigation:** optimize pipelines, mix of lightweight models + selective LLM usage
**In essence:**
Nothing fundamentally “breaks”—but without strong data discipline and infrastructure, **personalization accuracy and system responsiveness would degrade**. Our architecture is being built to ensure both **scale and quality improve together, not trade off.** |
Yes — AI/ML is built in-house and led by the founding team
Kunal Pandey (CTO & Co-Founder)
PhD Candidate at Jamia Millia Islamia with specialization in AI and computational research
Published in Physical Review D (Dec 2025)
Leads architecture of quiz generation, NLP pipelines, and personalization systems
Core expertise areas:
NLP for content classification, tagging, and structuring (exam-oriented taxonomy)
LLM-based question generation with controlled output pipelines
User modeling and recommendation systems (accuracy, response time, topic-level performance)
Designing closed-loop learning systems (Filter → Quiz → Adapt)
Execution approach:
AI is developed as part of the product core, not outsourced
Modular architecture enabling iterative improvements as data scales
In essence:
We have founder-level AI capability, ensuring deep control over the core technology and faster iteration compared to teams dependent on external AI resources. |
We use a combination of structured public data and proprietary user-generated data
1. Content Dataset (Input Layer)
Current affairs data sourced from publicly available, exam-relevant information
Processed through our in-house NLP pipeline (classification, tagging, structuring)
Status: Public/unlicensed raw data → transformed into structured internal dataset (owned layer)
2. Proprietary User Dataset (Core Asset)
Quiz responses, accuracy, response time, topic-level performance, and usage patterns
Mapped to specific content and difficulty levels
Status: Fully owned, continuously growing dataset (key moat)
3. Derived AI Datasets (Generated Layer)
AI-generated question bank with difficulty tagging and performance calibration
Content-to-question mappings and topic-level embeddings
Status: Proprietary, built in-house from our pipelines
4. Model Layer
Use of foundational models (LLMs) for generation tasks
Status: Licensed via API providers, with our own control layers and pipelines on top
In essence:
Raw input data: Public
Structured content + question bank: Proprietary (processed)
User behavior data: Fully owned (core advantage)**
This combination ensures that while inputs are accessible, the value and defensibility lie in our transformed datasets and user interaction data. |
**Closed-loop learning system with continuous data-driven improvement**
* **Feedback loop from real users (core driver):**
Every quiz attempt captures accuracy, response time, and topic gaps → feeds directly into model refinement and recommendation tuning
* **Continuous evaluation & recalibration:**
* Monitor question accuracy, difficulty distribution, and regeneration rates
* Recalibrate difficulty and tagging using aggregate performance data
* **A/B testing on live users:**
Test different recommendation strategies, quiz formats, and difficulty levels → optimize for accuracy improvement and retention
* **Hybrid modeling approach:**
Start with rule-based systems (for stability and cold start) → progressively layer ML models as data scales
* **Data enrichment over time:**
Expand across exams (SSC, UPSC, etc.) → broader datasets improve generalization and personalization
* **Human-in-the-loop validation (targeted):**
Sample audits of generated questions and outputs to maintain factual accuracy and exam relevance
* **Infrastructure for iteration:**
Modular architecture allows independent upgrades to NLP, quiz generation, and recommendation layers without system disruption
**In essence:**
Model performance improves through a **compounding feedback loop of user data → evaluation → refinement → better recommendations → more usage → more data.** |
**India-first, with a scalable global model**
* **Primary focus: India**
Targeting the ₹15,000 Cr government exam market (Banking, SSC, UPSC, Railways), where we already have validated traction (600+ paid users).
* **Why India first:**
* Massive, underserved aspirant base (2.5 crore+)
* High inefficiency (content overload, lack of guidance)
* Strong willingness to pay for outcomes
* **Global potential (long term):**
The core product is not content—it’s a **learning decision engine**, which can be applied to:
* Standardized tests (e.g., GRE, GMAT)
* Other exam-driven or skill-based learning markets
**In essence:**
We are building for **India now**, but designing a **globally scalable intelligence layer for learning.** |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/coSrcuTpYISA-UNMASTER-Generic-Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/coSrcuTpYISA-UNMASTER-Generic-Pitch-Deck.pdf |
https://youtu.be/uKZptg6MkYw?si=Cej3sKInKCkn021- |
**Yes — Unmaster is a mission-driven startup focused on improving learning outcomes at scale**
* **Core mission:**
Reduce inefficiency in exam preparation by helping students focus on *what actually matters*—not just consume more content
* **Problem with real impact:**
2.5 crore+ aspirants in India prepare for government exams, and **80%+ fail not due to lack of effort, but lack of direction**
* **Impact we create:**
* Reduce 40–60% of time wasted on deciding what to study
* Enable structured, personalized preparation for students without access to coaching
* Improve confidence and outcomes, especially for Tier 2/3 aspirants
* **Accessibility by design:**
* Affordable pricing (₹59/month)
* Mobile-first, low-bandwidth product → usable across India
* **Scalable impact through AI:**
Instead of relying on expensive coaching models, we build a system that delivers **personalized guidance at scale**, making quality preparation more accessible
**In essence:**
We are building not just a product, but a system that **democratizes effective exam preparation by replacing guesswork with data-driven decisions.** |
No. However, we come from a Tier-2/3 background and non-elite access pathways, and have built Unmaster fully bootstrapped—from problem discovery to a revenue-generating product—without relying on privileged startup networks or capital. |
N.A |
checked |
| Apr 22, 2026 @ 7:29 AM |
sunil punjabi |
sunil.punjabi@onicares.com |
https://www.linkedin.com/in/oneoni |
http://Na |
9833777765 |
CEO, CTO |
We met at Antler Cohort 6. Been working together now for a year. |
2 |
As a team we not only cover all vertical of a business, we have a strong emotional bond. Thats what keeps us alligned. |
ONI |
NA |
Bengaluru |
98% of pregnancy today is clinically unmonitored; ONI is worlds 1st AI ObGyn "Continuum Pregnanncy Care" helping improving feotal health and labour outcomes |
98% of pregnancy is clinically unmonitored, and today 1out of 2 mothers are high risk pregnancies. Health outcomes are also poor, 2 out 3 newborns have growth shunting. |
ONI delivers ObGyn-led, continuous support during & between consultations — enabling early detection and personalised intervention during critical developmental windows.
Much like IVF productised fertility, CPC - continuous pregnancy care has the potential to create a category at significantly larger scale, driven by:
• Predictable, recurring volume
• Low incremental capex
• Strong clinical and emotional engagement
• Long-term family-level brand and lifetime value
This is the category ONI is building toward — with clinical intelligence, voice-led workflows, and epigenetically informed recommendations forming the foundation. |
Our solution focus on supporting the mother clinically, but through the ObGyn |
Users |
Users |
Pregnant Mothers |
10,000 Cr |
5,000 Cr |
3,000 Cr |
Subscription |
Currently, no direct competitors |
we onboard obgyns, they inturn onboard mothers |
Wrok with large EMR cos like Akhil & Shivam which have 1,100 hospitals |
Our long vision is to build Women OS - from pre-conception (gynaecology, preconception, fertility) to Pregnancy to Postpartum to first 10 years of human life. |
Private ltd |
Na |
Yes |
Capital and mentorship |
scale |
open to both |
Yes |
Core engine |
We are building an AI layer on top of a structured clinical system.
We use foundational LLMs to understand a mother’s profile including obstetric history, medical history and risk parameters, and generate clinical pathways. This includes consultation schedules, antenatal tests by week, and what needs to be tracked during each visit.
All clinical data is stored in a structured format, and the AI works on top of this instead of raw text. For reports, we use an OCR pipeline to extract values and standardise them into our schema.
On the app side, we use voice AI as an interaction layer so both doctors and mothers can interact using voice along with normal UI.
Over time, we are building predictive models using our own data to identify risks early at different stages of pregnancy. |
Right now we are building the data pipeline.
Once deployed, we will have longitudinal pregnancy data including clinical records, test results, lifestyle inputs and consultation patterns across the full pregnancy journey.
This kind of structured, time-series pregnancy data is not easily available and becomes our core advantage after scale. |
Our defensibility comes from three things:
First, structured clinical data collected across the full pregnancy lifecycle.
Second, continuous interaction data from both doctors and mothers, including voice inputs.
Third, domain-specific pathways and workflows built into the system.
Over time, our predictive models will be trained on this proprietary dataset, which makes it hard to replicate. |
For LLM based outputs:
Clinical correctness validated with doctors,
Adherence to defined medical protocols,
Consistency across similar inputs
For OCR:
Extraction accuracy,
Field level validation against expected ranges,
For predictive models (once deployed):
Precision and recall for risk detection,
False positive and false negative rates,
Outcome correlation with actual clinical events,
We also keep human-in-the-loop validation, especially in early stages. |
Data is encrypted at rest and in transit.
Access is controlled using role-based access.
We support separation of identity and clinical data where required.
We are aligning with HIPAA and DPDP requirements.
Hospitals or doctors remain the data owners, and we act as the processor. |
Yes, since we are working in healthcare, any AI-driven clinical recommendations can come under regulatory scrutiny depending on the market.
We are positioning the system as a clinical support tool and not replacing doctor decision making.
As we expand globally, we will need to align with local healthcare regulations and data protection laws. |
The main challenges will be:
Maintaining consistency and quality of LLM outputs across large volumes.
Handling latency for real-time interactions like voice.
Ensuring data quality across different hospitals and geographies.
We are designing the system to be modular so models, pipelines and infra can scale independently. |
Yes, we have a small in-house team working on applied AI and backend systems.
Our expertise includes LLM integration, building agentic workflows for clinical pathways, structured data systems, OCR-based report parsing, and voice interaction layers.
We also build end-to-end data pipelines and scalable backend systems.
We will expand into dedicated ML roles as we scale. |
Currently we rely on:
Public medical guidelines and protocols for defining pathways.
Open source models for OCR and LLM capabilities.
We do not rely on any third-party proprietary patient datasets.
Our long-term focus is to build our own dataset from usage. |
We improve through usage.
We collect structured data from consultations, reports and outcomes.
We use doctor feedback to refine outputs.
We monitor errors and edge cases and update prompts, pipelines and models.
Once we have enough data, we will train and fine-tune our own models for better accuracy. |
We are building for global markets.
Right now we are focused on India to validate and scale the product.
The system is designed in a way that it can adapt to different geographies, protocols and regulations over time. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/rWTRf7S6QZhA-ONI-Investments-13-Apr-2026-AI.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/rWTRf7S6QZhA-ONI-Investments-13-Apr-2026-AI.pdf |
https://www.youtube.com/watch?v=eLpAo4rrHBY |
Our focus is Maternal and Child Health - SDG 3 - our clear goal is to impact health trajectories at the time of development, for a better society |
NA |
NA |
checked |
| Apr 21, 2026 @ 10:31 PM |
Sonika , KS Cheema |
sonika@digitruce.com |
http://www.linkedin.com/in/sonikatanwar1 |
http://www.digitruce.ai |
9607089012 |
CEO - Indian Airforce Veteran in Cybersecurity & Automation, MD - leading Entrepreneur |
We met while doing a Cybersecurity audit. Both wanted to work and build indigenous OT security systems and started talking about it. The dream realised in form of DigiTruce.ai |
two |
As a team - we are true to our core values- integrity, honesty and dedication.
Together- we can move mountains and are always ready for the mission. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 21, 2026 @ 4:04 PM |
Archana Parameswaran, Sundar Raj |
thefragmentapp@gmail.com |
https://www.linkedin.com/in/archanaparameswaran/ |
https://www.thefragment.app/ |
9108027243 |
Co founders, IT Product Management, Development, Deployment |
We met in our previous job, we were in the same team. We have worked for about 5 years counting the old company also. |
1 |
Our biggest strength is our resilience, clear vision for the product and our combined technical experience of 25+, which we bring along across different platforms like SaaS, IoT, web and mobile applications, fintech, and edtech. We’ve built and scaled products at every stage—from zero to launch, as well as improving and operating live systems. |
Fragment (from Texarch Private Limited) |
https://www.thefragment.app/ |
Bangalore, India |
Fragment is an AI powered zero set up scheduling platform, |
Skilled organizers juggle fragmented tools—complex, disconnected, requiring technical setup for scheduling, communication, and payments that lead to drop-offs. |
A system built to run multiple events across groups, aiming to include all organizers in the digital economy — regardless of technical skills. |
Fragment creates an "operational moat" by unifying the entire lifecycle of recurring logistics—AI scheduling, embedded fintech, and real-time communication—into a single, automated platform. Unlike static event tools, our defensibility lies in the deep vertical integration of these workflows, making Fragment the indispensable operating system that organizations cannot extract themselves from without disrupting their entire daily execution. |
MVP |
Users, Pilots, Signups |
Individual organizers, coaches, group event organizers. |
7392 Cr |
3304 Cr |
1 Cr |
Subscription, Booking Transactions, AI token, Market place |
Eventbrite, Luma, Meetup, WhatsApp, Google Sheets. |
App store downloads, reach out through linked in and content creation |
Demo videos, approach organizers, content videos |
Our vision is for Fragment to become a scheduling platform for India's 50M+ solopreneurs and expand into a full operating system for institutions and organizations with group and recurring events. Eventually expand internationally. |
Our company was incorporated in 2023 |
none |
Yes |
Stronger product positioning, help with market expansion, and mentorship in technology and marketing. |
Product Market Fit and initial paying customers. |
We would prefer hybrid |
Yes |
Automation layer |
We are implementing Agentic AI, to automate non conflicting scheduling for the users and eventually will implement more generative AI as well. |
Having both business users and their clients, our data would be more around the operational context and the lifecycle of the long term recurring event, payments and communication. |
We are enabling AI for our users for specific tasks and automations, this will be custom build and will not be useful or applicable beyond our application. Therefore this is not a competition to other AI companies especially if they are AI centric products. |
Our metrics would be to validate the AI's accuracy on the math and calendar logic and see, how often the AI gets it right without human involvement. |
With the payment gateways enabled, we will ensure there is the fintech grade security along and have a localized data residency. |
The application sits at an intersection of Fintech, Personal Data, and AI automation. We must comply with DPDP, RBI guidelines to digital payment and AI accountability. |
1. At a 10X growth, there can be latency, at a large scale we experience slow responsiveness and will be computationally expensive.
2. We may have to move to Small Language Models (SLMs) for 80% of routine tasks (updates, simple shifts) and reserving the "Large" models only for initial setup and complex conflict resolution.
3. Also at the scale we may need an enterprise style of use cases where it must role based access and ensure security for each users data within their own organization. |
Yes, we have an AI in-house developers. Our developers build AI agents (a combination of Agents with the AI models available). |
We are primarily working on the Agentic AI for automation at the moment. The data is from our own data base. |
1. We may have to improve on the operational rules not just storing it the choices and responses as memory.
2. We must enable a lot of security around the AI that it does not touch the database directly but the data is enable through a middle man API.
3. We may have to implement SLMs for routine tasks.
4. Enablement of role checks. |
We are building for India first and will expand to the global markets. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Xx6O8XsXg92a-Fragment-Pitch.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Xx6O8XsXg92a-Fragment-Pitch.pdf |
https://play.google.com/store/apps/details?id=com.thefragmentapp |
We are impact focused, we are building the foundational infrastructure that allows any mission-driven organizers/ organization to run their operations digitally to succeed at scale. |
No |
N/A |
checked |
| Apr 21, 2026 @ 2:03 AM |
Suchandana Ghoshal, Sougam Maity |
avistha2024@gmail.com |
http://www.linkedin.com/in/sougam/ |
https://avistha.com/%20 |
9932242598 |
Founder and CTO, Btech and MTech and Founder and CEO, BTech |
|
two |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 20, 2026 @ 8:34 PM |
Archit Shervi and Aditeya JC |
architshervi6@gmail.com |
http://www.linkedin.com/in/archit-shervi-433a1222b |
http://NA |
7259870616 |
Engineer and Lawyer |
We met in college and we have worked together over 5 years on multiple projects and hustles, we have worked on 2 startups before this, one of which was scaled to 30L ARR, while being bootstrapped, it was an entertainment space startup. |
one |
Understanding between the both of us and the communication amongst us. As a team we are also very consitant, as there has never been a time or a gig for which neither of us have showed up, we always work it out in a way where both of us or atleast one of us are avilable. |
Lawlet Ai |
NA |
Bangalore |
Lawlet AI is a legal-tech platform designed to streamline legal research, document analysis, and compliance for both students and professionals through AI-driven automation. |
We are solving the gross inefficiency and cognitive overload inherent in traditional legal education and research. |
Lawlet AI is the intelligent infrastructure for the next generation of legal professionals. We are moving the legal industry from search-based workflows to intelligence-based workflows.
While the current legal-tech landscape focuses on static databases, Lawlet AI provides a dynamic, AI-powered ecosystem that functions as a:
Synthesized Intelligence Layer: We don’t just find cases; we distill them. Our proprietary engine compresses thousands of pages of complex jurisprudence into high-signal summaries, reducing research time by 80%.
Tactical "Flight Simulator": We are closing the gap between academic theory and courtroom reality. Our AI-driven moot court simulations provide a low-risk, high-frequency training ground for students to develop oral advocacy and rebuttal skills—data points that traditional education ignores.
The Professional "Second Brain": Lawlet acts as a persistent knowledge graph for the user, organizing research, managing compliance, and identifying precedents that human eyes miss.
Why Now?
The legal industry is at a breaking point where the volume of data has outpaced human cognitive capacity. Lawlet AI isn't just an "assistant"—it is the necessary leverage required to practice law in an AI-first world. We are turning a labor-intensive bottleneck into a scalable, high-margin software advantage.
The Edge
We aren't just selling a tool; we are capturing the user’s workflow early in their career (as students) and scaling with them as they enter the workforce, creating a high-LTV (Lifetime Value) ecosystem with built-in switching costs. |
The Moat: High Switching Costs The "Student-to-Partner" Pipeline: We capture users at the university level. By graduation, Lawlet holds five years of their personal research, notes, and logic. Moving to a competitor isn't just a software change; it’s a lobotomy of their professional memory. Workflow Integration: We aren't a "tab" they open; we are the operating system where they summarize (FIRAC), practice (Moot Court), and track (Live Cause Lists). The Uniqueness: Specialized Intelligence Proprietary Data Synthesis: General AI hallucinates law. Our engine forces LLMs through a structured legal framework (Facts, Issue, Rule, Analysis, Conclusion), delivering zero-fluff, high-accuracy intelligence. The Training Edge: Our AI Moot Court is a proprietary "flight simulator." It builds a unique dataset on advocacy and rebuttal that generic models can't replicate. |
two |
one |
Primary (B2C Entry): Law Students (Undergraduate/Postgraduate). Focus: Academic survival and practical skill-building (Moot Court). Logic: Low CAC (Customer Acquisition Cost) via campus networks and a 5-year high-retention lifecycle. Secondary (B2B/Professional): Early-Career Practitioners & Boutique Firms. Focus: Research efficiency and document synthesis. Logic: High LTV (Lifetime Value) as students transition into the workforce, carrying their proprietary "Second Brain" (data) with them. Institutional: Law Universities. Focus: Practical training infrastructure. Logic: Large-scale licensing for specialized training modules (AI-driven simulations). The "Why Now?" (Market Pain) The legal sector is experiencing a data-to-human capacity mismatch. The volume of judicial output has exceeded human cognitive limits; Lawlet AI provides the necessary "intelligence layer" to bridge this gap, transforming a labor-intensive industry into a software-leveraged one. |
TAM (Total Addressable Market) ₹3.15 Lakh Crore ($38.1 Billion) This is the Global Legal Tech Market value. It represents the total worldwide potential for AI-driven legal automation, research, and infrastructure that Lawlet AI can eventually scale into. |
SAM (Serviceable Available Market) ₹10,000 Crore (~$1.2 Billion) This is the Indian Legal Services & Tech Market for students and early professionals. Basis: India has ~1.5 million registered lawyers and ~500,000 active law students. With a 13.8% CAGR in legal tech adoption and the rapid digitization of Indian courts (e-courts), this segment is the immediate pool of value Lawlet AI can serve. |
SOM (Serviceable Obtainable Market) ₹150 Crore - ₹200 Crore This is our 3-Year Revenue Target based on capturing the "High-Intent" segment. The Math: Targeting the top 150,000 law students (NLUs and premium private colleges) and 50,000 junior litigators in Tier 1 cities. Pricing Logic: At an average annual subscription (ARPU) of ₹7,500 – ₹10,000, capturing just 10% of this beachhead results in a ₹150Cr+ annual opportunity. |
Revenue Model: The "Growth-Link" Strategy We employ a multi-tiered subscription model designed to capture users early and increase extraction as they move into high-value professional roles. 1. B2C: The Student Tier (The "Beachhead") Model: Monthly/Annual SaaS Subscription. Price Point: ₹499/month or ₹3,999/year. Target: Law students in India (500k+ active). Strategic Value: This is our low-CAC entry point. We solve immediate academic pain (exams/moot courts) to build the data moat. 2. B2C: The Professional Tier (The "Retention" play) Model: Premium SaaS Subscription. Price Point: ₹1,499/month or ₹12,000/year. Target: Junior associates and independent litigators. Strategic Value: As our student users graduate, they "upgrade" to professional features (Live Cause Lists, Client Management, Advanced Precedents). Because their last 5 years of research are locked in Lawlet, retention is nearly 100%. 3. B2B: Institutional Licensing (The "Infrastructure" play) Model: Annual Per-Seat Licensing. Target: Law Universities and Tier-1 Law Firms. Value Prop: Law schools pay for the AI Moot Court "Flight Simulator" to improve student placement rates. Law firms pay for the "Intelligence Layer" to reduce the billable hours wasted on junior-level manual research. 4. The "Value-Add" Marketplace (Future Expansion) Model: Transactional/Commission-based. Feature: Certified Legal Templates and "Lawyer-to-Client" lead generation. Logic: Once we own the lawyer's workflow, we monetize the peripheral services they require. |
Legacy Giants (Indirect): SCC Online, Manupattra, LexisNexis. The Gap: Built as static "libraries." High friction, high cost, and zero intelligence layer. They ignore the student workflow and focus on senior partners. AI Wrappers (Direct): Generic "Chat with PDF" startups and GPT-based legal tools. The Gap: No structural guardrails. Prone to legal hallucinations. Lacks integration with Indian Bare Acts and real-time Court Cause Lists. EdTech Platforms: LawSikho, EBC Learning. The Gap: Human-led and content-heavy. High-cost, low-scale training that lacks daily utility or research automation. |
B2C: High-Velocity Campus Acquisition Campus Ambassador Program: We deploy student representatives in India’s top 500 law colleges (NLUs and Tier-1 private institutions). Ambassadors drive peer-to-peer adoption through localized WhatsApp/Telegram communities and campus events. Strategic Partnerships: Collaborating with student-run legal societies and moot court committees to integrate Lawlet AI into official competition preparation. Freemium Funnel: A "Moot Court Starter" tier to capture users instantly, converting them to paid tiers for advanced FIRAC summarization and exam-readiness tools. 2. B2B: Institutional Integration "Digital Infrastructure" Sales: Partnering with law universities to provide Lawlet as the official AI research and training partner. Value Prop for Schools: Improving national ranking and placement rates by providing students with the Lawlet "Flight Simulator" for practical legal training. 3. Professional Migration (The "Switching Cost" Play) Organic Transition: As student users graduate, they carry their pre-existing Lawlet "Second Brain" into firms. This creates a bottom-up adoption pressure within law firms. Alumni Networks: Targeting recent graduates with "Early Practitioner" discounts to ensure Lawlet remains their primary professional tool as they enter the workforce. |
Phase 1: The Beachhead (B2C Student Market) Campus Ambassador Network: Deployment of "Lawlet Leads" in top 500 law colleges (NLUs/Tier-1 Private). Ambassadors drive adoption through peer-to-peer workshops and WhatsApp/Telegram communities. Product-Led Growth (PLG): A Freemium "Moot Court" Tier. Students use our AI simulator for free competition prep, creating an immediate viral loop. Advanced research and FIRAC summarization tools are gated behind a subscription. Strategic SEO/LLMO: Optimizing for "Legal Search Intent." We target long-tail queries related to specific subjects and backlog prep, ensuring Lawlet is the top AI-recommended tool in search engines and LLM responses. 2. Phase 2: Institutional Expansion (B2B) University Partnership Program: Selling Lawlet as the "Digital Infrastructure" for Law Schools. Universities license the platform to provide students with a standard "Flight Simulator" for mandatory practical training. Placement Integration: Positioning Lawlet proficiency as a certification that makes students "Day-1 Ready" for Tier-1 law firms. 3. Phase 3: Professional Retention (The Bridge) The "Second Brain" Lock-in: As students graduate, their entire academic research history remains in Lawlet. Transitioning to a competitor involves a massive loss of personal data and work history. Direct Professional Marketing: Targeted LinkedIn campaigns for Junior Associates, focusing on ROI and Time-Efficiency—positioning Lawlet as a way to reduce billable hour fatigue. |
Our long-term objective is to transition Lawlet AI from a specialized academic tool into the autonomous operating system (OS) for the legal profession. We aim to own the entire professional lifecycle of a lawyer, moving from assisting research to executing complex legal workflows.
1. The Professional "Second Brain" (Years 1–3)
Objective: Achieve 15% market penetration among law students in India.
Outcome: Accumulate a proprietary dataset of user research habits, academic performance, and logic mapping. By the time our first cohort graduates, Lawlet will be their indispensable professional memory, ensuring near-zero churn as they enter law firms.
2. Agentic Legal Automation (Years 3–5)
Objective: Transition from generative AI to Agentic AI.
Outcome: Deploy autonomous agents capable of setting goals and executing end-to-end strategies (e.g., automated due diligence, predictive litigation analytics, and multi-jurisdictional compliance monitoring) with minimal human supervision.
3. Market Dominance & Global Expansion (Year 5+)
Objective: Expand the "India-first" model to other Commonwealth jurisdictions (UK, Australia, Canada) that share similar legal foundations.
Outcome: Become the primary interface for legal practice, where Lawlet is not just a "tab" in a browser but the infrastructure that connects students, practitioners, and the judiciary. |
Incorporated |
NA |
one |
Founder-Market Fit Alignment: Avinya’s focus on "Responsible AI" and "DeepTech" aligns with Lawlet’s mission to build an ethically grounded, structurally sound intelligence layer for the Indian legal system. Access to the IIT Alumni Ecosystem: As an AI-native startup, we seek to leverage Avinya’s deep network of technical mentors and industry veterans from the IIT ecosystem to refine our LLM architecture and scaling strategy. Strategic Mentorship: We are specifically applying to Avinya for its "Mentorship Pods." Transitioning from a student-centric MVP to a B2B institutional OS requires the operational maturity and GTM guidance that Avinya’s leadership provides. Validation & Networking: Being part of a cohort backed by the Government of Karnataka and ACSEL provides the institutional credibility necessary to partner with prestigious Law Universities (NLUs) and Tier-1 legal firms. Mission Alignment: Avinya’s commitment to building "resilient and responsible leaders" resonates with my philosophy of creating long-term, defensible infrastructure rather than short-term AI wrappers |
To achieve our mission of becoming the definitive "Operating System" for the legal mind, we aim to utilize the Avinya programme to reach the following high-stakes milestones: Product-Market Fit (PMF) Validation: Conduct structured pilot programs with at least 5 top-tier Law Universities (NLUs) to stress-test our AI Moot Court simulations and FIRAC-based summarization engine in real-world academic environments. Technical Scalability: Leverage the IIT Alumni Centre’s technical mentorship to optimize our AI architecture, specifically ensuring zero-latency and 100% accuracy when scaling judicial data processing by 10x. GTM Strategy Execution: Refine our "Campus Ambassador" acquisition model into a repeatable, high-velocity engine to achieve a user base of 25,000+ active law students within the cohort period. Fundraising Readiness: Utilize Avinya's Mentorship Pods to bridge the gap between an early-revenue MVP and a venture-scale platform. Our goal is to exit the programme ready to secure a Seed/Pre-Series A round via the Avinya Demo Day network. Institutional Credibility: Secure formal endorsements or strategic partnerships with legal regulatory bodies or Tier-1 law firms, using Avinya’s institutional backing and government ties (Karnataka IT/BT Dept) as a catalyst. |
yes |
one |
one |
For a VC questionnaire, you must move beyond "we use GPT-4." You need to describe a proprietary, multi-layered stack that ensures accuracy and legal defensibility.
Here is your AI architecture, broken down into clinical pointers:
Lawlet AI: System Architecture
We employ a Modular AI Stack designed specifically for high-stakes legal reasoning, where general-purpose LLMs typically fail due to "hallucinations."
Layer 1: The Synthesis Engine (Structured FIRAC Processing)
Unlike generic chat interfaces, our engine utilizes a Cascading Prompt Architecture to force-filter raw judicial text through the FIRAC framework (Facts, Issue, Rule, Analysis, Conclusion). This ensures that the output is not just a summary, but a structured legal brief.
Layer 2: Retrieval-Augmented Generation (RAG) & Grounding
To eliminate hallucinations, our models are "grounded" in our proprietary Indian Legal Knowledge Base. Every insight generated is cross-referenced against verified Bare Acts, statutes, and current Court Cause Lists using a high-density vector database (e.g., Pinecone or Weaviate).
Layer 3: Specialized LLM Fine-Tuning
We utilize Domain-Specific Small Language Models (SLMs) like Legal-BERT or BART, fine-tuned on Indian case law, for extractive tasks (identifying citations and precedents). This is orchestrated with larger models (GPT-4o/Claude 3.5) for complex abstractive reasoning and conversational simulations.
Layer 4: The Simulation Logic (Moot Court Logic)
Our AI Moot Court uses a State-Machine Architecture to manage the flow of oral advocacy. It evaluates user input against a "Scoring Logic" that measures logical consistency, rebuttal accuracy, and adherence to the "Rule of Law," providing data-driven feedback rather than just generic chat.
Layer 5: Data Persistence & Knowledge Graph
We build a User-Specific Knowledge Graph. Every case a student researches or simulates is mapped into their personal graph, creating a persistent "Second Brain" that grows in value over the 5-year academic lifecycle. |
Our proprietary data advantage is built on exclusive interaction data and structured legal datasets that cannot be scraped or replicated by generic LLMs.
Interaction-Based Advocacy Data: Through our AI Moot Court simulations, we are capturing unique datasets on oral advocacy, rebuttal patterns, and logical flow. This "performance data" allows us to fine-tune models on how winning arguments are constructed—a dataset that is currently non-existent in the public domain.
Structured FIRAC Knowledge Graph: We are building a massive, proprietary repository of human-verified, FIRAC-structured case summaries. While general AI has access to raw judgments, Lawlet owns the structured "logic maps" of those judgments, significantly reducing hallucination rates and increasing retrieval precision.
Hyper-Local Regulatory Data: We integrate real-time, often fragmented data from local Court Cause Lists, Government Gazettes, and Regional Bare Acts that are not consistently digitized or indexed by global search engines.
The "Student-to-Professional" Lifecycle Data: By capturing users at the university level, we possess a long-term data history of a professional's research evolution, areas of expertise, and cognitive gaps. This allows for hyper-personalized AI assistance that "knows" the lawyer's professional history better than any third-party tool. |
AI Component Defensibility
To a VC, an LLM API key is not a moat. Our defensibility stems from moving beyond general-purpose inference into specialized, workflow-integrated intelligence.
Here is why our AI cannot be easily replicated by a competitor or rendered obsolete by a foundational model update:
The "Flight Simulator" Data Flywheel: Foundational models (like GPT-4 or Claude) are trained on static public data. Through our AI Moot Court, Lawlet captures dynamic interaction data—how law students form rebuttals, where their logical fallacies lie, and how they structure oral arguments. We are building the only proprietary dataset on legal cognitive behavior, giving our AI an edge in simulating real-world legal combat that generic wrappers cannot match.
Structural Guardrails (The FIRAC Engine): General LLMs hallucinate because they predict the next best word. Lawlet’s AI doesn't just read raw text; it forces judicial data through a strict FIRAC framework (Facts, Issue, Rule, Analysis, Conclusion). We are systematically converting unstructured legal noise into structured, machine-readable logic graphs. Our moat is this growing repository of verified, structured legal logic.
Workflow Lock-In & Personalization: Our AI is not a stateless chat box. Over a 5-year law degree, Lawlet maps the user's specific research patterns, blind spots, and preferred precedents, becoming a hyper-personalized "Second Brain." A competitor offering a slightly faster LLM cannot overcome the switching cost of leaving behind five years of personalized AI context.
Hyper-Local RAG Architecture: We do not rely on the LLM's internal memory for legal accuracy. Our RAG (Retrieval-Augmented Generation) pipeline is grounded specifically in hyper-local Indian legal datasets—Bare Acts, Gazette notifications, and daily Court Cause Lists. This deterministic grounding layer ensures our AI operates with zero-hallucination utility, effectively neutralizing the threat of "smarter" general-purpose AI. |
Model Evaluation & Reliability Metrics
To ensure zero-hallucination legal utility, we evaluate our models far beyond basic NLP benchmarks (like BLEU/ROUGE). We measure deterministic accuracy and workflow efficacy using the following frameworks:
RAGAS Metrics (Retrieval-Augmented Generation Assessment): * Faithfulness Score: We measure how strictly the AI’s generated response is grounded in the retrieved context (Bare Acts, Judgments). A score below 0.99 triggers a manual review. This is our primary anti-hallucination metric.
Context Precision: Evaluates if the most legally relevant paragraphs of a 100-page judgment are prioritized in the retrieval layer.
FIRAC Completeness Rate (Custom Metric): We programmatically parse the generated case briefs to ensure all five structural constraints (Facts, Issue, Rule, Analysis, Conclusion) are populated accurately without omitting critical sub-issues.
Human-in-the-Loop (HITL) Edit Distance: We track how much a user alters an AI-generated summary or Moot Court rebuttal before saving it to their knowledge graph. A low edit distance proves the model is delivering high-signal, "ready-to-use" intelligence.
Latency & Time-to-First-Token (TTFT): Processing massive, unstructured legal PDFs requires heavy compute. We measure latency rigorously to ensure the UX remains frictionless for students during high-pressure exam or competition prep. |
Zero-Retention AI Processing (No Training on User Data): We strictly utilize enterprise-grade LLM APIs with zero-data-retention agreements. User inputs, case briefs, and Moot Court strategies are never used to train external foundational models (like OpenAI or Anthropic). We maintain strict tenant isolation to prevent data bleed between user accounts.
DPDP Act 2023 Compliance & Data Residency: We are built for the Indian regulatory landscape. All data is localized on domestic servers (e.g., AWS/Azure India regions) ensuring strict compliance with the Digital Personal Data Protection (DPDP) Act of 2023. We enforce data minimization and explicit user consent protocols.
Enterprise-Grade Encryption: All data at rest is encrypted using AES-256, and data in transit is secured via TLS 1.3. This ensures that even in the event of an infrastructure breach, the underlying legal notes and intelligence graphs remain cryptographically locked.
Pathway to Institutional Certifications: To secure B2B institutional deals with law universities and Tier-1 firms, our architecture is designed from Day 1 to map against SOC 2 Type II and ISO 27001 compliance frameworks.
Granular Auditability & RBAC: We implement strict Role-Based Access Control (RBAC) and maintain immutable audit logs for all data interactions. This provides a clear chain of custody, ensuring that institutions can trust Lawlet with sensitive academic and pre-litigation data. |
The 2026 IT Amendment Rules (SGI Compliance)
The Risk: The February 2026 amendments to the IT Rules enforce brutal compliance windows for Synthetically Generated Information (SGI). Platforms must label all AI-generated content and comply with 3-hour takedown notices for unlawful synthetic media, risking the loss of Section 79 "Safe Harbor" protection if they fail.
The Mitigation: Lawlet operates with mandatory provenance tracking. Every AI-generated case brief, summary, or Moot Court simulation includes embedded metadata and clear visual watermarks indicating it is AI-generated, ensuring strict compliance with the new SGI labeling mandates.
2. BCI Rules & Unauthorized Practice of Law (UPL)
The Risk: The Bar Council of India (BCI) and the Advocates Act, 1961, strictly prohibit non-lawyers (and algorithms) from providing direct legal advice. If Lawlet is perceived as replacing human legal judgment, it risks being shut down for UPL.
The Mitigation: We strictly position Lawlet as a B2B/Student "Intelligence Layer" and research assistant, not a public-facing legal advisor. Our UI enforces a "Human-in-the-Loop" architecture, requiring the user to verify citations before exporting. We sell efficiency to the lawyer, not advice to the client.
3. DPDP Act 2023 & Section 126 Privilege
The Risk: Uploading sensitive client data or pre-litigation strategies into an LLM risks violating the Digital Personal Data Protection (DPDP) Act of 2023 and breaking Lawyer-Client Privilege under Section 126 of the Indian Evidence Act.
The Mitigation: Absolute data sovereignty. We utilize localized Indian server infrastructure (AWS/Azure India) to satisfy data residency. We execute zero-data-retention agreements with foundational model providers, guaranteeing that no user query is ever used to train public models.
4. Emerging Liability in Agentic AI workflows
The Risk: As we scale toward autonomous legal workflows (e.g., an AI agent independently cross-referencing and filing basic compliance forms), current Indian tort and consumer protection laws place absolute liability on the developer if the "Agent" makes a critical error.
The Mitigation: We employ a "Read-Only" automation approach for Phase 1 and 2. Lawlet synthesizes and recommends, but requires a manual cryptographic signature or explicit "Accept" click from the human lawyer before any external action is executed. |
Vector Database Latency (The RAG Bottleneck): At 10x scale, naive KNN (K-Nearest Neighbors) searches within a vector database (e.g., Pinecone, Weaviate) holding millions of legal chunks will suffer severe latency degradation.
LLM Inference Rate Limits & Costs: Hitting external LLM APIs (GPT-4/Claude) at 10x volume will immediately trigger provider rate limits (Requests Per Minute/Tokens Per Minute), resulting in API throttling and exponential cost bloat.
Context Window Overload (The Needle-in-a-Haystack Failure): Attempting to stuff massive context windows with retrieved documents at scale degrades accuracy. LLMs suffer from "lost in the middle" syndrome, missing critical legal nuance when parsing too much text simultaneously.
Compute Bottlenecks for FIRAC Engine: Our proprietary FIRAC synthesis engine requires significant sequential processing. At 10x concurrent users, this sequential logic will create a queue bottleneck, resulting in unacceptably high Time-to-First-Token (TTFT) for users waiting for case summaries.
State Management in AI Moot Court: The Moot Court simulation requires maintaining complex, multi-turn conversational state and scoring logic. Scaling this requires massive read/write operations to a fast caching layer (like Redis). Without optimization, the simulation will lag, breaking the real-time "flight simulator" experience.
SGI Labelling Compliance Lag: As mandated by the 2026 IT Rules, every AI output must be watermarked and logged. At 10x scale, the overhead of real-time metadata injection and compliance logging will severely degrade inference speed if not processed asynchronously. |
Current in-house expertise is heavily indexed on Applied AI and System Orchestration:
RAG Architecture & Grounding: Deep expertise in building Retrieval-Augmented Generation pipelines, specifically structuring dense, unstructured data into vector embeddings for deterministic, low-hallucination retrieval.
Cascading Prompt Engineering: Advanced capability in enforcing structural guardrails on foundational models. This includes developing the proprietary logic that forces LLMs to synthesize case law strictly within the FIRAC (Facts, Issue, Rule, Analysis, Conclusion) framework.
Workflow Integration: Integrating AI endpoints with functional software infrastructure, managing state logic for complex conversational simulations (like the AI Moot Court), and handling API-based data ingestion.
Strategic Hiring Plan
Our immediate goal post-funding is to expand from an Applied AI structure to a Deep-Tech ML team. We are allocating capital to hire specialized NLP engineers and ML researchers to bring model fine-tuning (specifically SLMs trained on Indian jurisprudence) completely in-house, reducing our reliance on foundational API models and optimizing compute latency for scale. |
Publicly Available Government Datasets (Ingested & Processed): We utilize large volumes of public domain data, which we ingest, clean, and structure. This includes:
Indian Judgments: Unstructured case law from the Supreme Court of India and various High Courts.
Statutory Data: Indian Bare Acts, penal codes (BNS, BNSS, BSA), and government gazette notifications.
Live Court Data: Daily Cause Lists and localized court schedules.
Proprietary Owned Datasets (The Data Moat): We own all datasets generated organically within the platform.
The FIRAC Knowledge Graph: Our proprietary database of human-verified, FIRAC-structured legal logic graphs derived from raw judgments. This structured data is strictly owned by Lawlet.
Advocacy Interaction Data: The telemetry and interaction data generated from the AI Moot Court simulations. This dataset, measuring rebuttal speed, logical flow, and argument construction, is a unique, owned asset unavailable in the public domain.
User Research Telemetry: Anonymized data tracking research patterns, query intent, and cognitive gaps among law students and junior professionals.
Licensed Datasets (Strategic Partnerships):
We are currently negotiating API access to secondary legal data aggregators and regional legal reporting services to ensure our RAG architecture has the most up-to-date, hyper-local precedents where public data ingestion is delayed. |
The Proprietary Data Flywheel: We leverage interaction data generated from our AI Moot Court and simulation features. Every user interaction, argument structure, and rebuttal pattern acts as training data to continuously fine-tune our models on legal advocacy and cognitive behavior.
Human-in-the-Loop (HITL) via RLHF: We capture the "edit distance" when users modify AI-generated FIRAC summaries or research briefs. This provides a continuous stream of high-quality Reinforcement Learning from Human Feedback (RLHF), organically aligning the model with expert legal standards without the cost of manual data labeling.
Transition to Domain-Specific SLMs: We are systematically migrating compute-heavy extraction tasks from generalized LLMs (which are prone to hallucination) to custom-trained Small Language Models (SLMs). These SLMs are continuously fine-tuned exclusively on Indian jurisprudence to increase domain accuracy while driving down inference latency and API costs.
Dynamic RAG Pipeline Optimization: Our vector database is not static. We employ automated feedback loops that adjust chunking strategies and retrieval heuristics based on user query success rates. As new judgments and gazettes are published, the RAG architecture dynamically re-indexes to ensure the grounding layer is always current.
Automated Regression Testing (Golden Datasets): Before any model weight adjustment is pushed to production, it is programmatically stress-tested against a proprietary "Golden Dataset" of complex Indian legal scenarios. We use evaluation frameworks like RAGAS to ensure zero regression in context precision and absolute compliance with our zero-hallucination baseline. |
Phase 1: India as the Beachhead (The Ultimate Stress Test)
India is the world’s most complex and volume-heavy legal ecosystem, generating over 5 million pages of judicial data annually. By building our AI architecture to accurately parse the extreme complexity, backlog, and multi-linguistic nuances of the Indian legal system, we are forging an infinitely resilient intelligence layer. We are starting here to capture the largest single pool of high-intent law students globally (~500,000 active).
Phase 2: Commonwealth Expansion (Shared Legal DNA)
Because the Indian legal system is rooted in English Common Law, our core AI architecture (especially the FIRAC synthesis engine and Moot Court logic) is inherently exportable. Once we achieve institutional PMF in India, our immediate expansion targets are the UK, Australia, Singapore, and Canada. The underlying legal mechanics are identical; only the grounding datasets (local RAG vectors) need to be swapped.
Phase 3: The Global Thesis
While our initial Go-To-Market is hyper-localized, the long-term vision is an autonomous "Operating System" for the legal mind. The problem of cognitive overload and manual research fatigue is universal. By mastering the hardest market first (India), we ensure our AI architecture can seamlessly disrupt high-ARPU legal markets globally. |
|
|
|
|
|
|
| Apr 20, 2026 @ 6:47 PM |
Tushar Moitra , Priyanshu Sharma |
225100036@iitdh.ac.in |
https://www.linkedin.com/in/tushar-moitra/ |
http://suprsyncr.in |
9799007918 |
Tushar ( Founder, ML eng.) , Priyanshu (Founder, SDE) |
we both are batchmates studying at IIT DH. we are working on this from last 3 months |
All |
Our biggest strength as a team is our founder-level ownership mindset. We don’t just execute tasks and we treat every problem like it’s our own company on the line. That means we move fast, make decisions with conviction, and stay accountable for outcomes, not just effort. Each of us brings a different skill set, but we’re united by a shared drive to build, iterate, and improve constantly. This combination lets us turn ideas into reality quickly while staying resilient through challenges. |
Suprsyncr |
suprsyncr.in |
Dharwad, Karnataka |
AI that sell for you |
Businesses today sell across multiple channels like marketplaces, D2C websites, and social platforms, but managing pricing, inventory, customer queries, and order flows across all of them is highly fragmented and manual. This leads to lost sales, inconsistent customer experience, and poor decision-making. Existing tools are either too siloed or not intelligent enough to actually drive revenue. |
Suprsyncr is an AI-powered multi-channel sales co-pilot that helps businesses automatically manage and optimize their entire selling workflow. It connects across platforms and uses AI to handle pricing optimization, inventory syncing, customer engagement, and sales decisions in real time. Instead of just providing dashboards, it actively executes actions that improve conversions and revenue, acting like a 24/7 sales operator for the business.
A single platform that helps from creating your seller account to listing and selling product online. |
Most tools stop at analytics or basic automation. Suprsyncr goes one step further by combining real-time data integration with decision-making AI that not only suggests but executes revenue-driving actions. Over time, it builds a proprietary data layer across channels and learns business-specific patterns, making the system smarter and harder to replace. The defensibility comes from this continuously improving AI layer and deep integrations across fragmented commerce ecosystems. |
two |
Testimonials |
Primary: D2C brands, e-commerce sellers, and SMBs selling across multiple channels (marketplaces like Amazon/Flipkart, Shopify stores, social commerce). Secondary: Mid-sized brands scaling multi-channel operations and needing automation in pricing, inventory, and customer engagement. |
$2.4B |
$470M |
$35–45M |
SaaS subscription model (tiered pricing based on features, number of channels, and order volume). Additional revenue through usage-based pricing for AI actions (automated campaigns, optimizations) and enterprise integrations. |
Shopify apps ecosystem, Zoho Commerce, Unicommerce. |
Partnerships with e-commerce enablers, agencies, and Shopify ecosystem partners. Content and community-led growth targeting founders and operators. Performance-driven case studies to build trust and referrals. |
Start with a focused niche (high-growth D2C brands in India), solve one critical pain point deeply (sales + operations automation), and expand horizontally across workflows. Land with a strong ROI-driven use case, then expand via upselling additional automation features. Leverage integrations (Shopify, marketplaces) as distribution channels. |
To become the default AI operating system for commerce businesses globally, where selling is fully autonomous.
Suprsyncr should evolve from a tool into an AI layer that runs end-to-end sales, decisions, and execution across all channels, allowing businesses to scale without operational complexity. |
NA |
NA |
one |
We are looking for structured support to accelerate from MVP to a strong product with clear market validation. Avinya’s focus on early-stage founders, hands-on mentorship, and access to a strong network makes it the right environment for us to refine our product, sharpen our GTM, and avoid common early-stage mistakes. More than capital, we value guidance on building a scalable, defensible company. |
Achieve strong product validation with paying customers. Refine and validate our go-to-market strategy. Build a stable, scalable product with clear core use cases. Establish early revenue traction and repeatable growth channels. |
Yes (both) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 20, 2026 @ 5:06 PM |
Dipendu Biswas |
biswas.dipendu@gmail.com |
https://www.linkedin.com/in/dipendubiswas |
https://www.tarisa.tk/resolution/lms.html |
+91 9980796314 |
Dipendu is a Technology Architect and does product development. He has 25 years of experience in digital transformations for Capital One, Nationwide Insurance, Credit Suisse and ANZ Bank. He has professionally practiced technology architecture for Accenture, EY, Nokia and Wipro. Karatoa looks after talent acquisition and public relations. She has 5 years of experience in Information Technology training and education. |
We have met at networking events. We have worked together for 7 years. |
2 |
We have pitched to angel investors and venture capitalists and convinced them on the investment opportunity. We have received a soft commitment from an investor. We have received recognition for our venture development by Government of Karnataka, Government of India, Bank of Baroda and Reserve Bank of India. Press releases have covered us and our startup, which are social proofs. See press release here
https://pib.gov.in/PressReleasePage.aspx?PRID=1984364 |
Tarisa Technologies |
https://www.tarisa.tk/resolution/lms.html |
Bengaluru |
Software as a Service to determine the creditworthiness of small and medium businesses using alternative data sources. |
Micro, small and medium enterprises form a large portion in their contribution to the economy. They may not have credit history with the bureaus. They are often financially excluded as financial institutions find it prohibitive in terms of cost for lending in small ticket sizes. Financial analysts have to process volumes of disparate information about MSME and hence it takes lot of time. |
We have developed a platform for underwriters to assess financial health of small and medium businesses interactively. Biases of gender, economic status, industry and community are checked and removed. Our service determines the probability of default in repayment of loan by a borrower accurately. Small ticket sized loans are assessed in real-time. Large ticket sized loans are assessed in due course of time. Our platform learns about simple cash flows of businesses quickly. It requires adequate time to learn about complex cash flows of businesses. Banking and financial services require compliance with regulations. Hence, neural networks in our service have sufficient depth so as to provide explainability. Human in the loop label contextual data for training models. |
While business analytics is the base for processing huge volumes of data. It has to be extended with inferencing capabilities of AI for high productivity in loan processing. |
MVP |
Pilots, Testimonials |
Financial institutions |
489 billion USD |
2 billion USD |
150 million USD |
B2B subscription and Annual Maintenance Contract |
Lendingkart, Kaleidofin, Bluevine, Ondeck |
Founder sales |
Few financial institutions as design partners. Exhibits for financial industry trade shows. |
Our vision is to evolve an ecosystem of trust in business to business interoperability. |
Private Limited Company |
NA |
Yes |
Continuous venture acceleration |
Business connects and investment |
Yes |
Yes |
Core engine |
Databases like MongoDB, MySQL, Apache Hive are used for staging data. Initial training dataset is 70% of financial history available at any point in time. Machine learning models are progressively adapted as and when risk assessment is performed for MSME. Data is annotated by data scientists for samples and then annotations are automated. Production models are available for check bounce, account balance shortfall (monthly, quarterly), repayment default, tax invoices, annual returns and tax returns. Time for real-time training is available upto 1 minute. Time for batch training is available upto 1 month. A sample set from production is used for simulations and performance analysis. Predictions of loan default will be measured against repayment made within due date or loan underwritten. Probability of default is directly proportional to the rate of interest charged to borrower. Probability crossing defined thresholds will lead to rejection in loan sanctions. 5% inaccuracy will be tolerated. Imbalanced classification is analysed in the training dataset and Bayesian Optimisation is applied. We make real-time prediction when borrower is looking for loan from the financial institution. We make batch predictions when we need to project Non Performing Assets for the financial institutions. Time for batch prediction is available upto 1 month. Data volume is in terabytes. |
We scrape public data from the internet and filter them for consumption based on allowance. Generative Adversarial Network is used to generate synthetic data. |
Banking and financial services require compliance with regulations. Hence, neural networks in our service have sufficient depth so as to provide explainability. Human in the loop label contextual data for training models. |
Prediction of loan default is measured with respect to repayments made within the due date or loan underwritten. Machine learning is tuned periodically to reduce inaccuracies in predictions. Sample size is evaluated by data scientists for relevance of AI functions of our service in society. |
We comply with DPDPA and GDPR. We secure data at rest and in motion. |
We are a member of Self Regulatory Organisation in the Fintech sector which follows and adapts to RBI regulations. |
We require 8 GPU for 91 hours per month for pre-training models. Fine tuning of models require 1 GPU for 8 hours.
The computation is required for 25 hours per week. The capacity costs USD 98.32 per hour. So, the estimated cost is USD 9832.00 per month. Hence our computation for machine learning is scaled down. |
Our team has experience in machine learning pipeline. We have skills in Python, Pytorch, R packages, Transformers, Model training, finetuning, Generative Adversarial Networks, Retrieval Augmented Networks, Vector databases, Jupyter notebook |
Financial history is sourced from various data sources like core banking systems, credit bureaus, account aggregators. Initial training dataset is 70% of financial history available at any point in time. Machine learning models are progressively adapted as and when risk assessment is performed for MSME. Data is annotated by data scientists for samples and then annotations are automated. |
Prediction of loan default is measured with respect to repayments made within the due date or loan underwritten. Machine learning is tuned periodically to reduce inaccuracies in predictions. Sample size is evaluated by data scientists for relevance of AI functions of our service in society. |
We are building in India for local and global markets. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/4fR3sblIxLnT-Underwriter-copilot.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/4fR3sblIxLnT-Underwriter-copilot.pdf |
https://www.tarisa.tk/resolution/virtually.html |
In our survey, we have determined 62 million borrowers of finance in India. 300 million lives are affected by finance. Financial inclusion continues to be an unsolved problem. It affects contribution to GDP and Sustainable Development Goal for decent work and economic growth. As each underserved borrower completes a loan cycle with appraisal, access to finance and repayment of instalment, their credit usage is fed to the bureaus. With increasing credit history such businesses can seek finance at better terms. Success of our service is measured by completion of loan cycles and subsequent offer of further finance to MSME. Unsolicited loan requests by businesses returning to our service is also sought favorably. |
Woman co-founder |
ACSEL, GoK, KDEM |
checked |
| Apr 20, 2026 @ 4:59 PM |
Bryan Rodriges |
bryan@ai5labs.com |
https://www.linkedin.com/in/bryanrodrigues123 |
https://sentinel.ai5labs.com |
7975563057 |
Bryan is an Engineer working on ai systems from past 2 years. Chrysilla an software engineer at neflix on thier agentic harness engineering team |
We are siblings and have been working on side projects for the past 10 years |
1 |
Chrysilla brings in indepth expertise on the harness engineering from that is necessary for building and deploying robust models in regulated industry at massive scale. Bryan understands how to build systems on the software front and understands how to explore multiple use cases for deploying these agentic systems we are building for different use cases within industry |
ai5labs |
www.ai5labs.com |
Hubli Dharwad |
We unlock the enterprise bottleneck when in comes to providing agents with context to become usable within industry. |
Currently there is not substrate which enables lllms to have context and become usable within enterpise settings. This could be the case for smaller businesses as well. Sentinel unlocks that. |
We have building deterministic pipelines with extraction tools to map any and every incoming enterprise data source, this allows companies to actually be able to connect the data to llms and start using them for actual business use cases. building agents and increasing overall productivity |
our extraction pipeline is pretty unique and deterministic in nature there by allowing users building cross verification pipelines on top of it. "taking a loan form the bank? build a pieline ensuring all documents exist if they dont email them on missing documents also verify all the documents for name, age, adress, adhaar card" this simple task ideally needs so much effort |
MVP |
Pilots |
every single business could be our potential target. ideally enterprise |
10 billion usd |
2 billion usd |
200 million usd |
we charge a base subscription fee and top up credits |
legacy rpa tools, basware, procure etc |
none paying - we need to stress test it a lot though |
enterpise sales and linkedin |
Building agentic workflows for different domains like procurement, financing, hr, operations, maintenance, triage, etc on top of this platform data |
NA |
NA |
Yes |
for mentorhsipt connections and access to funding |
industry exposure |
yes |
Yes |
Core engine |
Sentinel is a multi-provider, RAG-first document intelligence platform. Stack:
- LLMs: Provider-agnostic factory (config-driven, zero-code swaps). Default: Fireworks MiniMax M2.5. Also wired: Anthropic Claude Sonnet 4, OpenAI GPT-4o, Groq Llama 3.3 70B, DeepSeek, local Ollama.
- Embeddings: Azure OpenAI text-embedding-3-large (1024-dim).
- Reranker: Voyage AI cross-encoder (optional, graceful fallback).
- Frameworks: LangChain 1.2 + LangGraph 1.0 (5-node state machine for chat: load_context → plan → execute_tools → generate → save). FastAPI backend, React/TS frontend.
- Document parsing: (layout-aware PDF/DOCX → markdown with bounding boxes).
- Vector store: Qdrant (hybrid dense + sparse BM25).
- Graph store: Neo4j (entity/relationship KG extracted per pipeline).
- Orchestration: Celery + Redis (extraction + indexing queues, idempotent with stable point IDs).
- Streaming: SSE over Redis pub/sub. |
Sentinel does not train on customer data. Our advantage is structural, not dataset-based: every processed document produces (a) user-defined structured extractions with confidence scores, (b) a
per-workspace entity graph, and (c) cross-reference verification traces. Over time, each customer's workspace becomes a private, queryable knowledge graph of their regulated document corpus — a
compounding switching cost. |
- Hybrid retrieval pipeline: dense + sparse BM25 fused via Reciprocal Rank Fusion (k=60) + Voyage reranking, with per-pipeline tunable weights — not off-the-shelf RAG.
- Deterministic verification layer: extraction confidence is composed from grounding (field-text-in-doc), format, and cross-field/cross-document rules — not just an LLM self-rating.
- Cross-document reference engine: multi-doc dependency rules (e.g. a PO must match a referenced contract) — hard to replicate for compliance workflows.
- Provider portability: we swap LLMs via config, so we are not locked to any single vendor's pricing or availability.
- Workspace-scoped knowledge graph: entities + relationships in Neo4j, scoped per tenant, expand retrieval context beyond pure vector search. |
- In-house SingleAxis SASF framework for model performance and reliability evaluation.
- Per-extraction confidence score (0–1) composed from grounding, format, and cross-field consistency; ≥0.85 auto-approves, below routes to human review (Review Inbox).
- End-to-end OpenTelemetry tracing: OpenLLMetry auto-instruments every LangChain/Azure OpenAI/httpx call; manual spans in extraction, indexing, chat tools. All spans tagged with workspace_id /
document_id / pipeline_id. Exported to Phoenix (Arize) with 100ms flush.
- Token + cost metering: every LLM call records prompt/completion tokens and USD cost asynchronously.
- Human-in-the-loop feedback: reviewer approvals/rejections in the Review Inbox become ground truth for prompt + schema iteration. |
- AuthN: Supabase Auth (JWT) via httpOnly Secure SameSite cookies; CSRF via Origin/Referer validation.
- AuthZ: RBAC (owner/admin/editor/reviewer/viewer), enforced at route level with require_permission().
- Tenant isolation: workspace_id foreign key on every entity; Qdrant filtered by project_id; Neo4j entities scoped by project_id.
- Encryption: in-transit + at-rest via Supabase managed Postgres and S3-compatible storage.
- Prompt-injection hardening: document content wrapped in tags with an anti-injection preamble; LLM instructed to treat document text as data, not instructions.
- Security headers: CSP, HSTS, X-Frame-Options middleware.
- No model training on customer data. |
- DPDP (India) / GDPR (EU): Sentinel processes regulated documents (compliance certs, tax filings, contracts); any cross-border model routing must respect data-residency. Today we default to a
US-hosted Fireworks endpoint — enterprise deployments can pin to EU/India-region providers via config.
- Sector-specific: customers in finance / insurance / healthcare may require SOC2 Type II, ISO 27001, HIPAA BAAs — on roadmap.
- AI-specific regulation: EU AI Act high-risk classification for decision-support systems — mitigated by our human-in-the-loop Review Inbox and full audit trail (OTel spans + Postgres history).
- Model-provider dependency: OpenAI/Anthropic policy shifts — mitigated by provider-agnostic factory. |
- Qdrant single-cluster throughput — solved by sharding per workspace or upgrading to Qdrant Cloud clusters.
- Celery queue saturation — extraction soft-limit is 10min; at 10x, LLM rate limits become the bottleneck before compute. Mitigation: per-provider rate-limit pools and batched embedding calls.
- Neo4j write contention on shared entity nodes — mitigation: per-workspace database or partitioned graph.
- Redis pub/sub fanout for SSE chat streaming — mitigation: move to dedicated pub/sub (NATS/Kafka) or per-region Redis.
- Embedding cost — 10x volume on text-embedding-3-large is linear-cost; we would evaluate cheaper 512-dim models or on-prem embeddings.
- Docling CPU cost for PDF parsing — horizontal worker scaling + caching of re-ingested docs. |
Chrysilla Rodrigues - AI Agent Harness engineer at netflix
Bryan Rodrigues - Building muliple data integrations using RAGs |
We do not train models. Customer data is processed in-workspace and never used for training. System-provided pipeline templates (Invoice, PO, Contract, Compliance Cert, Supplier Onboarding) ship with
synthetic demo datasets we authored. Evaluation corpora are either customer-owned (under contract) or synthetic. |
- Reviewer feedback loop: human approvals/rejections in the Review Inbox are labeled ground truth; used to tune extraction prompts and confidence thresholds per pipeline.
- Per-pipeline A/B: LLM provider, chunking config, and retrieval weights are JSON-configurable — we can A/B test per workspace without deploys.
- OTel-driven regression detection: every LLM call is traced; latency/cost/error regressions surface in Phoenix dashboards.
- Model upgrades via config: new providers (e.g. Claude 4.7, GPT-5) swap via config.yaml.
- Cross-reference rule engine: deterministic rules evolve with customer workflows and directly raise precision without retraining. |
Both. Architecture is region-agnostic (ASCII-safe storage paths, no hardcoded locale); LLM and embedding providers are swappable per deployment to satisfy India DPDP residency or EU GDPR. Initial GTM
focuses on regulated-document workflows (compliance, procurement, finance) where both Indian and global buyers have the same pain. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qRhytyn7Rd8n-Sentinel_Pitch_Deck.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qRhytyn7Rd8n-Sentinel_Pitch_Deck.pptx |
http://sentinel.ai5labs.com |
No |
I have a catholic minority certificate. not sure it that will help |
Shriti Muesli |
checked |
| Apr 20, 2026 @ 4:28 PM |
Christeena Jaya Prabhu |
jchristeena@el-ariah.com |
https://www.linkedin.com/in/christeena-prabhu-34aa2423/ |
https://elcrunchae.com |
8151977997 |
Founder and Director of El-Ariah Preserves Private Limited |
We first connected through a shared interest in building high-quality, consumer-focused brands, particularly in the food and nutrition space. Our early interactions quickly evolved into deeper collaboration as we discovered strong alignment in our long-term vision, execution style, and approach to building scalable businesses.
Over the past [insert duration], we have worked closely across both strategy and execution—ranging from validating product ideas and refining our go-to-market approach to building early traction and shaping the brand narrative. This period has allowed us to develop a strong working dynamic grounded in trust, clear communication, and complementary skill sets.
One of the founders, Christeena JayaPrabhu, is the Founder & CEO of El Crunchae, a brand focused on revolutionizing freeze-dried food. With over 19 years of experience in business operations, she brings deep expertise in building and scaling operations-driven businesses, along with a strong vision for transforming how consumers access nutritious, convenient food products globally |
1 |
Our biggest strength as a team is the combination of deep operational expertise and strong brand-building capability, coupled with a shared bias toward execution.
With over 19 years of experience in business operations, Christeena JayaPrabhu brings a proven ability to build and scale systems, manage supply chains, and translate vision into structured, repeatable processes. Her experience as Founder & CEO of El Crunchae—focused on innovating in the freeze-dried food space—adds a strong foundation in product development, quality control, and understanding evolving consumer preferences.
Complementing this is a strong focus on market insight, positioning, and rapid experimentation. As a team, we balance strategic thinking with hands-on execution, allowing us to move quickly from idea to validation while maintaining clarity on long-term goals.
What truly sets us apart is our alignment in vision and our ability to challenge each other constructively. We make decisions quickly, communicate openly, and take ownership across functions, which helps us maintain momentum even in uncertain or ambiguous situations.
This combination of operational depth, execution speed, and mutual trust enables us to build efficiently and adapt as we scale. |
El-Ariah Preserves Private Limited |
https://el-ariah.com |
Bangalore |
Building a next-generation freeze-dried food brand that delivers nutritious, convenient, and high-quality products for modern consumers. |
Reimagining everyday nutrition with high-quality, convenient freeze-dried foods. |
Our solution is a next-generation freeze-dried food platform that makes nutritious, real-food products accessible, convenient, and long-lasting without compromising on quality or taste.
We use advanced freeze-drying technology to preserve the natural nutrients, flavor, and texture of ingredients while removing moisture, enabling long shelf life without artificial preservatives. This allows us to deliver clean-label food products that are lightweight, easy to store, and quick to consume—ideal for modern, fast-paced lifestyles.
Our product range is designed to solve everyday gaps in nutrition and convenience, offering healthy alternatives for on-the-go consumption, travel, work, and home use. By combining operational expertise with a strong focus on product quality and consumer needs, we are building a scalable food brand that bridges the gap between health, convenience, and accessibility. |
Our solution is unique because we combine clean-label, high-quality freeze-dried foods with strong operational expertise to build a scalable, early-mover brand in an underdeveloped category in India. |
MVP |
Users, Pilots |
Our target customers are urban, health-conscious consumers who value convenient, clean-label, and nutritious food options, including working professionals, frequent travelers, fitness-focused individuals, and young households. |
4752 |
2850 |
901 |
B2C, B2B |
Our revenue model is B2C sales via D2C and marketplaces, with additional B2B distribution partnerships. |
We acquire customers through digital marketing (social media and performance ads), marketplace visibility, influencer collaborations, and partnerships with niche retail and wellness communities |
Our go-to-market strategy is to start with D2C and online marketplaces to validate demand and build brand awareness, supported by digital marketing and influencer-led education, followed by expansion into niche retail, wellness stores, and institutional B2B channels. |
Our long-term vision is to become a global leader in freeze-dried nutrition by making high-quality, nutritious, and convenient food accessible to everyday consumers, redefining how the world eats on the go without compromising health or taste. |
Private Limited |
NA |
No |
We are applying to Avinya to gain strategic mentorship, access to a strong founder ecosystem, and structured support to accelerate product-market fit and scale our go-to-market execution in the early stages of building our freeze-dried food brand. |
During the programme, we aim to validate product-market fit, refine our go-to-market strategy, build early traction, and establish a scalable operational and supply chain foundation for growth. |
Hybrid |
Yes |
Supporting feature |
We do not have a core AI architecture as our business is not AI-native; we primarily use standard digital tools and analytics platforms for marketing, sales tracking, and operational decision-making. |
We are building proprietary insights from customer feedback, purchase behavior, and product testing data within the freeze-dried food category, which is still nascent in India. |
We do not rely on an AI component for our core product; our defensibility comes from product quality, operational expertise, and early-mover advantage in the category. |
Not applicable, as we do not use AI models in our core product. |
We follow standard data protection practices for customer data collected through D2C channels and marketplaces, ensuring secure storage and compliance with applicable platform and legal requirements. |
Our risks are primarily in food safety, labeling compliance, and FSSAI regulations, which we actively adhere to as a food manufacturing brand. |
Not applicable, We use internally collected customer feedback, sales data, and product testing insights, which are fully owned by us.ot have an AI system in our product stack. |
No, we do not have an in-house AI/ML team, as our focus is on food product development, operations, and brand building. |
We use internally collected customer feedback, sales data, and product testing insights, which are fully owned by us. |
Not applicable to our current business model. |
We are initially building for India with a long-term vision to expand into global markets. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/SZkqUFTJYOrR-El-Ariah-Preserves-Pitch-Deck-Final.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/SZkqUFTJYOrR-El-Ariah-Preserves-Pitch-Deck-Final.pdf |
http://NA |
Yes, we are building an impact-focused startup aimed at improving everyday nutrition by making clean-label, minimally processed, and highly convenient food more accessible to modern consumers.
Our goal is to address the growing gap between health and convenience in today’s food system, where consumers often have to compromise nutrition for ease of use. Through freeze-dried food technology, we aim to preserve the natural integrity of ingredients while eliminating the need for preservatives, additives, and excessive processing.
In the long term, we also aim to contribute to reducing food waste and improving food preservation efficiency, while expanding access to nutritious food across different geographies and lifestyles. |
NA |
Suhruth Ujjni |
checked |
| Apr 20, 2026 @ 12:20 AM |
HARSH PARASHAR |
harshparashar@amoghconnect.com |
https://www.linkedin.com/in/harsh-parashar-0b8baa224/ |
http://amoghconnect.com |
7415732128 |
FOUNDER & CEO |
We have met our team via linkedin and offline startups events at multiple locations across India. |
All |
Biggest strength as a team to work together on deep tech startup problem challenges and finding their solutions. Thinking at root level, understanding the market and exactly building the solution that revolutionize the world. |
Amogh Ever Useful (Fornax EverUseful Private Limited) |
amoghconnect.com |
New Delhi |
We are building a privacy-first, on-premise federated AI platform that enables hospitals to leverage collective intelligence without ever sharing patient data. |
We are solving the fundamental conflict between AI advancement and patient data privacy in healthcare. Healthcare institutions generate vast amounts of valuable data—from electronic health records (EHRs) to imaging and diagnostics—but this data is highly sensitive, regulated, and fragmented. To build effective AI systems, large-scale data aggregation is typically required. However, existing solutions rely on centralized cloud-based models, which force hospitals to share or move patient data outside their secure environments. This creates serious challenges, including privacy violations, regulatory non-compliance (DPDP, HIPAA, GDPR), risk of data breaches, and loss of data ownership . As a result, hospitals face a dilemma: Either adopt AI and risk legal/privacy issues, Or protect patient data and miss out on AI-driven improvements in care. This has led to slow and limited adoption of AI in real-world clinical settings, despite its proven potential to improve diagnosis, treatment, and operational efficiency. Additionally, healthcare data is often siloed across institutions, meaning no single hospital has enough diverse data to build highly accurate and unbiased AI models. This results in weaker models, biased outcomes, and limited generalizability. At the same time, most AI systems lack clinical alignment and trust, as they operate without continuous feedback from doctors, leading to issues like incorrect recommendations or hallucinations. Therefore, the core problem we are solving is: How can healthcare institutions leverage the full power of AI—across large, diverse datasets—without ever compromising patient privacy, data ownership, regulatory compliance, or clinical trust? Our solution directly addresses this by eliminating the need for data centralization while enabling collaborative intelligence, making AI adoption safe, scalable, and trustworthy for healthcare systems. |
We are building an AI-first, privacy-first healthcare intelligence platform that addresses one of the biggest challenges in modern healthcare: enabling advanced AI adoption without compromising patient data privacy. Today, healthcare data is highly sensitive and fragmented, and traditional cloud-based AI solutions require centralizing this data, which introduces risks such as data breaches, regulatory violations, high latency, and loss of data ownership. As a result, hospitals are often unable to fully leverage AI despite its transformative potential .
Our core innovation lies in a paradigm shift: instead of sending data to AI, we bring AI to the data. We deploy our platform entirely on-premise within hospital infrastructure, ensuring that patient data never leaves the hospital’s secure environment. This is combined with a federated learning architecture, where multiple hospitals collaboratively train AI models without sharing raw data. Only encrypted model updates are exchanged, enabling collective intelligence while maintaining strict privacy and data sovereignty. This approach not only ensures compliance but also leads to more robust and unbiased AI systems trained on diverse real-world data .
A key differentiator of our platform is the integration of Reinforcement Learning from Human Feedback (RLHF) into the clinical workflow. Doctors and clinicians actively participate in validating AI outputs, ranking their relevance, and correcting errors. This human-in-the-loop system ensures that the AI continuously improves based on real clinical expertise, significantly reducing hallucinations and increasing trust and reliability—critical factors for adoption in healthcare .
We are not building just a model but a complete healthcare AI ecosystem. Our platform includes secure EHR ingestion, a local AI inference engine for real-time insights, a federated learning orchestrator for distributed training, an RLHF feedback loop, and a compliance and audit layer. Importantly, our system is designed with a compliance-first approach, adhering to regulations such as DPDP (India), HIPAA (US), and GDPR (EU), making it enterprise-ready from day one .
What makes us truly unique is the combination of on-premise deployment, federated intelligence, human-aligned learning, and built-in compliance in a single architecture. Unlike conventional cloud-based solutions, our platform enables hospitals to retain full control over their data while still benefiting from shared AI advancements. This creates a scalable and secure network where institutions collaboratively improve AI without exposing sensitive information.
Our long-term vision is to build a sovereign healthcare AI ecosystem, enabling national-scale collaboration, accelerating medical research, and providing AI copilots for clinicians—all while maintaining the highest standards of privacy, ethics, and trust. |
Our solution is unique and defensible because it combines on-premise AI deployment, federated learning, human-in-the-loop training (RLHF), and compliance-first architecture into a single integrated platform—something most healthcare AI solutions do not offer together. Instead of relying on cloud-based models that require sensitive patient data to be centralized, we ensure that data never leaves hospital infrastructure, preserving full data ownership and eliminating major privacy and regulatory risks . A key differentiator is our use of federated learning, which allows multiple hospitals to collaboratively improve AI models without sharing raw data, creating a powerful network effect while maintaining strict confidentiality. On top of this, we integrate continuous clinical feedback from doctors, ensuring the AI is not just accurate but also aligned with real-world medical decision-making—significantly increasing trust and usability in clinical settings . From a defensibility standpoint, our advantage lies in the deep integration with hospital infrastructure, regulatory compliance readiness (DPDP, HIPAA, GDPR), and the complexity of building a distributed, privacy-preserving AI system at scale. This creates strong technical, operational, and regulatory barriers for new entrants. Over time, as more hospitals join the network, our federated system becomes smarter and more valuable, creating a data network effect without actually owning the data, which is extremely difficult for competitors to replicate. |
Revenue |
Pilots, Signups, Testimonials |
Our target customers include hospitals and multi-specialty healthcare providers as primary users, who need secure, on-premise AI for clinical decision support and EHR intelligence. We also serve diagnostic centers and radiology labs that handle large volumes of sensitive imaging data and require accurate, privacy-preserving AI analysis. Additionally, our platform is designed for government health systems and public healthcare bodies to enable large-scale, compliant, and sovereign AI deployment across regions. We cater to research institutions and medical colleges for collaborative AI-driven research without sharing raw patient data, and in the long term, healthtech companies and hospital networks looking to build scalable, compliant AI solutions. Overall, we target organizations that require advanced AI capabilities while maintaining strict data privacy, ownership, and regulatory compliance. |
$150B+ |
$20B–30B |
$50M–100M |
Our revenue model is a B2B hybrid of subscription and enterprise licensing, where hospitals pay recurring fees for using our platform, along with one-time setup and integration charges, and additional usage-based pricing for AI services. |
Google Health, Microsoft Healthcare AI, IBM Watson Health; Epic Systems, Oracle Cerner; Aidoc, Zebra Medical Vision, PathAI. |
We acquire customers through hospital partnerships, pilot deployments, and institutional collaborations, supported by direct B2B sales, government and research tie-ups, and referrals within healthcare networks. AIIMS Delhi is our first customer. |
Our go-to-market strategy is to start with pilot deployments in leading hospitals, prove clinical and operational impact, and then scale through hospital networks, government partnerships, and research collaborations, supported by direct enterprise sales and referral-driven expansion. |
Our long-term vision is to build a sovereign, privacy-first healthcare AI ecosystem that transforms how intelligence is created and used in medicine. We aim to connect hospitals, diagnostic centers, and research institutions through a federated network, where each participant benefits from collective AI advancements without ever compromising patient data privacy or ownership.
Over time, we envision our platform becoming the default AI infrastructure layer for healthcare, powering everything from clinical decision support and intelligent EHR systems to real-time diagnostics and personalized treatment recommendations. A key part of this vision is developing AI copilots for doctors, nurses, and administrators, enabling faster, more accurate, and data-driven decision-making while reducing workload and burnout.
At scale, our federated approach will enable national and global healthcare intelligence networks, accelerating medical research, improving disease prediction, and supporting public health systems in real-time. By ensuring strict compliance, transparency, and human-in-the-loop learning, we aim to build AI systems that are not only powerful but also ethical, explainable, and trusted by clinicians and patients alike.
Ultimately, our goal is to redefine healthcare AI by making it secure, scalable, and universally accessible, creating a future where advanced medical intelligence can improve outcomes for millions without ever sacrificing privacy or trust. |
08.09.2025 |
0 |
one |
We are applying to the Avinya Program to gain mentorship, clinical and technical validation, funding and strategic support to scale our privacy-first healthcare AI platform, along with access to a strong ecosystem of healthcare experts, institutions, and investors to accelerate real-world deployment and impact. |
During the program, we aim to validate and scale our product through real-world hospital deployments, refine our technology with expert mentorship, achieve clinical and regulatory validation, and build strong partnerships with healthcare institutions, researchers, and investors to accelerate adoption. |
Yes |
one |
one |
Our AI architecture is designed as a privacy-first, distributed, and modular system that operates entirely within hospital environments while enabling collaborative intelligence across institutions.
At the core, we deploy an on-premise AI inference layer inside hospital infrastructure, where Electronic Health Records (EHR), imaging data, and diagnostic inputs are securely ingested and processed. This layer hosts large language models and specialized medical AI models that provide real-time clinical decision support, summaries, and insights without any data leaving the hospital.
On top of this, we implement a federated learning architecture, where multiple hospitals train shared models collaboratively. Instead of transferring raw patient data, each hospital trains the model locally and only encrypted model updates are shared with a central orchestrator, which aggregates and redistributes improved models. This ensures data privacy, ownership, and compliance while still benefiting from large-scale learning.
A key component is our RLHF (Reinforcement Learning from Human Feedback) loop, where doctors and clinicians continuously interact with the system by validating outputs, ranking suggestions, and correcting errors. This feedback is used to fine-tune models, ensuring they remain clinically aligned, accurate, and trustworthy over time.
Additionally, we include a compliance and audit layer that tracks all model decisions, ensures explainability, and maintains logs required for regulatory standards like DPDP, HIPAA, and GDPR. The system follows a zero-trust security architecture, with encryption, access controls, and strict data isolation.
Overall, our architecture combines on-premise intelligence, federated learning, human feedback loops, and compliance-first design to create a scalable, secure, and clinically reliable healthcare AI platform. |
Our primary data advantage lies in a privacy-preserving federated learning network, where we learn from diverse, real-world hospital data without ever owning or centralizing it. Through on-premise deployments and partnerships (including clinical environments like AIIMS), our models are continuously improved using live, high-quality clinical data and workflows, which is difficult for competitors to access.
Additionally, our RLHF (human feedback loop from doctors) creates a unique proprietary layer of clinically validated insights and model tuning signals, making our system more accurate and aligned with real-world medical decision-making over time.
Together, this creates a strong moat: a network-driven intelligence advantage built on private data access, clinical feedback, and continuous learning, which is hard to replicate without similar hospital integrations and trust. |
Our AI components are defensible because they combine privacy-first architecture, real-world clinical integration, and continuous learning loops that are difficult to replicate.
First, our on-premise + federated learning setup ensures models are trained on diverse, real hospital data without centralizing it, creating a network intelligence advantage that competitors relying on cloud data cannot easily access. Second, our RLHF (doctor-in-the-loop feedback system) generates proprietary, clinically validated training signals, making our models more accurate and aligned over time. Third, deep integration with hospital infrastructure and compliance frameworks (DPDP, HIPAA, GDPR) creates high switching costs and regulatory barriers.
Together, this results in a data + network + workflow moat, where performance improves with every deployment, making it extremely hard for competitors to match without similar scale, trust, and integrations. |
We evaluate model performance and reliability using a combination of technical, clinical, and safety metrics:
Model Performance Metrics: Accuracy, Precision, Recall, F1-score, and AUC-ROC for classification tasks; BLEU/ROUGE and semantic similarity for clinical text outputs.
Clinical Validation Metrics: Doctor agreement rate, relevance ranking (via RLHF), and reduction in diagnostic errors.
Reliability & Safety: Hallucination rate, confidence calibration, false positive/negative rates, and override frequency by clinicians.
Operational Metrics: Latency, response time, and system uptime for real-time hospital use.
Continuous Feedback Loop: Performance improvement tracked through RLHF signals and real-world usage outcomes.
This multi-layer evaluation ensures our models are not just accurate, but also clinically trustworthy, safe, and deployable in real-world healthcare settings. |
We handle data privacy, compliance, and security through a privacy-first, on-premise architecture combined with strong regulatory and technical safeguards.
All patient data is stored and processed within hospital infrastructure, ensuring it never leaves the secure environment. We use federated learning, where only encrypted model updates are shared—never raw data—preserving full data ownership and confidentiality.
From a compliance standpoint, our platform is designed to align with DPDP (India), HIPAA (US), and GDPR (EU), with built-in audit logs, access controls, and explainability to meet regulatory requirements.
On the security side, we implement a zero-trust architecture, including end-to-end encryption, role-based access control (RBAC), secure authentication, and continuous monitoring. This ensures that every access request is verified and all system interactions are tracked and protected.
Overall, our approach ensures maximum data privacy, regulatory compliance, and enterprise-grade security by design, not as an afterthought. |
Yes, there are regulatory risks, but our architecture is designed to minimize them.
Healthcare AI operates under strict regulations like DPDP (India), HIPAA (US), and GDPR (EU), which can impact how patient data is processed, stored, and used. Potential risks include changes in data protection laws, stricter AI regulations, requirements for clinical validation, and approval processes for AI-based medical decision systems.
However, our on-premise, privacy-first and federated learning approach significantly reduces these risks since no raw patient data leaves hospital infrastructure, ensuring strong data sovereignty and compliance alignment from the start. Additionally, our built-in auditability, explainability, and human-in-the-loop (doctor validation) further strengthen regulatory acceptance.
So while regulatory changes are a factor in healthcare, our system is proactively designed to adapt and remain compliant, making it more resilient compared to traditional cloud-based AI solutions. |
At 10x scale, challenges may arise in federated learning coordination (latency and aggregation complexity), model consistency and drift across diverse hospital data, scaling the RLHF feedback loop, variability in on-premise infrastructure, and increased security/compliance management complexity—these are scalability challenges, not fundamental limitations. |
Yes, we have an in-house AI/ML team skilled in LLMs, medical AI models, federated learning, RLHF, and real-world deployment, with expertise in handling healthcare data, optimizing models, and building scalable, on-premise AI systems. Fine-Tuning LLM's, handling tons of data, AIML deep tech scientific research and building new solutions for challenging AIML problems. |
We currently use a mix of public medical datasets and partner-provided clinical data for pilots. Public datasets are properly licensed/open-source, while any hospital data used is not owned by us but accessed under strict agreements, remaining fully within hospital infrastructure. Our system is designed so that data is never centralized or owned by us, ensuring compliance and privacy. |
We continuously improve model performance through a combination of federated learning, RLHF, and real-world feedback loops.
Models are regularly updated using federated learning across hospitals, allowing them to learn from diverse data without accessing raw patient data. At the same time, doctor feedback (RLHF)—including validation, ranking, and corrections—is used to fine-tune models and improve clinical accuracy. We also monitor performance metrics, drift, and real-world outcomes to retrain and optimize models over time.
This creates a continuous learning cycle, where the system improves with every deployment while remaining privacy-compliant. |
India and Global Market Both |
|
|
Yes, we are building a mission-driven, impact-focused startup centered on transforming healthcare through ethical, privacy-first AI.
Our mission is to make advanced healthcare intelligence accessible, secure, and trustworthy, without compromising patient privacy or data ownership. We are addressing a critical gap where hospitals are unable to adopt AI due to regulatory and privacy concerns, and enabling them to safely leverage AI for better diagnosis, faster decision-making, and improved patient outcomes.
Beyond technology, our impact lies in building a sovereign healthcare AI ecosystem, where institutions can collaborate, innovate, and advance medical research without exposing sensitive data. This has the potential to improve healthcare delivery at scale, support public health systems, and democratize access to high-quality medical intelligence, especially in resource-constrained environments.
Ultimately, our goal is to create a system where AI enhances human expertise in healthcare while maintaining the highest standards of privacy, ethics, and trust. |
NA |
NA |
checked |
| Apr 19, 2026 @ 7:29 PM |
PRAMOD KUMAR G |
pramodkumargopisetty@gmail.com |
https://www.linkedin.com/in/pramod-kumar-gopisetty-487861223/ |
https://enwinove.com/ |
7673989701 |
CEO, With Industrial expertise & a 2023 graduate with lots of passion towards solving two big problems for all " Carbon emission from industries & Water contamination" |
We team of 3 started working on this problem since from our college & its been 4+years we have been working |
one |
Unity though we all are not working full but when situations fall we unity for solving it, Decision making & the productivity output we all combinedly make is out par with any other teams |
ENWINOVE SOCIAL SOLUTIONS PVT LTD |
enwinove.com |
Anantapur |
CLOUDEE delivers solar-powered industrial heat and water treatment to cut costs, emissions, and waste. |
We’re solving the problem of high energy costs, carbon emissions, and inefficient water treatment in MSME industries. Many rely on electricity, LPG, or diesel for heating and struggle with treating wastewater affordably, leading to pollution and operational burden. Existing solutions are either costly, complex, or not practical for small industries. CLOUDEE addresses this by providing a clean, cost-effective solar thermal alternative that reduces fuel dependency while enabling decentralized water treatment and reuse without disrupting operations. |
CLOUDEE is a solar thermal system that generates high-temperature heat using a solar concentrator and uses it for industrial processes like water heating and evaporation-based treatment. It integrates with existing setups in a hybrid, retrofit manner, so industries can adopt it without changing their operations. The system reduces dependence on electricity and fossil fuels while enabling decentralized water treatment and reuse. Designed to be simple, scalable, and service-based, CLOUDEE helps industries cut costs, lower emissions, and manage water more sustainably. |
CLOUDEE is unique because it targets industrial heat + water treatment together, not separately. Most solutions focus only on electricity (solar PV) or only on wastewater, while we address a major overlooked need—process heat—using solar thermal. It’s also designed for real adoption: a hybrid, retrofit system that works with existing setups, so industries don’t have to change operations or risk downtime. Our approach focuses on practical heat delivery and efficient transfer, not just generation. On the business side, our WaaS/ETaaS model removes upfront cost barriers, which is critical for MSMEs. The defensibility comes from the integrated system design (optics + heat transfer + application) and the real-world performance data we’re building through pilots, making it hard to replicate without similar learnings. |
two |
one, Pilots |
Industries we can do Effluent treatment & can provide sustainable Heating (Manufacturing, Textile, Pharma....), For Individual Communities(Used water treatment & water conservation)... |
$50+ Billions |
$10Billion |
$1Billion in 5years |
B2B(Direct Product sales, Effluent Treatment as a Service for industries & also Sustainable heating), B2G, B2C |
VA tech wabag, Ion exchange, Indra water, Lambert tech innovations,... |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 19, 2026 @ 5:46 PM |
onali baldiwala |
support@oser.ai |
https://www.linkedin.com/in/onali-baldiwala |
http://www.oser.ai |
9925037866 |
founder |
We met through professional collaboration and shared interest in technology and innovation. Over time, we started working closely on projects, and for the past 4–5 years, we’ve been building and refining ideas together, leading to our focus on developing a GenAI SaaS platform. |
1 |
Our biggest strength as a team is our structured collaboration—combining technical expertise, problem-solving ability, and long-term alignment. With 4–5 years of working together, we’ve built strong trust, clear role distribution, and the ability to execute efficiently on complex GenAI SaaS challenges. |
oser.ai |
www.oser.ai |
Bangalore |
oser.ai enables operators to manage, optimize, and monetize EV infrastructure through a single AI-powered platform. |
Managing EV infrastructure and fleets requires multiple systems, creating complexity, data silos, and missed revenue opportunities. |
oser.ai helps businesses efficiently run EV networks and fleets by automating operations, improving utilization, and providing actionable insights. |
oser.ai’s defensibility lies in its agentic AI layer that integrates data across infrastructure and fleets, enabling automated decision-making and compounding operational intelligence. |
MVP |
Revenue |
OEM, CNO |
Global EV infrastructure & fleet software market |
EV charging operators & fleet managers needing unified platforms |
: Early adopters in target regions adopting AI-driven EV operations |
oser.ai generates recurring revenue through tiered subscriptions, with additional income from premium analytics, integrations, and enterprise services. |
Key competitors include ChargePoint and similar EV charging platforms, but oser.ai stands out by integrating charging, fleet operations, and agentic AI into a single system. |
oser.ai acquires customers through a hybrid strategy of inbound content, performance marketing, and targeted outbound sales to EV operators, fleets, and municipalities. |
oser.ai will deploy a hybrid go-to-market strategy combining product-led growth, targeted outbound sales, and ecosystem partnerships to penetrate the mobility sector and acquire 100+ clients within 6 months. |
Our vision is to build the intelligent backbone of future mobility by enabling autonomous, efficient, and scalable EV infrastructure worldwide. |
Founded and incorporated in 2023, the company is actively developing and deploying its AI-driven platform for EV infrastructure and fleet management. |
oser.ai is currently bootstrapped, with initial development and growth funded by the founders, demonstrating capital efficiency and strong execution. |
Yes |
We are applying to Avinya to raise capital, access strategic mentorship, and fast-track the growth of our AI-powered EV infrastructure platform. |
We aim to refine our product, validate market fit, and accelerate the growth of our GenAI-powered mobility platform through structured mentorship and execution. |
Yes, we are open to hybrid or virtual formats and committed to actively participating |
Yes |
Automation layer |
Our AI architecture combines scalable cloud infrastructure, real-time data processing, and agentic AI agents that continuously learn and automate decision-making across EV infrastructure and fleets. |
Our platform generates proprietary, real-time operational data across EV infrastructure and fleets, enabling our agentic AI to continuously learn, optimize, and create a compounding data advantage. |
Our AI is defensible through the combination of proprietary cross-system data, agentic AI orchestration, and continuous learning loops that create a compounding intelligence advantage. |
We assess model performance using accuracy, latency, and drift detection, and ensure reliability through continuous monitoring of uptime, error rates, and system resilience in real-world operations. |
We handle data privacy and security through end-to-end encryption, role-based access controls, and compliance with industry standards, supported by continuous monitoring, audits, and secure cloud infrastructure. |
Regulatory risks include changes in EV infrastructure standards, energy policies, and data protection laws. we mitigate these through a flexible architecture and proactive compliance practices. |
At 10x scale, challenges may include increased latency, data pipeline bottlenecks, and model drift. we address these through scalable infrastructure, continuous monitoring, and model optimization. |
Yes, our in-house AI/ML team has strong expertise in RAG systems, vector search, and scalable AI infrastructure, enabling real-time insights and intelligent automation across our platform. |
We are currently in the early stage and not yet using proprietary datasets, but plan to combine licensed data sources with platform-generated operational data to build a strong data foundation. |
We continuously improve our models through real-time data ingestion, feedback loops, drift detection, and periodic retraining, enabling our system to learn and adapt over time. |
oser.ai is designed for both India and global markets, with a focus on scalable infrastructure that enables expansion across diverse mobility ecosystems. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/igtKNQC1drMN-Fortitude-Mobility-Pitch-Deck_Final_2026.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/igtKNQC1drMN-Fortitude-Mobility-Pitch-Deck_Final_2026.pdf |
http://N/A |
Yes, we are building a mission-driven startup focused on transforming EV mobility by making charging infrastructure and fleet operations more efficient, intelligent, and sustainable through AI. |
N/A |
Applied independently |
checked |
| Apr 19, 2026 @ 3:40 PM |
Aayushi Singh |
aayushi@magicroll.ai |
https://www.linkedin.com/in/asaayushi2025/ |
https://magicroll.ai/ |
08102014565 |
Rana Vivek Singh (CEO, CTO). A GSOC(Google Summer of code) scholar, he previously built a Microsoft Teams application that was awarded as the world’s best app and was a founding team member at VideoWiki(backed by Wikipedia). With deep knowledge of AI product architecture, Rana has led multiple funded product ventures like MTN, AU Bank, Quickrr. Aayushi Singh(CPO). She is a recognized Outreachy alumna from the world’s largest open-source program for women in tech. With a strong background in AI-driven motion design and a proven track record of building a creative automation startup with over fifty lakh rupees in annual revenue, Aayushi brings unmatched expertise in gtm and product Pawan Kumar(COO) and a serial entrepreneur. He brings over ten years of experience across product delivery,and operations. He has worked with Zerodha, Capgemini, and edtech ventures, giving him a cross-functional edge in growth and team building. The founding team works full-time on Magicroll AI and is backed by leading institutional programs including Ministry of I&B, Samridh 2.0 MeitY, Startup Maharathi, STPI IMAGE CoE, Elevate Karnataka, and the Change Engine Accelerator supported by Y Combinator, MIT veterans and VP'S of MMT, Spinny. |
We have known each other for over five years, we first met as college mates and have collaborated on multiple tech projects since then. We have worked together in person |
All |
Our biggest strength is our team's rare combination of world-class technical depth and real market execution experience , a GSOC scholar and Microsoft award-winning builder leading product, an Outreachy alumna who has already built a creative automation startup to ₹50L+ ARR, and a serial operator with 10+ years across Zerodha, Capgemini, and edtech. Together, we've not only built and shipped AI products at scale, but have already validated MagicRoll AI with real government and media partners, cutting workflows from hours to minutes. That blend of pedigree, execution, and early traction, backed by MeitY, Ministry of I&B, and a Y Combinator-supported accelerator, is what sets us apart. |
Magicroll AI |
https://magicroll.ai/ |
Bangalore |
AI Video Infrastructure Built for Scale |
Our focus is on content editing bottlenecks, which remain a major hurdle with the models currently available in the market. For SMBs and enterprises, the use cases are very niche and highly direct, but they cannot be fulfilled by existing AI tools because they are templated, tend to hallucinate, lack guardrails, and do not support assisted fine-tuning with deep contextual understanding. We are solving exactly this gap. |
Magicroll AI is a deep-tech autonomous video editing engine that uses multimodal generative models, computer vision, and contextual learning to understand and learn a creator’s unique editing style, perform semantic scene understanding on raw footage, and automatically compile motion graphics, B-roll, and multi-language captions into platform-optimized video exports at scale.
Why now: Because SMBs and enterprises are stuck stitching together multiple disjointed “AI” tools for each part of the video workflow, and the current crop of AI editors is cluttered, templated, and often hallucinates forcing constant re‑prompting instead of delivering reliable, context‑aware outputs end‑to‑end. |
Editing Infrastructure Layer A proprietary video engine built from the ground up, not a wrapper on existing tools. It handles semantic scene parsing, motion graphics compilation, B-roll sequencing, and multi-format rendering in one unified pipeline, making speed, quality, and scale simultaneous outputs rather than trade-offs. This depth takes years to rebuild, creating a strong technical barrier to entry. Contextual Brand Identity Engine Goes beyond automation by learning a brand or creator's visual grammar over time, including pacing, tone, color language, caption style, and storytelling structure. This contextual memory compounds with every export, building a living brand style guide baked into the editing layer itself. For enterprises and government bodies managing consistent multi-language content at scale, it becomes a mission-critical dependency, not just a productivity tool. The Team as a Moat A Microsoft award-winning AI builder, a creative automation founder who has already scaled a venture to ₹50L+ ARR, and a decade-long operator across Zerodha and Capgemini, all working full-time. Backed by MeitY, Ministry of I&B, and a Y Combinator-supported accelerator, this combination of deep-tech expertise, go-to-market execution, and government-grade trust is not something a competitor can hire or buy overnight. Compounding Lock-In All three layers reinforce each other, making MagicRoll increasingly irreplaceable the more a team relies on it. |
Users |
Revenue, Pilots |
SMBs, Enterprises, Brands Agencies |
We are building for a converging global AI video and avatar market of roughly USD 150B+ by the early–mid 2030s, as video creation, editing, and synthetic presenters move fully to AI-first workflows. |
SAM — USD 12–18B The serviceable slice focused on AI-powered video editing and avatar-led content automation for enterprises, media houses, government bodies, and regional content creators across Asia-Pacific and emerging markets, where multilingual, high-volume video production needs are most acute and underserved. |
SOM — USD 400–600M MagicRoll's realistic near-term capture across Indian government and public sector content pipelines, edtech platforms scaling regional language courses, enterprise marketing teams automating brand content, and creative agencies requiring platform-optimized, multi-language video at scale, anchored by existing institutional partnerships with Ministry of I&B, MeitY, and STPI. |
Currently Creator Plan + Enterprise Model (Credit Based) + (Pay as you Use ) |
Invideo, Pictory AI, Captions AI |
We acquire customers through a combination of warm institutional referrals and direct enterprise outreach. Our government pilots with MIB, Aakashvani, and News Wing act as high-trust proof points that open doors with enterprise buyers. Marquee clients like Nandus, who generated 600K+ views using MagicRoll, create referral loops and social proof within the brand and agency ecosystem. Our backing from IIT Delhi Samridh 2.0, IIM Punjab, and Razorpay Rize gives us curated access to SME and enterprise networks directly aligned with our ICP. As we scale, we are building self-serve onboarding and in-product virality to shift from outbound to product-led growth, reducing CAC over time. Launched on Product Hunt and rank #1 globally giving visibility in US market |
Our GTM is structured around six levers with India enterprise consolidation as the primary focus, scaling paid revenue from our two validated ICPs: edtech platforms and brand or creative agencies. We run enterprise direct sales through institutional networks from IIT Delhi, IIM Punjab, and government programs, converting pilots into multi-seat licensing agreements. By June 2027 we launch AI-automated social distribution across Instagram Reels, YouTube Shorts, and TikTok, adding a strong stickiness driver. Our pay-as-you-go credit model lowers the trial barrier and scales revenue naturally with customer output. US market entry is planned for August 2026, targeting mid-market brands, marketing agencies, and edtech platforms using India PMF as the proof point. |
MagicRoll AI's long-term vision is to become the default AI video intelligence layer for every brand, institution, and creator that needs to communicate at scale across languages, formats, and platforms. We are building toward a world where a single content brief autonomously generates, edits, and distributes platform-optimized video in any language without human intervention. Our proprietary Cortex engine, co-developed with IIT Delhi and protected by a joint patent, is the foundation of this vision, a domain-specific model that generalist tools cannot replicate. As the global AI video and avatar market converges toward USD 150B+ by the early-mid 2030s, MagicRoll aims to own the infrastructure layer that powers this shift across emerging markets and beyond. |
Incorporated in Karnataka as a Private Ltd Entity |
1.32 Cr INR |
Yes |
Avinya's focus on accelerating AI-based startups in Karnataka aligns directly with where MagicRoll AI is right now — live in production, ICP validated, and ready to scale distribution with the right institutional support and network. As an Elevate Karnataka grant winner already embedded in the Karnataka startup ecosystem, Avinya feels like the natural next step to access curated enterprise introductions, investor relationships, and the mentorship needed. |
During the 4-month cohort, we want to convert our existing government and enterprise pilots into paid contracts, close our first multi-seat licensing deals with edtech and agency clients, and build the GTM muscle needed for our August 2026 US market entry. The seed funding, mentor network, and Demo Day visibility from Avinya would directly accelerate our path to the September 2026 fundraise. |
Yes |
Yes |
Core engine |
MagicRoll AI is built on a multimodal generative stack we call Cortex, our proprietary video intelligence engine trained specifically on advertising and branded content data. It combines computer vision for semantic scene understanding, large language model layers for script and caption generation, and a motion design automation layer for B-roll sequencing and graphics compilation. The architecture is designed as an end-to-end pipeline where raw footage goes in and platform-optimized, multi-language video comes out without manual intervention. Cortex is being co-developed with IIT Delhi under Samridh 2.0 with a joint patent filing in progress, making it a domain-specific model purpose-built for video production workflows. |
Our primary data advantage comes from our government pilots with MIB, Aakashvani, and News Wing, which gave us access to a rare corpus of regional language, institutional-grade video content that no commercial dataset replicates. Our client work with brands like Nandus, Beautifula, and Destora AI continuously feeds real branded video performance data back into Cortex, creating a proprietary feedback loop tied to actual audience outcomes. The joint patent co-filing with IIT Delhi further formalizes IP ownership over our training methodology and domain-specific model architecture. This combination of government-sourced, brand-performance, and regional language data is not something a generalist model or competitor can easily acquire or replicate. |
Cortex is trained exclusively on advertising and branded content data, giving it a domain specificity that generalist video AI tools like Runway or Pika fundamentally lack. The proprietary training data, institutional IP co-ownership with IIT Delhi, and a patent in progress create compounding legal and technical moats around the core model. Beyond the model itself, MagicRoll's contextual brand identity engine learns each client's unique visual grammar over time, making the system increasingly personalized and sticky with usage. Rebuilding this combination of domain-specific training, contextual memory, and institutional IP partnership would take a competitor years, not months. |
We evaluate Cortex across four primary dimensions: output quality scored against human editor benchmarks, platform engagement metrics from client video performance data, processing speed and rendering latency at scale, and caption accuracy across regional languages measured against native speaker review. Client repeat usage and credit consumption rates serve as our real-world PMF signal, with Nandus, Beautifula, and Destora AI providing continuous feedback loops. We also track semantic accuracy of scene understanding by comparing automated B-roll selections against manually curated edits. |
All client video assets and brand data are stored with strict access controls and are never used to train shared models without explicit consent, ensuring brand confidentiality is preserved. Our architecture is designed with data isolation per client, meaning one enterprise's style data and footage never bleeds into another's model outputs. Given our government pilots with MIB and Aakashvani, we have been held to institutional-grade data handling standards from early on, which has shaped our compliance posture. We are aligning with India's Digital Personal Data Protection Act 2023 as we scale and will implement SOC 2-equivalent controls ahead of our US market entry in August 2026. |
The most relevant regulatory risk is India's Digital Personal Data Protection Act 2023, which governs how user and client data is stored, processed, and shared, directly impacting our data pipeline and model training practices. For our government content workflows, there are content sensitivity and broadcast compliance requirements under the Ministry of Information and Broadcasting that we must navigate carefully, though our existing pilot relationships give us a strong compliance baseline. In the US market, evolving AI content disclosure laws and synthetic media regulations could require us to label AI-generated or AI-edited video, which we are building flagging infrastructure for ahead of our August 2026 entry. Copyright and IP ownership of AI-generated video outputs remains an emerging global regulatory question that we are proactively addressing through our IIT Delhi patent co-filing. |
The most immediate bottleneck at 10x scale would be rendering infrastructure, as our video compilation and export pipeline is compute-intensive and currently optimized for current client volumes rather than concurrent enterprise-scale workloads. Cortex's contextual brand memory system would also face latency challenges as the number of individual brand profiles being tracked simultaneously grows, requiring a shift from our current architecture to a more distributed model serving layer. Regional language caption accuracy could degrade at scale if we encounter dialects or low-resource language variants underrepresented in our current training data. We are proactively addressing this by designing Cortex for horizontal scalability |
Yes, we have a dedicated in-house AI team anchored by our CEO Rana Vivek Singh, a GSOC scholar with deep experience in AI product architecture who has led multiple funded AI ventures. Our CPO Aayushi Singh brings applied AI expertise in motion design and creative automation, with research published at ICoRD 2027 at IISc Bangalore on Design with AI on behalf of MagicRoll. Our team has grown from four to seven members this year with specific hires made against Cortex model development gaps identified from pilot learnings. The Cortex engine is being built entirely in-house with academic collaboration from IIT Delhi, ensuring both research depth and product applicability. |
We are using 3 types of datasets:
1) Those are available at platform like huggingface, civit and more and are licensed.
2) Synthetic Datasets, which we have created in house using our pipelines.
3) Inhouse datasets which we collect from our beta users and freemium users to train the model better. |
We have integrated the RL framework with multiple weights and biases which is smart enough to learn from the users heatmaps and prompts. Also have enabled a human in a loop structure in it. |
We are building for both, with India as the deliberate proving ground before global expansion. FY2027 is focused on consolidating paid enterprise revenue in India across edtech platforms, brand agencies, and government content pipelines, using this as the PMF foundation. US market entry is planned for August 2026, targeting mid-market brands, marketing agencies, and edtech platforms in a B2B-only motion with no consumer or self-serve play at this stage. The multilingual and regional content infrastructure we have built for India is itself a global differentiator, as the same architecture applies to any market where brands need to communicate across languages and platforms at scale. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qeul4hHBoRCc-Magicroll-Deck-Updated_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qeul4hHBoRCc-Magicroll-Deck-Updated_compressed.pdf |
https://www.loom.com/share/b2d166e0daef4458b9f11cd60be10a92?sid=aa6b6c13-497e-48ff-a343-f72136e67e78 |
We are building for both, with India as the deliberate proving ground before global expansion. FY2027 is focused on consolidating paid enterprise revenue in India across edtech platforms, brand agencies, and government content pipelines, using this as the PMF foundation. US market entry is planned for August 2026, targeting mid-market brands, marketing agencies, and edtech platforms in a B2B-only motion with no consumer or self-serve play at this stage. The multilingual and regional content infrastructure we have built for India is itself a global differentiator, as the same architecture applies to any market where brands need to communicate across languages and platforms at scale. |
Yes, MagicRoll AI is co-founded by Aayushi Singh, our CPO, who is a recognized Outreachy alumna from the world's largest open-source program for women in tech, making us a women-co-founded deep-tech startup. She has independently built a creative automation venture to ₹50L+ ARR, bringing both representation and proven execution to the founding team. MagicRoll was also recognized at the AI by Her Global Summit and ranked in the Top 30 globally, a platform specifically celebrating women-led innovation in AI. This is not a token representation but a core part of our founding DNA, with Aayushi leading product vision, GTM strategy, and research published at IISc Bangalore on behalf of the company. |
NA |
checked |
| Apr 19, 2026 @ 1:59 PM |
NagaSwetha Devarakonda and Nachiketa Anshu |
Swetha@tieopp.com |
https://www.linkedin.com/in/swethadn/ |
http://tieopp.com |
9343600316 |
CEO and CTO - 22 years of corporate tech experience across the globe |
Classmates - Know each other from last 26 years - have worked together for last 20 years on several engagements. |
All |
Our biggest strength is our ability to convert ideas into real, working systems with measurable outcomes.
We are not a purely conceptual team—we have a strong execution backbone. We combine product thinking, technology depth, and real-world problem understanding to build solutions that are already being used and validated.
For example, with TieOpp, we have:
Onboarded 5,000+ students from 150+ colleges in Karnataka and have MoU with 10+ leading colleges in Mysore and Bangalore
Built and deployed AI-driven employability readiness scoring
Enabled live internships and hiring outcomes
This reflects our core strength:
👉 We don’t just ideate—we build, deploy, iterate, and scale. |
TieOppConnect Private Limited |
tieopp.com |
Mysuru |
AI-powered employability platform that helps colleges measure, improve, and connect student readiness to real internships and jobs. |
We are solving the gap between academic learning and actual employability. Today, a large percentage of students graduate without being job-ready, while companies struggle to find candidates with the right skills and mindset. Colleges lack a structured way to measure employability, identify skill gaps, and guide students toward industry readiness. As a result: Students apply blindly without knowing where they stand Colleges cannot track or improve real placement outcomes Enterprises face high screening effort and poor candidate quality |
TieOpp is an AI-powered employability platform that helps colleges systematically measure, improve, and connect student readiness to real opportunities.
Our solution works in three layers:
Measure: We generate an Employability Readiness Index (ERI) using AI-driven skill assessments and Personality Insights to benchmark each student.
Improve: We provide structured roadmaps, targeted learning interventions, and an AI Video Coach to help students close identified gaps.
Achieve: We match students to internships and jobs through AI-based skill and personality alignment, reducing hiring friction for enterprises.
The platform integrates students, colleges, and enterprises into a single ecosystem—enabling data-driven placements instead of guesswork. |
Our uniqueness lies in combining measurement, improvement, and placement into a single closed-loop system—something existing solutions treat in isolation. Proprietary Employability Readiness Index (ERI): We don’t just list jobs—we quantify student readiness using AI-driven skill and Personality Insights, creating a standardized benchmark for employability. Closed-loop improvement engine: Unlike job portals or LMS platforms, we don’t stop at assessment. We provide structured roadmaps and interventions (including AI Video Coach) to actively improve outcomes. Multi-sided ecosystem with data network effects: As more students, colleges, and enterprises use TieOpp, our matching accuracy improves—creating a compounding data advantage that is difficult to replicate. Deep integration with colleges (distribution moat): We work directly with institutions, embedding into their placement workflows, making us part of their core system—not an external tool. |
Users |
Users, Pilots, Testimonials |
Colleges are our primary customers, while students are end users and enterprises are hiring partners in our ecosystem. |
₹50,000+ Cr (~$6B+) India has ~40 million higher education students. Assuming an average annual employability platform spend of ₹1,000–₹1,500 per student: 4 Cr students × ₹1,250 avg = ~₹50,000 Cr TAM |
₹8,000–₹10,000 Cr Initial focus on: Engineering, management, and professional courses ~8–10 million students across ~10,000+ colleges 80 lakh × ₹1,250 = ~₹10,000 Cr SAM |
₹100–₹250 Cr Realistic capture: 200–500 colleges ~1–2 lakh students onboarded 1.5 lakh × ₹1,500 avg = ~₹225 Cr SOM |
We monetize through institutional subscriptions and scale additional revenue via hiring, skilling, and certification layers built on top of our employability data platform. |
Our competitors fall into three categories, none of which solve the problem end-to-end like TieOpp: Job Portals (e.g., Naukri, LinkedIn): Focus on job listings and applications, but do not assess or improve student employability. Assessment Platforms (e.g., HackerRank, AMCAT): Provide skill testing, but lack continuous improvement pathways and institutional integration. LMS / Skilling Platforms (e.g., Coursera, upGrad): Offer courses, but are not connected to measurable employability outcomes or hiring pipelines. |
We follow a focused B2B2C go-to-market strategy, with colleges as the primary entry point. Direct outreach to colleges (Primary channel): Founder-led and relationship-driven outreach to principals, TPOs, and management, supported by demos, pilot programs, and MoUs. On-ground ecosystem presence: Participation in events, hackathons, and academic partnerships (e.g., TieCon, college engagements) to build trust and visibility. Student-led adoption within colleges: Once onboarded, students drive internal virality through peer usage, improving engagement and retention. Enterprise onboarding through demand pull: Companies are attracted by access to pre-assessed, job-ready talent, and are onboarded via targeted outreach and existing networks. Digital channels (supporting): LinkedIn outreach, email campaigns, and content-led engagement to generate inbound interest. |
Our GTM is institution-first, outcome-driven, and network-led—creating a compounding advantage as more students, colleges, and enterprises join the platform. |
We are building the infrastructure layer for employability—where readiness data replaces resumes and drives how talent is developed and hired at scale. |
Private Limited Registered on 15 Feb 2024 |
Bootstrapped |
Yes |
We see Avinya as a force multiplier—helping us move faster from traction to scale by unlocking the right mentorship, partnerships, and ecosystem leverage. |
We aim to scale to 100+ colleges, strengthen product-market fit, build enterprise demand, and establish a repeatable go-to-market engine for rapid growth. |
Yes, we are flexible and open to hybrid or virtual participation, and fully committed to engaging consistently across all programme activities. |
Yes |
Supporting feature |
Our AI architecture is built as a modular employability intelligence layer that supports assessment, scoring, matching, and guidance.
At a high level, it includes:
Data Ingestion Layer: student profile data, resume data, assessment responses, internship/job role requirements, and interaction history
Feature & Scoring Layer: structured extraction of skills, behavioral indicators, and assessment performance to generate our Employability Readiness Index (ERI)
Matching Engine: AI-based matching between students and opportunities using skill relevance, readiness scores, and Personality Insights
Recommendation Layer: personalized roadmaps, skill-gap identification, and intervention suggestions to improve employability
Feedback Loop: outcome data such as applications, shortlists, internship progress, and hiring results continuously improve scoring and matching quality over time
The architecture is designed to be modular, explainable, and scalable, so that each layer can improve independently while contributing to a stronger closed-loop employability system. |
We are building a proprietary employability dataset combining readiness scores (ERI), skill and personality signals, and real hiring outcomes—creating a closed-loop data advantage that improves our models over time. |
Our AI is defensible because it is built on proprietary, outcome-linked data and a closed-loop learning system integrated deeply with institutions.
Proprietary, outcome-linked dataset:
Our models are trained on structured data combining skills, Personality Insights, and real hiring outcomes—not just resumes or test scores—making predictions more contextually accurate.
Closed-loop learning system:
We continuously learn from the full lifecycle (assessment → improvement → application → hiring), allowing our scoring and matching models to improve over time.
Institutional integration (data access moat):
Direct integration with colleges ensures consistent, large-scale, and structured data inflow that competitors cannot easily replicate.
Multi-dimensional intelligence (not single-signal AI):
We combine skill signals, behavioral traits, and progression data, making our models richer than typical keyword or test-based systems.
Switching cost & compounding advantage:
As colleges and enterprises rely on ERI and historical data, switching away means losing accumulated insights—creating a strong retention and data moat. |
We measure success not just by model accuracy, but by outcomes—how effectively our AI improves employability scores and converts candidates into real hiring outcomes. |
We implement privacy-by-design with user consent, RBAC, encryption, and secure cloud infrastructure, aligned with DPDP Act and global best practices, along with audit trails and controlled data usage for AI models. |
Key risks include data protection (DPDP), AI fairness, and education sector regulations. We mitigate these through privacy-by-design, explainable models, institutional partnerships, and compliant cloud architecture. |
At 10x scale, the main risks are inconsistent input data, model drift across diverse cohorts, rising inference costs, and operational bottlenecks in quality control. We address this through stronger data pipelines, monitoring, modular AI services, and feedback-driven retraining. |
Yes, our AI is led by Pruthvi Raj Venkatesh, a GenAI scholar with 23+ years of experience, international publications, and patents. The team specializes in NLP, matching systems, and scalable AI architectures for employability scoring and recommendations. |
We primarily rely on proprietary data (student profiles, assessments, and hiring outcomes) collected with consent, supplemented by public skill taxonomies. We are not dependent on licensed datasets, and our core data is owned and continuously growing. |
We use a closed-loop system combining outcome-based feedback, periodic retraining, segment-wise monitoring, human feedback, and A/B testing to continuously improve model accuracy and relevance. |
India-first, with plans to expand into global emerging markets such as Southeast Asia and the Middle East. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/1tJxOxLz2p2w-Pitch_Deck_TieOpp_ACSEL.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/1tJxOxLz2p2w-Pitch_Deck_TieOpp_ACSEL.pdf |
http://tieopp.com |
Yes, we are mission-driven, focused on improving employability outcomes by helping students become job-ready and connecting them to opportunities, especially for underserved institutions. |
Women led from Tier2/3 city of Mysuru |
KDEM Sudheer Shankar |
checked |
| Apr 18, 2026 @ 10:49 PM |
Anil D,Madhu J R |
anil.dandin@gmail.com |
https://www.linkedin.com/in/anil-d-9b224872/ |
http://www.easyeducation.in |
+918762012312 |
CEO,CSE software developer,CTO,Software developer |
We are the Husband and wife,Started this startup past 7 months we are working on this. |
1 |
As a team we are strong in Software development and Well experienced teacher from NIIT background to deliver the product. |
Easy Education |
www.easyeducation.in |
Bengaluru |
**AI-powered learning platform delivering interactive 2D/3D lessons, AI tutoring, and analytics to make quality education affordable and accessible.** |
Students rely on rote, text-heavy learning that fails to build real understanding, while schools lack affordable tools for interactive teaching and real-time performance insights. |
We are building an AI-powered AR/VR learning platform that uses immersive 3D animations to make complex concepts easy to understand. Our solution transforms traditional textbooks into interactive visual experiences, supported by an AI tutor for personalized guidance and real-time learning analytics, making high-quality education engaging, effective, and accessible on existing devices. |
Our solution combines AI, AR/VR, and immersive 3D learning into one affordable platform that works on existing school devices—removing the need for expensive hardware. It is defensible through our proprietary interactive content pipeline, AI tutor trained on structured curriculum (NCERT-mapped), and a growing ecosystem of student data and learning analytics that continuously improves personalization and outcomes. |
MVP |
Signups, Testimonials |
Our target customers are K–12 students (Grades 8–12), schools (private and government), teachers, and coaching institutes—especially in Tier 2/3 markets seeking affordable, high-quality digital learning solutions. |
₹83,000 CRORES |
₹10,000 CRORES |
8–10 CRORES |
We operate a hybrid B2B + B2C SaaS model: schools pay a per-student annual subscription for platform access, while students can subscribe directly at an affordable yearly price. Additional revenue comes from AI-powered add-ons (tutoring, analytics), competitive exam prep, and teacher productivity tools offered on a subscription or usage basis. |
Your current positioning is good, but saying “we have no competitors” would hurt credibility. You **do have competitors**, but your edge is different. Use this: **Main competitors include BYJU'S, Vedantu, Unacademy, and Toppr, along with emerging AI learning tools. However, unlike them, we focus on affordable, AI-powered AR/VR learning with immersive 3D content and real-time analytics, designed to work on existing school infrastructure—making advanced learning accessible at scale.** |
We acquire customers through direct school partnerships and pilot programs, supported by demos and educator networks. Additionally, we leverage digital marketing, social media, and referrals to reach students directly, along with collaborations with NGOs and institutions for wider adoption. |
We follow a bottom-up + institutional GTM: start with pilot programs in schools to demonstrate outcomes, convert them into paid subscriptions, and scale through referrals and educator networks. In parallel, we drive B2C adoption via digital marketing and social media, while partnering with NGOs and institutions to expand into Tier 2/3 markets at scale. |
Our long-term vision is to build a global AI-powered learning ecosystem where immersive AR/VR and 3D experiences replace rote learning, making education highly interactive, personalized, and accessible to every student—turning any classroom or device into a smart learning environment. |
Incorporated |
NA |
Yes |
We are applying to Avinya to accelerate the development and validation of our AI-powered AR/VR learning platform. The program’s mentorship, ecosystem access, and support will help us refine our product, strengthen our go-to-market strategy, and scale our impact in making high-quality education affordable and accessible to students across India. |
During the program, we aim to: Pilot our solution in multiple schools and validate learning outcomes Enhance our AR/VR and AI capabilities for immersive education Build strong partnerships with institutions and educators Refine our business model and prepare for large-scale deployment Secure early traction and readiness for funding and expansion |
Yes, we are fully open to hybrid and virtual participation and can actively engage in all program activities in either format. |
Yes |
Core engine |
Our AI architecture is built on a modular, cloud-native stack combining LLMs, RAG, and multimodal capabilities. We use LLM APIs (such as DeepSeek/Claude/Gemini) for the AI tutor, integrated with a Retrieval-Augmented Generation (RAG) pipeline over structured NCERT-aligned content. The system supports text, voice, and image inputs, enabling multimodal learning experiences.
The backend is powered by serverless architecture (AWS Lambda, API Gateway) with DynamoDB for user data and entitlements, and S3 + CloudFront for scalable content delivery. We use embeddings and vector search for context retrieval, ensuring accurate, subject-bound responses.
The AI layer integrates with AR/VR and 3D content modules to provide guided explanations, adaptive learning paths, and real-time feedback. The system continuously improves through user interaction data and learning analytics. |
We are building a proprietary dataset of structured, curriculum-aligned learning content mapped to NCERT, combined with user interaction data such as learning patterns, concept-level performance, and engagement behavior.
This includes how students interact with 3D/AR/VR content, AI tutor queries, and assessment responses. Over time, this creates a strong data moat that improves personalization, recommendation accuracy, and learning outcomes—making our AI system more effective and difficult to replicate. |
Our defensibility comes from a combination of proprietary curriculum-aligned content, multimodal AI (text, voice, and visual understanding), and deep integration with AR/VR 3D learning experiences. Unlike generic AI tools, our system is subject-bound (NCERT-mapped) and continuously improves using real student interaction data, learning patterns, and performance analytics. This creates a strong data moat and personalized learning engine that becomes more effective over time and difficult to replicate. |
We evaluate performance using both AI and learning outcome metrics, including:
Accuracy and relevance of responses (against curriculum benchmarks)
Retrieval precision in RAG pipelines
Student engagement (time spent, interaction rate)
Learning improvement (pre/post assessment scores)
Response latency and system uptime
Feedback signals (user ratings, corrections, fallback rates)
These metrics help us continuously refine model quality and ensure reliable learning experiences. |
We follow a privacy-first approach with secure, cloud-based infrastructure. All user data is encrypted in transit and at rest, with strict access controls and authentication mechanisms (e.g., Cognito-based identity management). We minimize data collection, anonymize learning analytics, and ensure compliance with applicable data protection standards. We also implement role-based access, secure APIs, and monitoring to prevent unauthorized access and ensure system integrity. |
Yes, potential risks include data privacy regulations (especially for minors), content copyright considerations (curriculum usage), and evolving AI governance policies. We mitigate these by using our own structured content with proper references, implementing strong data protection practices, and staying aligned with regulatory guidelines. Our architecture is flexible to adapt to future compliance requirements as policies evolve |
At 10x scale, the main challenges would be increased latency, higher inference costs, and load on retrieval systems (RAG pipelines). Real-time AI responses and multimodal processing (voice, image, AR/VR interactions) could face performance bottlenecks.
We are addressing this through a serverless, auto-scaling architecture, model optimization, caching strategies, and efficient vector search systems to ensure scalability without compromising user experience. |
Yes, we have an in-house team with expertise in AI integration, prompt engineering, RAG systems, and full-stack development. The team has hands-on experience working with LLM APIs (Claude, Gemini, DeepSeek), building multimodal AI applications, and deploying scalable cloud-based systems on AWS.
We also combine technical expertise with deep product understanding in EdTech, enabling us to design AI systems specifically optimized for learning outcomes. |
We primarily use structured, curriculum-aligned content based on NCERT textbooks, which we transform into our own proprietary explanations, formats, and interactive learning modules.
In addition, we generate and own our own datasets through 3D/AR/VR content, assessments, and user interaction data (learning behavior, queries, performance). Over time, this creates a growing proprietary dataset. We ensure all content usage follows appropriate guidelines and is used as reference while building original learning experiences. |
We follow a continuous improvement loop driven by data and feedback. This includes analyzing student interactions, AI tutor queries, and learning outcomes to refine prompts, retrieval systems, and content.
We use feedback signals (ratings, corrections), A/B testing, and performance metrics to optimize accuracy and personalization. As our dataset grows, we enhance our RAG pipelines and fine-tune system behavior to deliver more precise, context-aware, and effective learning experiences over time. |
We are building for India first, focusing on scalable adoption in Tier 2/3 markets, and plan to expand to global markets with localized content and curriculum alignment. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/g79B7bGQfbeD-Final-Pitch-1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/g79B7bGQfbeD-Final-Pitch-1.pdf |
https://www.linkedin.com/posts/anil-d-9b224872_easyeducation-edtech-aiineducation-ugcPost-7450126794386509824-hfx-?utm_source=share&utm_medium=member_desktop&rcm=ACoAAA9Xd5YBAjw-HMNNNSgMn1B01-uh3QjKJKY |
Yes, we are a mission-driven startup focused on making quality education accessible and affordable for every student. We aim to solve the gap between traditional rote learning and real conceptual understanding by using AI, AR/VR, and immersive 3D learning.
Our primary focus is on students in Tier 2/3 regions who lack access to advanced learning tools. By enabling engaging, visual, and personalized education on existing devices, we reduce dependency on expensive coaching and infrastructure, improving learning outcomes at scale. |
NA |
K-TECH |
checked |
| Apr 18, 2026 @ 9:03 PM |
Rishabh Choudhary, Nishant Gupta, Malyaj Masane, Hitesh Shivkumar, Kaustubh Trivedi |
rishabhc1003@gmail.com |
https://www.linkedin.com/in/rishabh-choudhary-ba150b21b/ |
http://NA |
9808421111 |
Nishant-CEO, Btech Computer Science, Rishabh-COO, Btech Mechanical, Hitesh-CTO, Btech Computer Science, Malyaj-CPO, Btech Mechanical, Kaustubh-CMO, Btech Mechanical |
All of us met at the start of our Engineering in College, and since then for 4 years we have been room mates. We have been working on BlankSage for about 4 months now. |
All |
Our Biggest Strength is our Team bonding as we we have been friends for 4 years together we can work with speed and efficiency. Also all of us being from different fields and background gives us an edge over all the different aspects of a Startup including Technical, Business and Management. |
BlankSage |
blanksage.com |
Bengaluru |
AI First Exam Evaluation Infrastructure |
We are solving the problem of Slow and Inconsistent nature of manual evaluations across the education system today. |
BlankSage is an AI-first examination infrastructure built to eliminate the inefficiencies of manual evaluation across India’s education system.
We automate the assessment of answer sheets across multiple formats, including handwritten exams, coding tests, MCQs, and digital assignments.
BlankSage delivers accurate, rubric-driven evaluations with complete transparency. Every result includes a detailed breakdown of marks, clearly showing where a student performed well and where they need improvement.
By combining speed with consistency, we enable institutions to deliver near-instant feedback, so students can understand and act on their performance immediately after an exam. We reduce teacher effort by 90% so they can focus on important things like creating better learning material for students or contributing to Research |
BlankSage is defensible because it is not just an AI tool, it is a full-stack evaluation infrastructure tightly integrated with how assessments are created, conducted, and graded. We not only handle correction but also giving feedback to students focusing on their learning and improvement |
Users |
Users, Testimonials |
Schools, Junior Colleges, Higher Education Institutes, Coaching Centers, Individual Educators, Individual Students |
1,75,00,00,00,00,000 Rupees |
7,90,00,00,00,00,000 Rupees |
1,20,00,00,00,00,000 Rupees |
B2B Saas, D2C |
Vedaai, Codetantra, Examai, Classi |
We acquire customers through direct outreach to schools and coaching institutes, pilot programs, and referrals driven by measurable time and cost savings in evaluation. Starting with our own College and School |
We target schools and coaching institutes, land through pilot programs with teachers, demonstrate ROI via faster and consistent evaluation, and expand institution-wide through referrals and administrative buy-in. |
BlankSage aims to become the global infrastructure for assessments, powering how exams are created, evaluated, and analyzed across schools, universities, and professional certifications.
Over time, we are expanding into an AI-native LMS where evaluation, feedback, and learning are seamlessly integrated into a continuous, personalized education loop. |
Incorporated as Pvt. Ltd. |
NA |
Yes |
We are fresh college graduates and dont have as much industry experience as other. we would like to connect more so we can have a deeper understanding of the problems. The financial support would help accelerate our growth |
deeper understanding of the problem, better business sense, accelerate the growth of our platform |
Hybrid |
Yes |
Automation layer |
We use a serverless, asynchronous architecture to handle AI workloads.
Requests go through a FastAPI microservice that acts as a gateway, handling preprocessing (OpenCV, PyPDF) and routing jobs to an AWS SQS queue. A separate GPU-based microservice consumes these jobs and runs the AI models.
For on-prem deployments, we use Qwen-3.5-VLM (35B) for OCR and basic evaluation, while exploring Gemma-4 models for improved reasoning. This setup ensures full data control but is expensive to run 24/7 due to GPU costs.
For our cloud SaaS, we optimize for cost and scalability by using Gemini 3 Flash for OCR and GPT-4o-mini for grading and content generation. |
For the time being, we do not have this advantage. But currently, we are in the process of procuring pilot deals, demo programs and case study groups, in which we will collect anonymized data (legally) for model fine tuning and improvement.
We have the required infrastructure, expertise and knowledge to be able to procure, clean and fine-tune/train models once this data is acquired, and all our systems are being built keeping in mind that it has to be honed and perfected using collected data. |
Our system isn’t a single AI model, it’s a multi-step workflow involving preprocessing, artifact generation, and output creation.
Our context engineering has been refined across diverse, messy real-world data, giving us strong performance in unstructured scenarios.
This creates a clear edge: achieving this level of reliability requires significant testing, effort, and domain expertise, making it difficult for competitors to replicate without substantial investment. |
Our OCR models are regularly checked for their performance using Character Error Rate (CER) and Word Error Rate (WER) as the major metrics. Apart from that, we also track the mean, median and p95 latency for the models to process requests. Throughput is also measured using pages per second.
We are still currently in the process of devising accuracy and performance based metrics for exam evaluations and corrections/grading. |
We design our systems to align with standards like GDPR, ISO 27001, and SOC 2.
All PII is encrypted at rest (AES) and in transit (SSL/TLS). Our infrastructure runs on Amazon Web Services with strict IAM-based access control.
We support production-grade multi-tenancy. Authentication is handled via WorkOS (SOC 2 compliant), while authorization is managed in-house using role-based access control with fine-grained feature flags. |
We face three key risks:
1. Data for improvement – Ensuring compliant use of non-PII data to enhance grading and extraction systems.
2. Content IP – Managing rights and permissions for user-uploaded reference materials used in evaluation.
3. Regulatory alignment – Staying compliant with policies like National Education Policy and Ministry of Education guidelines. |
We use a distributed, asynchronous (producer–consumer) architecture, where scaling is primarily handled through autoscaling on Amazon Web Services by increasing consumers as workload grows.
Our current limitation is limited large-scale testing. In preliminary tests, a single L40S GPU handled ~9,000 OCR pages/hour (the slowest step). Beyond that, scaling should be linear by adding GPUs, but real-world concurrency performance still needs validation. |
We do not have an official full time AI/ML team. The current product building has mainly been done by our founding team of 5 members.
But, out of this founding team, both the CEO and CTO have excellent AI/ML expertise and domain knowledge. We are both well versed with the latest technologies and modern AI/ML approaches.
Additionally, we both have multiple patents, conference and journal publications at high quality venues (IEEE, Springer, Scopus Indexed work etc). |
For OCR fine-tuning, we use the IAM Handwriting Dataset (~5,600 labeled sentences).
For end-to-end validation, we currently rely on a small dataset of 20–25 corrected answer scripts with corresponding question papers and answer keys.
This dataset will expand in May–June through planned case studies with students, tuition teachers, and test-based learning institutes. |
As customer data grows, we’ll periodically fine-tune and retrain models, benchmarking performance against the current baseline.
Production models will be monitored continuously, with checks every 3–6 hours to detect performance drops and trigger alerts.
Long term, we aim to explore dynamically learning models that adapt to incoming data, though this requires rigorous validation for reliability and trust. |
Our current aim is for Indian markets, due to our expertise and domain knowledge regarding the Indian Education System. But our systems and architecture are generalized in nature, allowing for usage in global market scenarios as well in foreign education systems with very little modification. Our system is modular through well defined interfaces and is easily modifiable for client needs. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/poLxvtEwbZ1w-BLANKSAGE-PITCH-DECK-1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/poLxvtEwbZ1w-BLANKSAGE-PITCH-DECK-1.pdf |
https://drive.google.com/file/d/1ma31TrWwapiWHQeUEJldL80d4xpYWT08/view?usp=sharing |
Yes, this is an impact-focused startup.
We’re tackling a systemic issue in education: inconsistent and error-prone evaluation that affects fairness, student confidence, and learning outcomes at scale. Our goal is to ensure every student is assessed accurately and consistently, regardless of who evaluates them.
The impact is direct and measurable, reduced evaluation errors, faster feedback cycles, and more equitable outcomes for students across schools, coaching centers, and institutions.
Long term, we aim to enable a shift from marks-based evaluation to feedback-driven learning, improving how students understand and act on their performance. |
NA |
Linkedin, Startup Grants India |
checked |
| Apr 18, 2026 @ 12:31 PM |
efdghjkl;' |
xgfhcvjkbln;m', |
http://zxfchgvjkhbnlj;mk',; |
http://www.retdfyghujk;l.com |
2345678901 |
.otiruetyrsgxfcvnbnnkjiyutytdfxfvcbnjhuytfdfcvbjhguytdfx |
loiy8ti7rueyrsgzxfbvcgvhbkjlkm.nklgiu.r7lie875rfyugbhpg98xoi6cjrtvgbvo58rugiyturdrgfx |
All |
liutyuhdfxgbv nbjkgjh |
rxdtfcygvhbjk |
www.uhij.com |
gyvhbji |
wasedrftgyhuij8y7t6r5drxfcgvhu. |
jiuyftdfxcgvhbjhoiyt7r6td |
sdrhtjfkghllbykvtjchrxgfrghkbvcedhvyfrcfeserjcufythdrtsxhcsvdbmfhgfcftsrtmgfxb |
8y76ryertsegfxchvjgkjb |
two |
Signups |
etdrfyguhjk |
cfvbn |
bvm |
tfyuih |
576r8t79y8u9 |
ghjlk |
dyrtfugihij |
ycgvhbkjn |
09u89t7r86tfuyhuo8ygt7iyv |
ytfugihojp |
234567u8io |
one |
ytguhijokpo |
rydtfugihojik |
xcvbnm |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 17, 2026 @ 6:32 PM |
Archana Parameswaran, Sundar Raj |
thefragmentapp@gmail.com |
https://www.linkedin.com/in/archanaparameswaran/ |
https://www.thefragment.app/ |
9108027243 |
Co founders, IT Product Management, Development, Deployment |
Raj and I worked together in our previous organization in the same project, when we discussed the idea, we decided to work towards it. We know each other for 4 years now. In this product we have worked for more than 2 years together. |
one |
Our biggest strength is our resilience, clear vision for the product and our combined technical experience of 25+, which we bring along across different platforms like aaS, IoT, web and mobile applications, fintech, and edtech. We’ve built and scaled products at every stage—from zero to launch, as well as improving and operating live systems. We’ve led end-to-end execution: shaping product vision, building and managing teams, and delivering reliable, user-focused solutions. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 17, 2026 @ 11:06 AM |
Sandeep Kumar Suna |
sandeep.suna@gmail.com |
https://www.linkedin.com/in/sandeep-suna/ |
http://NA |
9986333161 |
16 Years of Corporate Experience. Ex-Oracle & Ex-Microsoft |
We have been discussing about the idea since last 6 months. I was exploring talent who support my vision and have the skill set to collaborate and I was introduced by a mutual friend. Since then we are collaborating well. |
two |
We have Gen Y & Gen Z combination. One had 16+ years of experience with big enterprises on Strategic Planning & Execution whereas the other have deep technical expertise to convert requirement into workable solution. We together fill the gap and perfectly solving problem for India's Education & Employment Challenges. |
Techprapt Solutions Private Limited |
NA |
Bangalore |
Our vision is to empower each individual to build India's Own Technology and become self-sufficient as a country. |
We are building product which primarily solve Indian problem and empower them toward better opportunities & Growth |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 17, 2026 @ 10:43 AM |
Sujata N Bhavikatti, Kusuma Jolad |
sujatabhavikatti1879@gmail.com |
http://www.linkedin.com/in/sujata-bhavikatti-a947b0290 |
https://www.mpsitlabs.in/index.html%20 |
09036140581 |
Sujata Bhavikatti - Principal Investigator will lead the complete project design, implementation, testing, validation of results and reporting. Sujata completed her PhD in digital VLSI domain" Cloud video transcoder-low power approach" . She is heading 2 funded projects. |
We met at Krishi mela in Dharwad in the year 2018. During casual meet as both are from same place, expressed interest as an entrepreneur. We stayed in touch and collaborated during our next visit to exhibition. After PhD I began the work of developing products and scale up the business. |
two |
Our biggest strength is the ability to collaborate effectively with diverse background and experience. One is expert in technical skills and entrepreneur contacts. Co founder has strong managerial skills and marketing experience. This helps in connecting with 3rd party for outsourcing of project work and build contacts for local market. |
Manjunatheshwara Power Systems labs Innovation and technology PVT LTD |
https://www.mpsitlabs.in/index.html |
Belagavi |
Technology for the farmers and society benefit |
Enable small and medium farmers to access the market updates such as new agri tools, facilities by government, fertilizers, awareness of local plants and their use, overall to increase the income |
App in regional language that enables connecting small and medium farmers to the agricultural tools available in market, reduce the labor and cost requirement, support in finding ways to improve the income using agriculture techniques, information about warehouses and identify the medicinal plantations grown randomly in field and suggest where to connect and how to use those for medicine, cosmetics. |
The solution connects to farmers as per their vegetation, local essence and customized solution as per the geographical nature of field and gives innovative solutions improve their income with minimum investment |
one |
one |
small and medium scale farmers of North Karnataka region and then extend to south Karnataka finally to regions. Ayurvedic medicinal plants and research units, Cosmetics companies |
|
|
|
|
AgroStar, DeHaat etc. Our key competitors include platforms like AgroStar, DeHaat, and AgriApp, which offer advisory and input marketplaces; however, we differentiate by providing a localized, AI-driven, and integrated solution that also addresses unexplored areas like medicinal plant utilization and hyper-local farmer needs. |
Through on ground farmer networks, partner with FPOs, local Entrepreneurs, |
with geographically divided database of a. medicinal and herb requirement centers, 2.local plants and herbs 3. utilisation possibilities, 4. cosmetics industries |
My vision is to increase the vegetation knowledge of farmers and village community so that the vegetation is conserved. this also has direct impact on environment conservation. The dependency on chemicals reduces. Build an income model for farmers so that their dependency on others for information and awareness reduces. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 16, 2026 @ 9:12 PM |
Arpit Agrawal |
arpit@angoor.ai |
https://www.linkedin.com/in/arpit-agrawal-273813137/ |
https://angoor.ai |
09131289932 |
Arpit Agrawal (CEO - Sales and Operations), Anuj Agrawal (CTO - Tech and Product), Rishabh Kumar (CPO - Product and Design) |
The founding team met at IIT Bombay, where we built a strong academic and professional bond over shared interests in technology and entrepreneurship. We have been working together for over 4yrs, collaborating on projects that laid the foundation for Angoor AI. Our complementary skill sets spanning AI/ML, product development, and business strategy have allowed us to move fast and stay aligned on both vision and execution since day one.
Arpit and Anuj are siblings, Anuj and Rishabh are batchmates |
All |
Our biggest strength is the combination of deep technical expertise and sharp business acumen rooted in our IIT Bombay background. We've built a genuinely AI-native product from the ground up, not a wrapper on top of existing tools, which speaks to our engineering depth. At the same time, we've been able to close real customers across two distinct ICPs (D2C Shopify brands and Real Estate aggregators), reaching $40K ARR, which shows our ability to translate technology into market traction.
Beyond credentials, we move fast as a team. We've shipped complex features like multi-agent shared context management, omnichannel flow builders, and API tool integrations, things even well-funded competitors haven't cracked with a lean team of 10. Our culture of ownership and execution is what sets us apart. |
Angoor AI Private Limited |
https://angoor.ai |
Bengaluru |
AI native Customer Interaction Platform for B2C businesses |
B2C businesses lose customers and revenue managing fragmented tools across WhatsApp, Instagram, email, and calls — with no unified AI layer to automate and personalise the entire customer journey. |
Angoor AI is an AI-native CRM that automates the entire customer lifecycle from pre-sales to support across WhatsApp, Instagram, email, calls, and Shopify, from a single platform. Our multi-agent system shares context across all channels, so every customer gets a personalised, seamless experience without businesses needing to scale human resources. What took multiple tools and teams now runs on one intelligent platform. |
1. Truly AI-native architecture - Unlike legacy CRMs bolted with AI, we were built ground-up with multi-agent shared context management, meaning all our agents understand each other's actions in real-time across every channel. No competitor is doing this at our level. 2. Omnichannel Flow Builder - Our multi-channel flow builder with simplified business logic (written in plain English, executed by our proprietary Angoor Engine) is something none of our direct competitors have built. This creates deep workflow lock-in once a business is onboarded. 3. Data moat over time - As businesses run on Angoor AI, we accumulate rich customer interaction data that powers increasingly personalised AI agents trained specifically on their customers. The longer they stay, the smarter their agents get - making churn nearly impossible at scale. Combined with our IITB founding team's technical depth and early traction with real customers, we're building a platform that becomes infrastructure - not just a tool. |
Revenue |
Users, Revenue, Testimonials |
D2C Shopify brands and Real Estate aggregators |
$184B |
$785M |
$5M |
Monthly Active Customers (Customer sending a message in a month is considered as an active engaged customer (inbound unique customer)) |
Salesforce, Yellow AI, Freshdesk, Bitespeed |
1. Community & Network (Primary - Current Stage) and cold outreach Leveraging our IITB alumni network, GrowthX, and Razorpay founder communities for warm introductions. This has been our most cost-effective channel and drove our initial $100K ARR. Trust and word-of-mouth conversions are highest here. 2. Channel Partnerships - partners working on a commission model 3. Paid & Organic (Scaling) - SEO-driven content targeting high-intent keywords |
Phase 1 (Now) - Closing initial customers: 15 customers in pipeline through founder networks. Use these as case studies and testimonials to build credibility. Phase 2 (Scaling) - Target D2C Shopify brands and Real Estate aggregators through referral programs, agency partnerships, and SEO content. WhatsApp and Instagram are our primary outreach channels for these ICPs. Phase 3 (Growth) - Expand globally using paid ads, influencer marketing in the B2B SaaS space, and a self-serve onboarding model to reduce sales cycle length. |
To become the AI infrastructure layer for B2C customer interactions globally, the single platform that powers every touchpoint between a brand and its customers, across voice, WhatsApp, Instagram, email, and beyond.
The way we see it, what Gumloop proved for B2B internal automations, Angoor AI will prove for consumer-facing B2C automations, one intelligent, unified stack replacing the fragmented mess of tools brands stitch together today. Five years from now, any enterprise B2C brand should be able to run their entire communication stack, pre-sales to support, outbound to inbound on Angoor AI, without writing a single line of integration code. |
Incorporated in Oct 2023 |
2Cr |
Yes |
Avinya is built for AI-native startups, and that's exactly what we are. Most generic accelerators don't deeply understand agentic systems or omnichannel AI the mentorship here is actually relevant to what we're building. Bengaluru is also where we want to scale our enterprise sales, and ACSEL's network gives us the right room to do that. The timing aligns perfectly we have traction, we need to scale, and Avinya is the right launchpad for that next step. |
We want to use the programme to do three things: accelerate product development on our voice AI and agentic flows, close our first enterprise customers in Karnataka leveraging ACSEL's network, and get fundraise-ready for our next round with a sharp investor narrative and demo day presence. Beyond numbers, we want to give back to the cohort as much as we take we've learned a lot building Angoor AI and believe the best accelerator experiences are ones where founders learn from each other as much as from mentors |
Open to Hybrid participation |
Yes |
Core engine |
Angoor AI is built on a multi-agent, multi-modal, omnichannel architecture with three core layers:
1. Agent Layer - Multiple specialised AI agents (sales, support, logistics, etc.) operate simultaneously across channels. What makes us different is shared context management; all agents share the same context window, understanding each other's actions in real-time and handing off to the right agent or human seamlessly.
2. Channel Layer - Native integrations across WhatsApp, Instagram, email, voice, and Shopify. Our architecture is provider-agnostic, meaning that when platforms like Meta release new APIs or features, we can integrate without restructuring our core.
3. Data Layer - A CDP-esque ETL pipeline that pulls customer data from multiple sources (Shopify, legacy platforms, external APIs), structures it, and feeds it to agents in real-time. This gives every agent unified, governed access to customer context at every touchpoint.
On top of this sits our Angoor Engine, a proprietary orchestration layer that translates business logic written in plain English into executable agentic workflows, making complex automation accessible without engineering effort.
We use frontier models for voice and conversation, with a feedback loop directly to model providers to continuously improve real-world performance. |
Our data advantage compounds over time in two ways:
1. Interaction Data Flywheel - Every customer conversation across WhatsApp, Instagram, voice, and email that runs through Angoor AI feeds back into our platform. Over time, this builds a rich, structured dataset of B2C customer behaviour patterns, intents, drop-off points, and conversion triggers that we use to continuously improve agent performance. The longer a brand stays on Angoor AI, the smarter their agents get, making churn nearly impossible.
2. Unified Customer Profiles - Unlike point solutions that only see one channel, we aggregate every touchpoint, including purchases, conversations, support tickets, and calls, into a single customer profile. This gives us a depth of behavioural context that no single-channel tool can replicate, and it becomes a genuine moat as the dataset grows.
The honest answer is we're still early in fully exploiting this advantage, but the architecture was deliberately designed to capture and structure this data from day one, which puts us well ahead of competitors who'd have to rebuild their foundations to get here. |
Four things make our AI genuinely hard to replicate:
1. Multi-agent shared context - All our agents share the same context window across every channel simultaneously. Competitors either handle one channel at a time or stitch agents together loosely. Ours is natively orchestrated, a fundamentally different and harder architecture to build.
2. Proprietary Angoor Engine - Our orchestration layer translates plain English business logic into executable agentic workflows with full context management. This isn't a prompt wrapper; it's a purpose-built execution engine that took significant engineering effort to get right.
3. Data moat - Every interaction on our platform deepens customer profiles across channels. Agents trained on a brand's own customer data become increasingly accurate over time, creating switching costs that grow with usage.
4. Provider-agnostic infrastructure - We're not locked into any single LLM or channel API. We can swap underlying models, integrate new platform features, and adapt to the rapidly evolving AI landscape without restructuring our core, something most competitors can't do without significant rework.
Ultimately, our defensibility isn't just one feature - it's the compounding effect of architecture, data, and execution that gets harder to catch up with every month we're ahead. |
We evaluate across three layers:
1. Conversation Quality Metrics
Intent recognition accuracy - did the agent correctly understand what the customer wanted?
Resolution rate - percentage of queries fully resolved without human intervention
Escalation rate - how often and at what points agents hand off to humans
First response time and average handling time across channels
2. Business Outcome Metrics
Conversion rate – did the AI interaction lead to a purchase, booking, or lead qualification?
Customer satisfaction score (CSAT) collected post-interaction
NDR resolution rate for D2C clients (Shopify order-related flows)
Lead qualification accuracy for real estate and aggregator ICPs
3. Model-level Reliability Metrics
Hallucination rate - tracked through human-in-the-loop review flags
Context retention accuracy across multi-turn conversations
Latency per response, especially critical for voice agents, where sub-second responses matter
Fallback rate - how often the model fails to generate a usable response
Beyond internal metrics, we maintain a direct feedback loop with our frontier model providers, sharing real-world conversation failures to improve model performance in our specific B2C use cases, something most platforms at our stage don't do. |
We are in the process of GDPR and ISO-27001 compliances |
A few we're actively tracking:
1. WhatsApp & Meta Policy Changes - As a platform built heavily on Meta's APIs, any changes to their business messaging policies directly impact us. We mitigate this by maintaining official BSP status and staying close to platform roadmaps, while keeping our architecture provider-agnostic so we're never over-reliant on a single channel.
2. Data Privacy Regulations - GDPR in the UK, DPDP Act in India, and anticipated US regulations around AI and customer data all apply to us. We're building compliance infrastructure proactively rather than reactively.
3. AI-specific Regulation - Emerging regulations around automated customer interactions, consent for AI-driven calls, and disclosure requirements are evolving fast. We stay ahead by building human-in-the-loop flows as a default, not an afterthought.
The honest answer is these risks are real but manageable, and being architecture-agnostic and compliance-first from day one is our best hedge. |
Three things we're actively thinking about:
1. Context Management at Volume - Our multi-agent shared context window is our biggest strength, but also our biggest scaling challenge. At 10x load, maintaining real-time context synchronisation across thousands of simultaneous conversations will stress our orchestration layer significantly. We're already engineering solutions here, but it remains our most critical scaling risk.
2. Voice Latency - Our voice agents need sub-second response times to feel natural. At 10x concurrent calls, infrastructure costs and latency management become non-trivial problems. We're currently stress-testing this.
3. Human-in-the-loop Bottlenecks - At the current scale, our escalation flows work smoothly. At 10x, if escalation rates aren't driven down first through better AI accuracy, the human oversight layer becomes a bottleneck that defeats the purpose of automation.
The good news is we've deliberately built a provider-agnostic, modular architecture so we can swap, upgrade, and scale individual components without rebuilding the whole system. We know where the cracks are, and that's half the battle. |
Yes. Our core team of 10 includes engineers who work directly on AI/ML infrastructure, agent orchestration, and model integration. The founding team from IIT Bombay brings strong technical depth in machine learning, systems design, and full-stack development.
Specifically, our in-house expertise covers:
Agentic orchestration - building and maintaining our multi-agent shared context management system and the Angoor Engine
LLM integration & prompt engineering - working with frontier models across conversation, voice, and multimodal use cases
Voice AI - active engineering on inbound and outbound voice agents, including latency optimisation and real-world performance tuning
Data pipelines - CDP-esque ETL infrastructure for structuring and feeding customer data to agents in real-time
We also maintain direct relationships with our model providers, giving us a feedback loop that most platforms at our stage don't have access to. Our roadmap includes LLMOps for prompt tuning and fine-tuned vertical models for specific enterprise use cases, which will deepen our in-house ML capability significantly over the next 12 months. |
We work with two categories of data:
1. Client-owned Customer Data
The primary datasets powering our agents are owned by our clients' customer conversations, purchase histories, CRM records, and interaction logs from platforms like Shopify, WhatsApp, and Instagram. This data is accessed under explicit client agreements, processed through our ETL pipelines, and never used across tenants or for model training without consent.
2. Foundation Models
For language and voice capabilities, we use frontier models from third-party providers under commercial licensing agreements. We don't train base models from scratch - our differentiation is in the orchestration, context management, and workflow layer built on top of these models.
We don't currently own large proprietary training datasets, but our architecture is deliberately designed to accumulate structured interaction data over time, which will form the basis of vertically fine-tuned models for enterprise clients on our roadmap.
To be straightforward: our current data advantage is architectural, not dataset-based. The proprietary dataset moat is something we're actively building toward. |
Three ways:
1. Real-world Feedback Loops
Every human-in-the-loop escalation, agent fallback, and CSAT score is a signal. We systematically capture these to identify where agents fail with wrong intent, lost context, poor resolution, and feed that back into prompt tuning and workflow improvement. We also maintain direct relationships with our frontier model providers to share real-world failure cases and influence model improvements.
2. LLMOps & Prompt Tuning
On our near-term roadmap is a dedicated LLMOps layer for systematic prompt versioning, A/B testing, and performance benchmarking across conversation flows. This moves us from an intuition-driven to a metrics-driven model improvement.
3. Vertical Fine-tuning
As we accumulate sufficient interaction data per industry vertical D2C, real estate, and student housing, we plan to fine-tune models specifically for those use cases. A model trained on thousands of Shopify customer conversations will outperform a generic LLM on order queries, returns, and NDR resolution every time.
The compounding effect is that every new client, every new conversation, and every new escalation makes the system measurably smarter, which is ultimately what makes the platform harder to replicate over time. |
India and UK |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/wkPTkP96tj7Q-Angoor-AI-Investor-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/wkPTkP96tj7Q-Angoor-AI-Investor-Deck.pdf |
https://www.youtube.com/watch?v=XdPsbyEoRbw&t=1s |
Honestly, we're building a business first, but the impact is real, and we don't shy away from it.
The businesses we serve are mostly small and mid-sized B2C brands that simply cannot afford to hire large customer support and sales teams. For them, Angoor AI isn't a nice-to-have; it's the difference between being able to compete with larger players or not. When a D2C wellness brand like Sanfe can resolve customer queries around the clock without burning out a small team, or when a UK student housing platform can qualify leads at 2 am across time zones, that's genuinely levelling the playing field.
At a broader level, we believe the next wave of AI value creation isn't in B2B productivity tools for large enterprises; it's in giving every B2C brand, regardless of size, the same quality of customer experience that only the biggest players could afford before. That's the mission we're quietly working toward.
We're not saving the planet, but we're giving smaller brands a real shot at building lasting customer relationships, and we think that matters. |
No |
Amit |
checked |
| Apr 16, 2026 @ 7:23 PM |
Manish shah |
manish@tansorpay.in |
https://www.linkedin.com/in/manish1712/ |
https://www.tansorpay.in/ |
07020147645 |
A designer who taught himself AI/ML, enrolled at IIT Patna, and founded India's first biometric UPI terminal is a memorable founder narrative. The design background explains why TansorPay's UX is frictionless. The AI/ML education explains how you can credibly lead a computer vision company. Most investors will remember that combination. |
We met online, and we have been working together for a year. |
All |
Our biggest strength is the **fusion of technical rigour and operational design.** Specifically, we leverage:
Multidisciplinary Expertise: We combine high-level AI/ML engineering with professional project management, allowing us to build complex biometric systems that actually work in real-world environments.
Rapid Execution: We possess the agility to move from a technical concept to a live pilot in high-traffic retail sectors, backed by a proven track record of institutional recognition.
Strategic Resilience: With a solid foundation in the fintech ecosystem (DPIIT, Razorpay Rize), we have the credibility and the network to scale complex hardware-software integrations. |
TansorPay |
https://www.tansorpay.in/ |
Mumbai |
Face-scan payments — no phone, no card needed. |
We are solving the friction and device dependency of modern checkout experiences. Currently, retail and hospitality payments require customers to rely on physical items—smartphones, credit cards, or cash. This leads to slower transaction times, inconvenience (forgotten wallets, dead phone batteries), and a fragmented customer experience. TansorPay solves this by completely removing the need for physical hardware on the consumer side, enabling instant, device-free payments through biometric face-scans. |
We provide a biometric payment infrastructure that enables instant, device-free checkout.
By combining specialised point-of-sale hardware with a highly secure processing backend, we allow customers to complete transactions using only a facial scan—eliminating the need for cards, cash, or smartphones. |
Our solution is highly defensible because it offers true device-free autonomy, completely eliminating the need for consumer hardware by relying entirely on secure facial biometrics. To support this, we provide purpose-built hardware and software infrastructure engineered specifically for the high-speed demands of physical retail. Furthermore, our early pilot launches and backing from elite fintech ecosystems create strong first-mover network effects and deep merchant lock-in. |
MVP |
Users |
Our target customers are high-traffic cafes, restaurants, and retail businesses seeking to provide a frictionless, device-free checkout experience for their patrons. |
50000 Cr |
12000 Cr |
120 Cr |
Terminal SaaS, Transaction Fee, Integration API, Hardware Margin. |
NA |
We acquire customers through targeted B2B pilot programs in high-traffic retail locations and by leveraging our strategic partnerships within elite fintech networks to drive merchant adoption. |
Our go-to-market strategy centers on executing targeted pilot programs in high-traffic Mumbai retail and hospitality venues to prove early traction, while leveraging our established fintech ecosystem networks to scale merchant adoption. |
Our long-term vision is to make "India pays with face" a reality by making TansorPay the ubiquitous standard for physical transactions across the country. We aim to deploy our biometric POS at every single payment touchpoint—from local kirana stores and shopping malls to hospitals, metros, and public transit networks. |
Done |
NA |
Yes |
We are applying to Avinya to secure crucial seed funding to scale our biometric POS infrastructure and to gain strategic mentor support to accelerate our go-to-market execution. |
During the programme, we aim to leverage the seed funding and mentorship to successfully scale our initial pilot operations, optimize our proprietary biometric hardware for mass deployment, and accelerate our merchant acquisition |
hybrid |
Yes |
Core engine |
Our AI architecture operates on a hybrid edge-cloud model: our purpose-built POS hardware locally processes real-time face detection and liveness anti-spoofing to ensure millisecond checkout speeds. At the same time, our secure cloud backend utilises advanced deep learning algorithms to match encrypted biometric templates and authorise the instantaneous tokenisation of payments. |
Our proprietary data advantage is the continuously growing, real-world dataset of facial biometric vectors mapped directly to secure financial tokens across diverse Indian demographics, which constantly trains our AI for unmatched accuracy and liveness detection. |
Our AI component is highly defensible because it tightly integrates our edge-cloud processing architecture with a continuously growing, real-world dataset of facial vectors mapped directly to secure financial tokens, creating a compounding data advantage that software-only competitors cannot easily replicate. |
We evaluate model performance and reliability by strictly monitoring False Acceptance Rate (FAR) and False Rejection Rate (FRR) for biometric accuracy, Liveness/Presentation Attack Detection (PAD) for anti-spoofing security, and end-to-end millisecond latency to ensure real-time checkout speeds. |
We handle privacy and security by strictly adhering to India's DPDP Act and the RBI’s 2026 digital payment guidelines, utilising edge-computing on our POS devices to instantly tokenise facial vectors so that no raw biometric images are ever stored or transmitted. |
The primary regulatory risks involve evolving RBI mandates concerning Additional Factor of Authentication (AFA) for purely device-free payments and stringent compliance requirements under the DPDP Act for biometric consent, both of which we actively mitigate through our zero-storage tokenisation architecture. |
If our user base scales 10x, the primary breaking point will be the 1:N vector search latency in our cloud backend, where matching a live face against a massively expanded database could delay our millisecond checkout speeds unless we transition to advanced distributed database sharding. |
No |
Owned |
training and improving the model. |
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/YtYAa3cGhWe3-TansorPay_Pitch.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/YtYAa3cGhWe3-TansorPay_Pitch.pptx |
https://youtube.com/shorts/MqBLDEuRAtQ?si=sGd6airJiySzISdm |
Yes, we are building a mission-driven startup focused on democratising digital payments across India by removing the dependency on smartphones and physical cards. By making device-free biometric checkouts the standard at every touchpoint—from local kirana stores to public transit—we ensure seamless financial inclusion and true transactional accessibility for everyone. |
NA |
NA |
checked |
| Apr 16, 2026 @ 5:01 PM |
Mr. Kiran Kumar D, and Mr. Ravi Kumar N |
ravi@harithavani.in |
https://www.linkedin.com/in/kk8892632742/ |
https://shop.harithavani.in/ |
08892632742 |
Kiran Kumar CEO has 12 years experience on business development and product research. Mr Ravi Kumar Heads production and Installation |
We meet at badminton tournament. We have been working since 5 years |
1 |
Our biggest strength as a team lies in our ability to combine deep problem understanding with practical, scalable innovation. We are not just building biogas plants—we are solving real-world challenges of waste management, clean energy access, and usability. By developing a sealed, odour-free system with patented balloon-based gas collection, we have addressed the key barriers that have historically limited biogas adoption. Additionally, our focus on **end-to-end ownership—from manufacturing to installation to continuous monitoring of every unit through unique IDs—**ensures consistent performance, customer trust, and long-term impact. This blend of innovation, execution, and customer-centric thinking is what sets us apart and drives our growth. |
Harithavani Renewable systems Pvt Ltd |
https://shop.harithavani.in/ |
bangalore |
Harithavani Renewable Systems builds compact, odour-free biogas solutions that convert everyday organic waste into clean cooking fuel and organic fertilizer. |
We are solving the problem of unmanaged organic waste and rising dependence on expensive, fossil-based cooking fuels by enabling households and institutions to convert their daily waste into clean biogas and organic fertilizer through compact, odour-free, and easy-to-use systems. |
We provide compact, fully sealed biogas systems powered by our patented balloon-based gas collection technology that convert daily organic waste into clean cooking gas and nutrient-rich organic fertilizer; designed for both urban and rural use, our systems eliminate issues like odour, leakage, and maintenance, and are supported by a smart monitoring approach where each installation is tracked through a unique ID to ensure consistent performance and long-term reliability. |
Our solution is defensible because it solves the core limitations that have historically restricted biogas adoption—odour, leakage, and usability—through a patented, fully sealed system with balloon-based gas collection, making it viable for urban environments where traditional systems fail. Beyond hardware, we differentiate through a product + service model, where every installation is assigned a unique ID and continuously monitored, ensuring performance, reliability, and long-term customer engagement. This combination of proprietary technology, superior user experience, and a data-driven operating layer creates strong barriers to replication and enables scalable, consistent deployment across diverse segments. |
Revenue |
Pilots, Signups |
Our target customers include urban households, apartments and gated communities, hotels and restaurants, institutions (such as hostels, schools, and temples), and rural households with access to organic waste, all of whom are looking to reduce waste disposal costs, lower LPG consumption, and adopt sustainable, decentralized energy solutions. |
Rs. 60,000 crore |
Rs. 20,000 crore |
Rs. 100 crore |
Our revenue model is primarily driven by the sale and installation of compact biogas systems, with additional recurring revenue from annual maintenance and monitoring services. We also generate value through customized solutions for institutions and bulk waste generators, and plan to expand into subscription-based service models, slurry (organic fertilizer) monetization, and potential carbon credit opportunities, creating both one-time and recurring revenue streams. |
biotech India, GAS Today, Sistembio |
We acquire customers through a combination of direct sales, strategic partnerships, and awareness-driven marketing. This includes on-ground demonstrations, referrals, and installations in visible locations (apartments, institutions, restaurants) that act as live showcases, along with digital channels like social media and inbound inquiries. We also collaborate with waste management partners, builders, and local bodies to access bulk customers, while leveraging word-of-mouth and performance-based trust to drive repeat and referral sales. |
Our go-to-market strategy focuses on high-impact early adopters and visibility-driven scaling. We begin with institutions, apartments, and HoReCa (hotels, restaurants, cafés) where daily waste generation is consistent and ROI is clear, enabling faster decision-making and strong reference cases. These installations act as live demonstration sites, driving word-of-mouth and credibility. We combine direct sales and on-ground demos with partnerships with waste management companies, builders, and local bodies to access bulk customers. Parallelly, we use digital channels and storytelling to build awareness around waste-to-energy benefits. As we scale, we plan to expand through a channel partner/distributor network in Tier 1 and Tier 2 cities, supported by a service and monitoring layer to ensure performance, retention, and recurring revenue. |
Our long-term vision is to build a decentralized waste-to-energy ecosystem where every household, community, and institution can convert their organic waste into clean energy and organic fertilizer at the source. We aim to make biogas a mainstream alternative to LPG, especially in urban settings, by combining innovative hardware with a smart monitoring network. Over time, we envision scaling into a pan-India and global cleantech company, enabling energy independence, reducing landfill waste, and contributing meaningfully to climate and sustainability goals. |
Pvt Ltd |
NA |
Yes |
We are applying to Avinya to accelerate our journey of building a scalable, decentralized waste-to-energy solution. Beyond capital, we value Avinya’s focus on purpose-driven innovation and long-term impact, which strongly aligns with our mission to solve waste management and clean energy challenges. We believe Avinya’s ecosystem, mentorship, and strategic guidance can help us strengthen our technology, refine our go-to-market strategy, and scale responsibly, while staying true to our vision of creating sustainable and inclusive energy solutions. |
During the programme, we aim to validate and scale our business for the next stage of growth. Specifically, we want to (1) achieve strong market traction through pilot deployments and customer validation, (2) refine our unit economics and financial model to make the business highly scalable and investor-ready, (3) strengthen our manufacturing and supply chain to support larger volumes, and (4) build the right go-to-market and partnership strategy for rapid expansion across cities. Additionally, we seek mentorship to sharpen our positioning, improve our storytelling, and prepare for successful fundraising and long-term scale. |
Yes, we are open to both hybrid and virtual participation and are flexible to engage in the format that best aligns with the programme’s structure and requirements. |
No |
Not applicable |
We currently do not rely on heavy AI-driven decision-making; however, we are building a lightweight, scalable data and monitoring architecture around our biogas systems. Each installed unit is assigned a unique digital ID, through which we capture operational data such as usage patterns, gas output consistency, and maintenance needs. This data layer enables predictive servicing, performance optimization, and customer support.
Going forward, we plan to integrate AI/ML models to analyze aggregated system data for failure prediction, output optimization, and demand forecasting, thereby improving efficiency at scale. Our architecture will evolve as a combination of IoT-enabled data collection, cloud-based storage, and AI-driven analytics, creating a smart, connected decentralized energy network. |
Our primary data advantage comes from our installation-level monitoring approach, where each biogas system is assigned a unique digital ID and tracked over time. This allows us to build a proprietary dataset on real-world waste input patterns, gas output performance, user behavior, and maintenance cycles across different environments (urban homes, apartments, institutions).
Over time, this creates a valuable, hard-to-replicate dataset that helps us optimize system performance, improve product design, and enable predictive maintenance. As we scale, this data layer becomes a strong moat, allowing us to develop AI-driven insights for efficiency, reliability, and demand forecasting, which traditional biogas players typically do not capture. |
Our AI component is defensible because it is built on proprietary, real-world operational data that competitors typically do not capture. Each installed system is uniquely identified and monitored, allowing us to continuously collect data on waste input, gas output, user behavior, and maintenance patterns across diverse use cases. This creates a compounding data advantage—the more systems we deploy, the better our models become.
Additionally, our AI is tightly integrated with our proprietary hardware and service layer, enabling use cases like predictive maintenance, performance optimization, and system-level benchmarking. This hardware + data + service integration makes it difficult for competitors to replicate without similar scale and long-term deployments, creating a strong and defensible moat over time. |
“We measure performance through prediction accuracy, reduction in system downtime, and real-world consistency of gas output across installations.” |
We take a privacy-by-design approach, ensuring that data collection, storage, and usage are secure, minimal, and compliant with applicable regulations.
Data Minimization: We primarily collect system-level operational data (e.g., gas output, usage patterns) and avoid capturing sensitive personal data unless absolutely necessary.
User Consent & Transparency: Customers are informed about what data is collected and how it is used, with clear consent mechanisms.
Secure Infrastructure: Data is stored on secure cloud platforms with encryption (in transit and at rest), along with controlled access and authentication protocols.
Access Control: Strict role-based access ensures that only authorized personnel can view or manage data.
Compliance Readiness: We align with relevant data protection frameworks (such as India’s DPDP Act) and follow best practices for data governance.
System Security: Regular monitoring, backups, and safeguards are in place to prevent data loss or unauthorized access.
Overall, our focus is to ensure that while we build a data-driven platform, customer trust and data security remain uncompromised. |
es, there are some regulatory considerations, but they are largely manageable and, in many cases, supportive of our solution.
Waste Management Regulations: Compliance with local municipal rules for handling organic waste (e.g., segregation at source, on-site processing norms) is required, but these policies actually favor decentralized solutions like ours.
Safety & Installation Standards: Biogas systems may need to adhere to basic safety guidelines (pressure handling, gas storage, installation practices), especially for institutional deployments.
Subsidy & Policy Dependence: Changes in government incentives or subsidy structures for biogas/renewable energy could impact adoption in certain segments.
Local Approvals: In some cases (apartments, commercial spaces), permissions from local authorities or housing societies may be needed.
Overall, the regulatory environment is more of a tailwind than a risk, as governments are actively promoting waste management, clean energy, and alternatives to LPG. |
At 10× scale, the biggest risks are not the models themselves but the data and system consistency around them.
Data Quality & Standardization: With rapid deployment across diverse environments, inconsistencies in input data (waste type, usage patterns, sensor reliability) can reduce model accuracy.
Hardware–Software Variability: Differences in installation quality or hardware performance can introduce noise, affecting predictions and reliability.
Sparse or Delayed Data Streams: If connectivity or data capture is inconsistent, real-time monitoring and predictions can weaken.
Model Generalization: Models trained on early datasets may not generalize well across new geographies and use cases without continuous retraining.
Operational Scalability: Managing and acting on insights (e.g., servicing thousands of units) can become a bottleneck without strong backend systems.
To address this, we are building for scale with standardized data pipelines, robust IoT integration, continuous model retraining, and automated alerting systems, ensuring reliability improves—not degrades—as we grow. |
We currently do not have a dedicated in-house AI/ML team, as our primary focus has been on building and deploying robust hardware and validating real-world use cases. However, we are developing a data-driven foundation through system-level monitoring, and we work with technical collaborators/advisors when needed.
As we scale, we plan to build a small, focused AI/ML team with expertise in IoT data processing, predictive analytics, and time-series modeling to enable features like performance optimization, failure prediction, and demand forecasting. Our approach is to layer AI on top of strong real-world data, ensuring practical and high-impact applications rather than purely experimental models. |
We primarily use internally generated, proprietary datasets collected from our installed biogas systems. This includes data on waste input patterns, gas output levels, system performance, and maintenance cycles across different user environments. This data is owned by us and forms the foundation of our monitoring and future AI capabilities.
In addition, we may use publicly available reference datasets and benchmarks (e.g., standard biogas yield ranges, environmental data) for calibration and validation purposes. These are typically open-source or publicly accessible, and we do not rely on any third-party licensed datasets as a core dependency.
Overall, our focus is on building a proprietary, real-world dataset, which becomes more valuable and defensible as we scale deployments. |
We plan to continuously improve model performance through a data-driven feedback loop tightly integrated with our deployed systems.
Continuous Data Collection: As more units are installed, we capture diverse real-world data on waste input, gas output, and system behavior, improving dataset quality and coverage.
Model Retraining: We periodically retrain models using updated data to ensure they adapt to new geographies, use cases, and operating conditions.
Performance Monitoring: We track metrics like prediction accuracy, error rates, and system uptime to identify drift or degradation early.
Feedback from Field Operations: Insights from servicing and customer usage are fed back into the models to improve predictions and reliability.
Segmentation & Localization: Models are refined for different customer segments (households, restaurants, institutions) to improve precision.
Iterative Deployment: We follow a test–learn–deploy cycle, rolling out improvements incrementally to ensure stability.
This approach ensures that our models become more accurate, reliable, and scalable as our deployment base grows. |
We are building for both India and global markets, with India as our primary launch and validation market. India offers a strong need for decentralized waste management and clean cooking solutions, allowing us to refine our product, unit economics, and deployment model at scale.
In parallel, we see significant global potential in regions across Asia, Africa, and Latin America, where organic waste management and access to affordable clean energy are major challenges. Our long-term vision is to build a globally scalable, decentralized waste-to-energy platform, starting from India and expanding to similar high-impact markets. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/BFQ3JdUPJYrk-Harithavani-Seed-fund-Investor-Pitch-Deck-1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/BFQ3JdUPJYrk-Harithavani-Seed-fund-Investor-Pitch-Deck-1.pdf |
https://youtu.be/HU84Ff-bLPw?si=2vm9qyna55wMzOUE |
Yes, we are a deeply mission-driven and impact-focused startup. Our core mission is to solve two critical challenges—organic waste management and access to clean, affordable energy—by enabling decentralized conversion of waste into biogas and organic fertilizer.
Every installation directly contributes to reducing landfill waste, lowering methane emissions, and decreasing dependence on fossil fuels like LPG, while also promoting sustainable agricultural practices through organic slurry. By making biogas systems compact, odour-free, and accessible, we are unlocking adoption in segments where traditional solutions have failed, especially in urban environments.
Our goal is not just to sell products, but to build a scalable ecosystem that drives environmental sustainability, energy independence, and circular economy adoption at a large scale. |
NA |
Social media |
checked |
| Apr 16, 2026 @ 4:13 PM |
Vedanth Yelemane ; Sripad Raghavendra ; Pramod Naik |
vedanth@creviz.io |
https://www.linkedin.com/in/vedanth-chikkapura-yelemane/ |
https://www.creviz.io/ |
9742259640 |
vedanth - Strategy and GTM ; sripad - platform and tech architect ; pramod - tech architect |
Our core team is built on deep, decade-long relationships and a shared history of building enterprise software together. We are a tightly knit triad, rather than a recently assembled team.
Vedanth & Sripad (13+ Years): The Domain Obsession
Sripad and I first met during our engineering studies at Visvesvaraya Technological University (VTU). Our working relationship began over a decade ago when we co-founded our first company in 2013. We were attempting to build the exact custom business solutions we are building today, utilizing the Salesforce/Force.com platform. While we ultimately had to sunset that venture due to the severe limitations and deficiencies of the platform at the time, that shared experience cemented our working dynamic and our obsession with the SME operational gap.
Sripad & Pramod (The Technical Core):
Sripad and Pramod forged their working relationship at a previous organization. Crucially, they were tasked with collaborating to build an internal low-code platform from the ground up. This intensive, high-stakes engineering environment perfectly synchronized their technical cadence and laid the architectural groundwork and mutual trust required to build the Creviz App Builder today.
Together, our intersecting histories mean we have already bypassed the typical "co-founder dating" phase. We have weathered failed platforms, built complex low-code architecture together, and are now perfectly aligned to execute on Creviz. |
All |
Our biggest strength is our deep domain obsession combined with a perfectly complementary execution triad: Business Strategy, Platform Architecture, and Engineering Excellence.
We have seen firsthand how rigid tools like Odoo and Zoho fail SMEs, and we are not tourists in this space. In fact, this is our second, much more mature attempt at solving this exact problem. Twelve years ago, we attempted to build custom SME workflows using Salesforce/Force.com. We learned the hard way what the market truly needed, and today, we have the right mix of maturity, technical capability, and Gen AI advancements to finally solve this at scale.
We each own our domains with complete mutual respect:
Vedanth (Strategy & GTM): Drives product vision and growth. With global experience across the US, Brazil, and India, and an MBA from a top entrepreneurial B-school, he leverages a deep worldwide alumni network to open doors, secure early heavyweight clients, and execute go-to-market strategies.
Sripad (Platform & Vision): The visionary behind the Creviz platform. He brings an entrepreneurial mindset to complex platform architecture and AI integration, focusing relentlessly on simplifying workflows and enabling non-technical users through robust no-code technology.
Pramod (Engineering & Execution): The technical engine. With 10+ years of exponential growth from QA to Full-Stack to Software Architect, he ensures unparalleled technical excellence. His rigorous work ethic and ability to translate complex SME constraints into scalable code keep our product delivery flawless.
We are obsessed with this space because we believe that empowering the world’s 500M+ SMEs with tailored, AI-driven operational tools is one of the most impactful things we can do. We are building alongside them, listening to them, and we’re just getting started. |
Creviz Software Solutions Private Limited |
https://www.creviz.io/ |
Bengaluru |
AI Powered No-Code Platform for Smart SMEs |
Small and medium enterprises (SMEs) form the backbone of India’s economy, contributing over 30% of GDP and employing 110 million people. Yet growth is slowed by a lack of affordable, adaptable technology. Unlike large corporations with ERP systems and IT teams, SMEs still depend on spreadsheets, manual processes, or fragmented tools. The result is inefficiency, errors, and limited visibility. Even a modest 1–2% efficiency gain across India’s 60 million SMEs could unlock a 0.5–1% uplift in GDP, highlighting the massive socio-economic opportunity. Current solutions don’t meet SME realities. ERP systems like Odoo or Zoho are rigid, consultant-dependent, and take months to implement. On the other hand, generic no-code platforms are not SME-friendly. They enable quick prototypes but often break in production and cannot adapt when workflows change, leaving SMEs exposed to operational failures. Creviz bridges this gap. It is the first AI-powered no-code platform purpose-built for SMEs, combining AI-driven app creation, error-proof interfaces, and built-in analytics to deliver enterprise-grade capability at SME scale. |
At its core, Creviz provides an end-to-end, AI-native operational platform that transforms how SME owners build, deploy, and scale their internal workflows. By combining natural language application generation with guided, consultant-free deployment and intelligent data analysis, the platform delivers a complete digital ecosystem. For the first time, SME leaders can rapidly digitize their unique workflows, manage complex operations without technical friction, and plan proactively with strategic insights once reserved for large enterprises.
Key Innovations:
Error-Proof AI Application Generation – SMEs describe their workflows in plain language. The system then regularly scores blueprints derived from these initial prompts and ongoing conversations. Once the system estimates the blueprint is error-proof, production-ready applications are built instantly.
SME-Friendly UI – Guided simplicity ensures immediate adoption by non-technical staff without the need for external implementations.
Flexible, Scalable Architecture – Applications evolve seamlessly alongside business needs, supporting highly dynamic and customized constraints.
AI Reporting & Planning Agent – Automatically structures cross-departmental data, generates instant reports, benchmarks performance against industry standards, and provides foresight for growth planning.
Creviz merges intelligent application creation, guided simplicity, and strategic foresight. It is more than a tool—it is the ultimate digital partner for SME owners, empowering them to thrive today and shape tomorrow, fueling India’s next wave of growth. |
The Unique Insight: > GenAI is currently transforming enterprise software because large companies have the internal IT teams to leverage it—winning the "build vs. buy" debate. However, SMEs are being left behind. They lack the technical staff, capital, and prompting expertise to build their own solutions using native AI models. Our Unique Solution: A Digital Workforce Creviz bridges this gap by delivering an AI-native operational platform designed explicitly for the SME owner. We aren't just providing software; we are deploying specialized AI agents that act as "digital employees": The PM Agent: Interfaces directly with the SME owner to understand operational bottlenecks and translate them into app requirements. The RPA Agent: Ingests, reads, and analyzes any document format across various customer use cases, automating tedious data entry and making the apps friction-free. The Reporting Agent: Accesses silo-free data across the entire platform to generate insights, index performance, and benchmark against global industry standards. Our Defensibility (The Moat): Single-Platform Data Gravity & Universal Agents Our strongest moat is our unified architecture. The market is currently being flooded with fragmented SaaS solutions and ad-hoc, "vibe-coded" AI apps, which create isolated data silos. Because Creviz builds every operational workflow on a single, unified platform, our AI agents have complete, cross-departmental context. Crucially, this standardized foundation means our AI agents are universally adaptable—they deploy and function seamlessly across entirely different SME customers and highly diverse custom applications without any technical barriers or integration friction. No standalone tool or fragmented tech stack can replicate the holistic operational control our ecosystem provides. |
Revenue |
Revenue |
SMEs which have atleast 10cr+ revenue but are using excel or spreadsheets to manage their operations |
400M SMEs globally, $2-$3T adoption potential |
30M SMEs worldwide with Annual Revenues between $1-$5M |
1500 SMEs (year 3) - 20000+ SMEs(year 6) |
One time Implementation cost + application saas billing |
The SME automation market is split between two extremes. ERP platforms such as Odoo, Zoho, and ERPNext are feature-rich but demand 3–6 month deployments, external consultants, and high ongoing costs. According to industry sources, 55–75% of ERP implementations fail to meet their objectives, and as many as 75% face delays, overruns, or project failure. This underscores why SMEs need affordable, adaptable alternatives designed around their unique operational realities On the other end, generic no-code platforms like Replit, Lovable, Emergent allow quick prototypes but rarely survive in production. SMEs struggle once workflows change—be it new compliance requirements, supplier shifts, or scaling demands—causing breakdowns that directly disrupt operations like order tracking, finance reconciliation, or inventory management. Creviz bridges this gap by delivering SME friendly, production-ready, adaptable apps powered by AI. Its differentiators include: AI-driven App Creation → instantly creates robust applications, not fragile prototypes. Continuously adaptable workflows → evolve seamlessly as SMEs grow, avoiding costly rebuilds. SME-first, error-proof UI → usable by non-technical staff, reducing reliance on IT consultants. AI Reporting & Planning Agent → a unique capability that lets SME owners generate instant reports, benchmark performance, and gain strategy insights—absent in both ERPs and generic no-code tools. USP: “Creviz is the only AI-powered no-code platform for SMEs that delivers production-ready apps and generates strategic AI insights— the ultimate partner for SME owners. |
Phase 0 to 1 (Current): Founder-Led & Unscalable We are currently acquiring customers through direct, 1-on-1 interactions, specifically targeting SME owners who have struggled with or abandoned traditional ERP implementations. Our primary channels are personal networks, industry events, and warm introductions leveraged through our core team's worldwide MBA alumni network. Phase 1 to 10 (Mid-Term): Referrals & Partnerships Our initial customers are industry heavyweights who are highly satisfied with our deployed solutions and are actively generating word-of-mouth referrals. To scale this, we are establishing channel partnerships with small-to-mid-sized software implementation firms, SME associations, CFO/CA forums, and financial outsourcing companies. Scale (Long-Term): Inbound & Outbound Engines We are laying the groundwork for a digital-first acquisition engine. We have hired our first SEO/GEO lead generation specialist to build domain authority and rank for high-intent keywords. Concurrently, we are running targeted outbound campaigns as we continuously narrow down our exact buyer persona. |
We have achieved Product-Problem-Solution fit and are actively iterating toward full Product-Market Fit. Our GTM strategy is built on a critical market insight: SME owners lack the time and technical expertise to use DIY app builders. They need a guided, "Do-It-For-You" approach. Step 1: Tech-Enabled Service (The Wedge) Instead of just selling software, we are building an in-house team of Business Analysts (BAs). Our BAs act as the bridge—they consult with the SME to map out their day-to-day operations and operational leakages. The BAs then use our proprietary App Builder to quickly deliver custom, edge-case-proof workflow applications. Step 2: Product Refinement via Internal Feedback Operating as our own first customer allows us to stress-test the App Builder in complex, real-world manufacturing and business environments. The continuous feedback loop from our BA team ensures the platform matures rapidly. Step 3: Platform Licensing (The Scale) Once the App Builder is sufficiently robust, our GTM shifts from a direct-service model to a scalable PaaS/SaaS model. We will license the App Builder to external agencies, independent implementation partners, and directly to enterprise SMEs, empowering them to build and manage their own applications. |
To become the default, AI-powered operational infrastructure for the world’s 500M+ SMEs.
Traditional ERP systems are rigid, expensive, and result in massive implementation failures, while fragmented SaaS solutions create isolated data silos. Furthermore, native GenAI is currently only accessible to enterprises with dedicated IT teams. Our long-term vision is to eliminate these barriers entirely.
By perfecting our proprietary App Builder, we will transition Creviz into a globally licensed platform. We will empower a worldwide network of independent business analysts and implementation partners to spin up hyper-customized, edge-case-proof operational workflows in a fraction of the time and cost.
Ultimately, because every custom application will run on our single, unified architecture, we will instantly equip every SME with our universally adaptable AI agents. We envision a future where these "digital employees" seamlessly manage cross-departmental operations, reporting, and workflows without any technical friction. We are democratizing the operational control and AI power of a Fortune 500 company for every SME, with zero technical overhead. |
Incorporated in Bengaluru Karnataka as Pvt Ltd |
NA |
Yes |
We are applying to Avinya because it is one of the few accelerators that deeply understands the intersection of AI, B2B SaaS, and operational workflows. We are not looking for generic startup advice; we are looking for highly specific operational mentorship to help us transition from our current "service-led wedge" (our BA team) to a highly scalable, AI-agent-led SaaS platform. We are specifically drawn to the deep, relevant domain expertise of the ACSEL board and operating team: Subrata Mandal (Core Operating Team): As the Co-Founder & COO of iFIX Tech Global, his experience in building scalable B2B enterprise software and intelligent automation platforms is exactly aligned with Creviz. His insights into workflow engines and deploying "autonomous digital care" are invaluable as we refine our own AI "digital employees" (our RPA, PM, and Reporting agents). Rajiv Kuchhal (Board of Directors): We are currently navigating the complexities of scaling a B2B business. Rajiv’s operational background in scaling Infosys BPO and OnMobile, combined with his deep-tech investing lens at Exfinity Venture Partners, makes him the ideal mentor to stress-test our hybrid pricing model and our transition from a service-heavy to a product-heavy GTM. Finally, because our target market relies heavily on trust, the backing of the IIT Alumni Centre Bengaluru (IITACB) and the Government of Karnataka provides the exact institutional credibility we need to close early SME partnerships and access early pilot deployments. |
Over the 4-month Avinya programme, our goal is to achieve true Product-Market Fit, transition our GTM strategy from "founder-led" to "scalable," and position ourselves for our next round of institutional funding. Specifically, we want to achieve the following milestones: 1. Stress-Test our AI Architecture: > We want to leverage the technical mentorship of the ACSEL ecosystem to refine our AI agent infrastructure. We aim to validate our unified data architecture with experts like Subrata Mandal and Dr. Praseeda K. to ensure our agents can seamlessly manage cross-departmental operations across diverse SME constraints without data silos. 2. Execute our Phase-2 GTM (Channel Partnerships): Currently, we acquire customers via unscalable 1-1 founder networks. During Avinya, we want to leverage the IITACB network and expert guidance to build our first institutional channel partnerships—specifically targeting mid-sized implementation firms and CFO forums who can act as distribution channels for our App Builder. 3. Refine the SaaS Economics: We want to work with strategists like Rajiv Kuchhal to lock in our hybrid revenue model. We aim to define the exact unit economics of our usage-based AI pricing (token usage for RPA/Reporting) versus our core SaaS seat licenses. 4. Demo-Day Readiness & Scaling: By the end of the programme, we aim to use the initial ₹10 lakh seed funding to finalize our SEO/GEO inbound engine, secure 5-10 highly sticky SME pilot deployments, and pitch successfully on Demo Day to raise the capital needed to globalize the Creviz platform. |
Yes we are open to hybrid and virtual participation. |
Yes |
Core engine |
Creviz follows a layered AI architecture built around structured metadata generation:
1. Input Layer
Conversational interface (chat-based)
Accepts natural language requirements (e.g., “Create a leave management system”)
2. Intent & Slot Extraction Layer
Multi-level intent classification:
Application → Modules → Forms → Rules → Actions
Slot-filling system to extract:
Entities (fields, workflows, roles)
Constraints (validations, conditions)
Maintains conversation state + intent stack
3. Context & Memory Layer
Session memory (short-term)
Metadata memory (long-term structured storage)
Versioned metadata with diff tracking
Optional vector layer (for semantic recall of past components)
4. Metadata Generation Engine (Core IP)
Converts structured intent → platform-specific JSON schema
Generates:
Forms
Workflows
Business rules
APIs
Supports iterative refinement via conversational edits
5. Execution Layer
Metadata is directly rendered into:
UI (React renderer)
Backend logic (Java services)
Database structure
6. Feedback Loop
User corrections → diff engine → improves prompts/templates
Continuous learning via usage patterns (not raw user data training) |
We are building proprietary structured enterprise intent → metadata datasets, which include:
Natural language → structured application definitions
Industry-specific workflows (HR, operations, CRM)
Field-level patterns (validations, dependencies, rules)
Over time, this becomes:
A high-quality supervised dataset for enterprise app generation
Difficult to replicate without:
Real user interactions
Iterative corrections
Domain-specific abstraction
Additionally:
Metadata usage patterns (how users modify generated apps)
Component-level performance insights |
Our defensibility comes from 3 layers:
1. Metadata Abstraction Layer (Core Moat)
We don’t generate code → we generate structured, executable metadata
This creates:
Consistency
Editability
Explainability
2. Learning Loop from Edits (High-value signal)
Most competitors stop at generation
We capture:
What users change
Why they change it
This builds a proprietary correction dataset
3. Deep Product Integration
AI is not a feature — it is embedded into the platform lifecycle
Tight coupling with:
Form engine
Rule engine
Workflow engine
4. Domain Focus (SME workflows)
Focus on companies (50–500 employees)
Repeated patterns → stronger AI accuracy over time |
We use a mix of offline + online evaluation metrics:
A. Accuracy Metrics
Intent classification accuracy
Slot extraction accuracy
Metadata completeness score (% of required fields filled)
B. Output Quality Metrics
% of generations requiring manual correction
Time to usable application (TTUA)
Number of iteration cycles per feature
C. Reliability Metrics
Hallucination rate (invalid fields / rules generated)
Schema validation success rate
API execution success rate
D. User-Centric Metrics
Task completion rate
Drop-off during app creation
User satisfaction (qualitative feedback) |
Data Handling
No training on customer data without explicit consent
Separation of:
User data
Metadata
System logs
Security
Encryption:
At rest (AES-256)
In transit (TLS 1.2+)
Role-based access control (RBAC)
Compliance (current + roadmap)
GDPR-aligned data handling practices
Data residency options (planned for enterprise customers)
Audit logs for all changes
AI-specific safeguards
Prompt isolation (no cross-tenant leakage)
PII masking (planned enhancement) |
Yes, but manageable:
Key Risks
Data privacy regulations (GDPR, India DPDP Act)
AI explainability requirements (future regulations)
Industry-specific compliance (HR/payroll data)
Mitigation
Metadata-driven approach improves:
Traceability
Explainability
No black-box execution — everything is structured |
Likely bottlenecks:
Latency in multi-step reasoning
Complex workflows → longer response times
Context management
Large applications → context window limits
Cost scaling (LLM usage)
High inference cost per generation
Consistency across sessions
Maintaining state across long conversations
Mitigation Strategy
Caching of metadata patterns
Incremental generation (modular approach)
Fine-tuned smaller models for specific tasks
Async generation for heavy workflows |
Founding team drives AI architecture and implementation
Expertise includes:
LLM prompting & orchestration
Backend systems (Java, distributed systems)
Frontend rendering (React-based dynamic UI)
Metadata-driven system design
Planned:
Dedicated AI/ML hires for:
Fine-tuning models
Evaluation pipelines
Data pipeline engineering |
Currently using:
Public LLMs (via APIs)
Internal synthetic datasets:
Generated application scenarios
Predefined workflows
Owned datasets (growing):
User interaction logs (anonymized)
Intent → metadata mappings
Correction/diff datasets
Licensing
No external proprietary datasets used currently
Plan to build fully owned dataset over time |
Short-term
Prompt engineering improvements
Better slot-filling and validation logic
Rule-based guardrails
Mid-term
Fine-tuned models on:
Metadata generation
Intent classification
Long-term
Reinforcement learning from user corrections (RLHF-like loop)
Industry-specific models (HR, CRM, Ops)
System-level improvements
Feedback loops from:
User edits
Failed generations
A/B testing of prompts and outputs |
This is a classic accelerator question, and it is usually a trap. If you say "Only India," they might think your Total Addressable Market (TAM) is too small. If you say "Global from day one," they might think you lack focus and are spreading your limited resources too thin.
The most successful answer to this question is the "Build in India, for the World" narrative. You want to show them that India is your ultimate stress-test and launchpad, but your architecture and ambition are inherently global.
Here is how you can structure this perfectly, leveraging your team's background and your current traction:
The Polished Version
Are you building for India, global markets, or both?
We are building a global platform, but India is our strategic launchpad and ultimate stress test. >
The operational challenges that SMEs face—lack of dedicated IT staff, siloed data, and rigid legacy systems—are universal. However, the Indian SME ecosystem is uniquely complex, highly cost-conscious, and operates under extreme constraints. Currently, we are deploying our early applications for local manufacturing and industrial clients. We firmly believe that if our App Builder and AI agents can seamlessly handle the edge cases and constraints of Indian SMEs, the platform will be robust enough to scale anywhere in the world.
Furthermore, we are globally minded by design. Our core team brings direct go-to-market and operational experience across the US, Brazil, and India. Once we solidify our playbook and perfect the platform's underlying AI architecture here, we will leverage our international MBA alumni networks and external implementation partners to expand our reach.
Our ultimate ambition is not restricted by geography; we are building the default AI operational infrastructure for the world's 500M+ SMEs. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/yJLe6pdb6r8e-Creviz-pitch-deck-acsel.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/yJLe6pdb6r8e-Creviz-pitch-deck-acsel.pdf |
https://drive.google.com/file/d/1LZW6el20guaVHEoJrR-iL4gJY2hLAqLg/view |
Yes, we are fundamentally mission-driven. Our core impact thesis centers on digital inclusion and economic empowerment for the underserved SME sector. >
SMEs are the backbone of the global and Indian economies, driving job creation and grassroots innovation. Yet, they are systematically left behind during major technological shifts. Traditional ERP software was built and priced for large corporations, resulting in massive implementation failures for small businesses. Today, GenAI is following the same trend—it is widening the gap by empowering enterprises that can afford dedicated IT teams, while SMEs are left struggling with manual processes and disconnected data silos.
We are not just building "cool tech"; we are solving a highly meaningful socio-economic problem. By giving SMEs access to an error-proof App Builder and affordable "digital employees" without requiring internal tech teams, we are actively leveling the playing field.
When an SME owner can finally gain holistic operational control and scale their business without being bottlenecked by technology, they hire more people, reduce waste, and grow sustainably. Our ultimate mission is to ensure that the 500M+ SMEs worldwide are not left behind in the AI revolution, but are instead empowered to lead it. |
NA |
NA |
checked |
| Apr 16, 2026 @ 12:54 PM |
Urooj Akhtar khan Inamdar |
uruinamdar@gmail.com |
|
|
09632342375 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 15, 2026 @ 11:03 PM |
Tushar Shinde, Nitish Mishra, Nitesh Tripathi |
tushar@vaaniresearch.com |
https://www.linkedin.com/in/dstartupkid/ |
https://vaaniresearch.com |
9404434273 |
Tushar Shinde • Researched on RL @ Indian Institute of Science (IISc), Bangalore • Product, Strategy, GTM, Positioning @ Vaani • Ran Product Portfolio @ OneCard (FPL Technologies - Unicorn) • Product Manager - NLU Enterprise Search @ Hypersonix Inc. (San Jose, USA) • Senior Data Scientist @ Publicis Groupe • Co-founded and scaled to previous ventures to $100K ARR |
Me (Tushar), Nitish, and Nitesh are building Vaani because we believe that Voice AI has the potential to transform how we interact with digital systems. While there is strong market demand for voice-based products and services, the state of the supporting infrastructure and Indic AI Models remains poor. We are determined to address this gap.
We’ve known each other for 8 years. Nitesh and Tushar are classmates at the Indian Institute of Science (IISc), and know Nitish from the IIT-Madras campus via a mutual friend. We'd earlier taken separate paths: Tushar into product and startup leadership (building and scaling 0–1 products), Nitesh into world-class deep tech research (with extensive work in Graph systems, NLP, and speech synthesis), and Nitish into building high-scale infrastructure (processing peta-bytes of Data at SG Brain Center). Now, those paths converge at Vaani. |
All |
We founders complement each other: Tushar leads GTM, product vision, fundraising and strategy. He’s customer-obsessed, moves with urgency, and has shipped products in the past that reached millions. Nitesh leads AI research & IP, building the entire voice AI stack from first principles and Nitish leads engineering. We've shipped our infra and application layer for Vaani within months, secured a first institutional funding round, brought business pilots to production with actual real-world results, and built a team of believers with limited resources. Currently, we are 9 people strong.
Vaani is more than a business for us. And we’re just getting started. |
Vaani AI Research Private Limited |
www.vaanivoice.ai |
Bengaluru, Karnataka |
Vaani builds the voice modelling and infrastructure layer that every Voice AI application (in linguistically complex markets) runs on. |
Emerging markets (like India, SEA, Gulf, LatAm) with high language-mix have high voice-based interactions every year, yet 60% are misrouted, mishandled, or dropped due to inadequate Voice AI infrastructure. Existing voice AI systems, predominantly built by Western companies, fail on critical fronts. Voice AI systems today fail in real-world environments like India, where conversations are multilingual, code-mixed, and often noisy. Existing solutions rely on fragmented pipelines (speech-to-text → LLM → text-to-speech), leading to high latency (800ms–2s), poor turn-taking, and unnatural conversations. Additionally, most global models are not trained for low-resource languages and dialects, resulting in low accuracy and unreliable deployments. This makes it difficult for enterprises to adopt voice automation at scale for critical workflows. |
Vaani provides a unified voice AI infrastructure and model layer optimized for real-world conversational environments. It reduces latency, improves multilingual understanding, and enables natural, context-aware voice interactions. The platform combines proprietary speech models, APIs, and deployment tools to help enterprises and developers build and scale voice systems reliably. By focusing on India and other emerging markets, Vaani enables voice interfaces that work across languages, accents, and conditions where existing systems fail. |
Vaani is building a full-stack voice AI infrastructure layer, combining proprietary speech models with orchestration and deployment capabilities. Its architecture is designed for low-resource, multilingual environments and focuses on real-world conversational challenges such as turn-taking, code-mixing, and noise handling. Unlike existing solutions that depend on multiple external APIs, Vaani optimizes the entire pipeline to reduce latency and improve reliability. The company also positions itself as a research-led organization, developing foundational speech models for emerging markets. |
Revenue |
Revenue |
Vaani serves enterprises, developers, and platforms that require scalable voice automation. Key segments include BFSI, fintech, healthcare, logistics, and customer support operations with high call volumes and multilingual users. The platform also targets SaaS companies, system integrators, and developers through APIs and self-serve tools. Additionally, research institutions and AI teams form a segment for model licensing and co-development. |
160000000000 |
50000000000 |
3500000000 |
Vaani follows a multi-stream revenue model. The primary stream is usage-based pricing for its platform and APIs, typically charged per minute of voice interaction. Enterprise deployments are monetized through pilot fees and annual contracts for production use. Additionally, the company generates revenue through licensing of its speech and text-to-speech models. This combination allows both scalable usage revenue and high-value enterprise contracts. |
Vaani operates in a competitive landscape that includes global voice AI infrastructure providers such as ElevenLabs, Deepgram, and Cartesia, as well as platform companies like Vapi, Retell AI, and Bolna. In India, competitors include Gnani.ai and Sarvam AI. However, most competitors either focus on individual components (e.g., TTS or APIs) or are optimized for Western markets. Vaani differentiates by building a full-stack, research-driven voice infrastructure tailored for multilingual, low-resource environments. |
We are currently acquiring customers through a combination of product-led growth and enterprise deployments. The self-serve platform and APIs will enable rapid adoption by developers and SMBs, while enterprise engagements will drive high-value deployments and reference customers. The company is also building partnerships with system integrators and distribution partners to expand reach. Over time, scaling will be supported by expanding language coverage, improving model capabilities, and entering international markets such as Southeast Asia and the Gulf. |
We are currently acquiring customers through a combination of product-led growth and enterprise deployments. The self-serve platform and APIs will enable rapid adoption by developers and SMBs, while enterprise engagements will drive high-value deployments and reference customers. The company is also building partnerships with system integrators and distribution partners to expand reach. Over time, scaling will be supported by expanding language coverage, improving model capabilities, and entering international markets such as Southeast Asia and the Gulf. |
Building the ElevenLabs of India - an infrastructure and models that lets every business, app, and service on the planet speak naturally to its users. |
April, 2025 |
23600000 |
Yes |
For the mentorship support, Enterprise doors and Grant |
To get more clarity on GTM, ICP and Scaling up from 1 to 100. |
Virtual preferred |
Yes |
Core engine |
Vaani is a foundational speech synthesis model built on a transformer-based TTS architecture. It combines a text encoder, acoustic model, and neural vocoder to generate human-like audio. The system supports streaming inference with sub-100ms TTFA (time-to-first-audio), enabling real-time voice AI applications. Models are fine-tuned per voice persona with distinct emotional and tonal profiles. |
We have curated a proprietary corpus of Indic language speech data covering Hindi, Tamil, Kannada, Telugu, Bengali, and other regional languages of 4.3 Lakh Hours of Speech, which wasich is roughly 50 Terabytes. |
Defensibility comes from three layers: (1) proprietary Indic speech data that is expensive and slow to recreate; (2) persona-level voice models with consistent emotional DNA, not just sound profiles; (3) low-latency streaming infrastructure optimised for voice AI agents. Together, these create deep switching costs for customers who build voice products on Vaani's voices and API. |
Key metrics: MOS (Mean Opinion Score) via human eval panels for naturalness; WER (Word Error Rate) for pronunciation accuracy; TTFA (time-to-first-audio) for latency; RTF (real-time factor) for throughput. We also track prosody consistency scores across utterances per persona. Automated regression tests run on each model update; human eval panels run fortnightly on sampled outputs. |
All audio data is encrypted at rest (AES-256) and in transit (TLS 1.3). We follow a data minimisation policy, user inputs are not stored beyond the inference session by default. Voice talent data is collected under signed consent and licensing agreements. Our infrastructure is SOC 2 aligned, and we maintain access control audit logs. For enterprise clients, on-premise deployment options are available. |
Key risks include: (1) emerging deepfake/synthetic voice regulations in India and the EU (AI Act) that may require disclosure when AI voices are used; (2) copyright ambiguity around voice cloning for specific public figures; (3) data localisation requirements under India's DPDP Act for storing voice data. We are proactively building consent workflows and usage disclosure APIs to stay ahead of these. |
Primary bottleneck is GPU inference throughput, current architecture is not yet fully batched for concurrent streaming sessions. At 10x load, latency spikes are expected without horizontal GPU scaling. Secondary risk is our model serving layer, which needs migration from single-region to multi-region. We have a 6-month roadmap covering batched inference, autoscaling, and CDN-based audio caching to address this. |
Yes. Our team includes 5 ML engineers with experience in speech synthesis, acoustic modelling, and low-latency inference. |
Opensources + Recorded + Bought from 30 different sources. |
Continuous improvement pipeline: (1) production feedback loop, flagged low-quality outputs are reviewed and added to fine-tuning sets; (2) A/B model evaluation with MOS panels before each release; (3) active learning to identify edge cases (rare phonemes, code-switching patterns); (4) quarterly voice persona refreshes with new data. We also track competitor benchmarks and run adversarial tests on prosody and naturalness. |
India and other emerging markets like Southeast Asia (SEA), LatAm, Gulf and others with significant language mix complexity. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/wmqAI9bvp4tq-Vaani-Research-IC-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/wmqAI9bvp4tq-Vaani-Research-IC-Deck.pdf |
https://www.youtube.com/watch?v=1LrZPHrPZpw |
We envision a world where every business, regardless of size or location, can speak to its customers as naturally and empathetically as humans do, in their language, with their cultural context, 24/7, and without the limitations. Over the next decade Vaani’s infrastructure is aiming to power a billion human-like AI conversations a day, becoming the invisible voice layer of global commerce, public services, and everyday life. |
OBC |
IISc |
checked |
| Apr 15, 2026 @ 10:44 PM |
Ritesh Kumar / Indrajeet Yadav |
ritesh.kumar@rupeeit.in |
https://www.linkedin.com/in/ritesh-kumar-9711a117a/ |
https://www.rupeeit.in/ |
9643097342 |
Founder & CEO / CTO |
Through common friend and we are working for last 8 months |
2 |
Our biggest advantage is speed, focus, and execution.
We’re a team of 3, and in under 3 months we built and launched RupeeIt on both the Play Store and App Store — entirely in-house.
While most AI products are horizontal, we went vertical. We built a domain-specific AI search engine for Indian stocks and mutual funds that can outperform general-purpose models on relevance, depth, and actionable insights.
Large teams move slower. We move fast, ship constantly, and stay obsessed with one problem: making financial information radically better for Indians.
Small team. Clear vision. Extreme velocity. |
Rupeeit |
https://www.rupeeit.in/ |
Bangalore |
India's AI-first money search engine - Ask anything about Stocks & Mutual funds and get insights in 2 mins that matter . |
India has 140M+ retail investors, but there is still no trusted, simple, and intelligent way for them to understand money. Most people make stock and mutual fund decisions through Google searches, WhatsApp forwards, YouTube influencers, fragmented apps, or expensive advisors. The result is noise, misinformation, and poor financial outcomes. Financial information in India is still a privilege. The top 1% get access to experts, research, and clarity. The other 99% get confusing content, biased advice, and generic tools. At the same time, existing platforms solve only one part of the problem — news, education, broking, or data. No platform combines trusted guidance, real-time intelligence, and personalized answers in one place. |
RupeeIt is building the financial intelligence layer for India.
Today, Indian consumers use a fragmented stack for money decisions: YouTube for learning, Google for research, news apps for updates, brokers for execution, and advisors for trust. None of these products deliver fast, personalized, and decision-ready answers in one place.
We replace that workflow with a single AI-native product. Users ask any question about stocks, mutual funds, or personal finance and receive actionable insights in under 2 minutes.
Product Architecture
1. Artha AI
Our proprietary finance AI engine built specifically for India. It ingests live market signals, multiple data sources, and news flows, then uses a multi-agent system to generate high-confidence outputs. We optimize for decisions, not chat responses.
2. Expert Content Layer
We pair AI with trust. Certified professionals create short-form educational content that helps users understand concepts, products, and market behavior in simple language.
3. Intelligence News Layer
We convert noisy financial news into concise, personalized intelligence. Instead of reading 100 headlines, users see only what matters to their portfolio and why. |
What makes RupeeIt unique is that we’re building the full financial intelligence stack for India, while everyone else is building point solutions. Today, users piece together 4–5 products: news apps for updates, YouTube for learning, brokers for execution, screeners for research, and advisors for trust. We combine all of that into one AI-native product. Our defensibility comes from six layers: 1. Proprietary Vertical AI Artha AI is purpose-built for Indian finance. It processes live market signals, multiple data sources, and domain-specific workflows for stocks and mutual funds. Generic LLMs are horizontal; we are vertically optimized. 2. Multi-Agent Architecture We use multiple specialized agents with consensus scoring instead of a single model response. That creates better reliability and decision-quality outputs in a category where trust matters. 3. Closed Data Flywheel Every query improves our understanding of investor intent, ranking quality, and personalization. More users create better answers, which drives retention and more users. 4. Trust Layer We pair AI with verified expert content from certified professionals. In finance, trust compounds slowly and becomes a moat. 5. Full-Stack Experience Competitors own one layer. We own search, education, intelligence, and eventually transactions. That creates stronger engagement and lower churn. 6. Execution Speed We’re a team of three and launched on both app stores in months. Small teams with sharp focus often out-execute larger incumbents. |
Users |
Users |
Our initial target audience is 22–40 year old salaried, mobile-first professionals in Tier 1 and Tier 2 India who actively invest or want to invest in stocks and mutual funds. They are digitally native users already using apps for investing, consuming finance content online, and searching regularly for better money decisions. Core segments: Young professionals doing SIPs and starting wealth creation Retail investors managing direct stock portfolios Aspiring first-time investors looking for trusted guidance Time-poor working professionals who want fast, actionable insights |
₹37,800 Cr |
₹10,080 Cr |
₹227 Cr in 3 years with 10 lac user base |
. Subscription Consumers pay for premium access to advanced AI insights, deeper stock and mutual fund research, personalized alerts, and exclusive expert content. 2. Ads High-intent financial audiences attract premium advertisers across brokers, insurers, lenders, wealth products, and financial brands. 3. Enterprise SaaS We license our AI intelligence layer, research tools, and white-labeled solutions to Mutual fund distributors , Fintech startups and Investment advisors 4. Sub-Broker / Distribution Commissions We earn commissions by helping users discover and transact in products such as mutual funds, broking accounts, insurance, and other financial products through partner networks. |
Multibagg.ai / Moneycontrol / Zerodha varasity |
Influencer lead marketing and Referral |
Go-To-Market Strategy (Year 1) 1. Creator-Led Acquisition Partner with finance creators and certified experts to drive trust-based user acquisition through short-form content on Instagram, YouTube, and social platforms. This is our primary top-of-funnel channel. 2. WhatsApp Community Growth Build high-engagement WhatsApp cohorts for daily market insights, retention, referrals, and habit formation. Finance products grow well through trusted peer networks. 3. Referral Flywheel + Performance Marketing Use referral incentives and targeted paid campaigns to scale efficiently, with a Year 1 CAC target of ₹150–₹200 per user. |
Our long-term vision is to build the default interface for money in India.
Today, consumers use separate products for search, news, learning, advice, and transactions. We think that stack collapses into one AI-native product.
RupeeIt starts by helping users make better decisions on stocks and mutual funds. Over time, we expand into every major financial workflow: saving, investing, borrowing, insurance, taxes, and retirement.
The product becomes a personal CFO for every Indian — always available, trusted, personalized, and action-oriented.
If Google organized information, we want to organize financial decisions for a billion people. |
Incorporated |
0 |
Yes |
Mentorship – Access to experienced operators and investors who can help us build and scale a product serving millions of users. Funding – Capital to accelerate product development, AI infrastructure, hiring, and growth execution. |
Build a scalable product roadmap — refine the product into a high-retention, scalable platform ready to serve millions of users. Reach 10K users — acquire and activate our first 10,000 users while validating strong engagement and repeat usage. |
YES |
Yes |
Core engine |
1. Data Intelligence Layer
Continuously ingests 70+ live signals and 30+ data sources including markets, mutual funds, news, and sentiment to create a real-time financial knowledge base.
2. Multi-Agent Reasoning Layer
A proprietary 6-agent system where specialized agents analyze risk, sentiment, valuation, relevance, and user intent, then generate consensus-scored outputs for higher accuracy.
3. Decision & Personalization Layer
Transforms 100+ raw inputs into 3–5 ranked actionable insights personalized to each user’s portfolio, goals, and risk profile. |
1. Unique User Intent Data
Every query on RupeeIt captures high-intent financial questions around stocks, mutual funds, portfolio actions, and market events. This creates proprietary demand-side data on what Indian investors actually want to know.
2. Structured Finance Knowledge Graph
We continuously organize fragmented Indian financial data across live market signals, funds, news, sentiment, and market behavior into a decision-ready intelligence layer. This is difficult to replicate in clean, usable form.
3. Persona Creation & Personalization Layer
We build dynamic user personas based on risk appetite, goals, investing experience, and behavior, allowing Artha to generate highly relevant answers instead of one generic response for all users. |
Our AI is defensible because it is built as a vertical financial intelligence system for India, not a generic LLM wrapper.
1. Proprietary Multi-Agent Architecture
Artha uses a 6-agent system where specialized models analyze signals such as risk, sentiment, valuation, news impact, and user intent, then generate consensus-scored outputs. This produces higher-quality financial answers than a single-model response.
2. Proprietary Data Layer
We combine 70+ live signals, 30+ financial data sources, market news, and user intent data into a structured intelligence layer focused on Indian stocks and mutual funds. Competitors can access models, but not our integrated data stack.
3. Personalization Engine
We build user personas based on goals, risk appetite, and investing behavior, allowing Artha to tailor outputs to each user. Personalized answers create better outcomes and stronger retention.
4. Continuous Learning Flywheel
Every query improves ranking, relevance, and decision quality. As usage scales, the product becomes smarter and harder to replicate.
5. Cost-Efficient Infrastructure
Most workflows are deterministic with selective model usage, allowing us to serve users at low cost while maintaining scale. |
1. Answer Quality
We track relevance, factual accuracy, freshness of market data, and whether the output clearly explains what happened, why it matters, and next actions. Internally, we benchmark ranked outputs (3–5 actions from 100+ inputs) against expected financial outcomes.
2. Consensus Reliability
Because Artha uses a multi-agent system, we measure inter-agent agreement scores, confidence thresholds, and conflict rates before responses are surfaced. Low-confidence outputs are constrained or reprocessed.
3. Personalization Performance
We track whether answers improve engagement across user personas based on risk profile, goals, and experience level. Metrics include repeat queries, session depth, and retention.
4. Speed & Cost Efficiency
For consumer finance, latency matters. We monitor time-to-answer, system uptime, and per-user serving cost (target low-cost deterministic workflows).
5. Expert Validation Loop
We regularly share Artha’s outputs with Investment Advisors, RIAs, and Mutual Fund Distributors to validate usefulness, accuracy, and real-world decision quality. In finance, expert trust is a key reliability signal. |
GUARD RAILS AND SOME IN HOUSE PARAMETERS |
NO |
1. Infrastructure: Data Freshness, Latency & Cost
At 10x scale, the biggest risk is maintaining real-time market data pipelines and fast response times while keeping multi-agent inference costs efficient.
2. Personalization & Quality Control
As users grow, handling millions of different personas, portfolios, and edge-case queries while maintaining answer accuracy and relevance becomes the core challenge. |
Yes |
YES , 1. NSE / BSE APIs & Market Data Sources
We use exchange-linked and market data sources for stock prices, indices, corporate actions, and related market information.
2. Mutual Fund & News APIs
We use mutual fund datasets, NAV feeds, and financial news APIs to track funds, market events, and breaking developments in real time.
3. Proprietary Internal Data Layer
We clean, normalize, and structure this raw data into our own intelligence layer used by Artha for search, ranking, and decision-ready insights. We also build first-party user intent and persona data from platform usage. |
1. User Query Flywheel
Every search helps us better understand investor intent, improve ranking quality, and identify gaps in answers. More usage creates better outputs over time.
2. Better Data + Expert Feedback
We continuously expand live market signals, news sources, and financial datasets, while validating outputs with Investment Advisors, RIAs, and MFDs to improve reliability.
3. Model Optimization Layer
We regularly fine-tune prompts, routing logic, agent orchestration, and personalization systems so Artha becomes faster, more accurate, and more relevant with scale. |
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/HnDDIHZ8AJvM-deck.rupeeit_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/HnDDIHZ8AJvM-deck.rupeeit_compressed.pdf |
https://www.youtube.com/watch?v=fe9BSIdpLSM.%20AND%20%20%20%20https://www.linkedin.com/posts/ritesh-kumar-9711a117a_rupeeit-fintech-ai-ugcPost-7449545731381567488-XrLQ?utm_source=share&utm_medium=member_desktop&rcm=ACoAACpi4iUBluAr4t-HjL0vl_53TDUtYBd4a9s |
Yes. RupeeIt is mission-driven at its core. We are building to democratize financial intelligence in India.
Today, quality money guidance is available to a small percentage of people through advisors, premium research, or privileged networks, while most Indians rely on noise from social media, WhatsApp forwards, and fragmented sources. This leads to poor financial decisions and missed wealth creation opportunities.
Our mission is to make trusted, personalized financial guidance accessible to every Indian at near-zero marginal cost through AI.
If successful, RupeeIt can help millions of people make better decisions on saving, investing, and growing wealth — regardless of income level or background.
We believe financial confidence should not be a privilege. It should be infrastructure. |
NO |
Chittaranjan Savadi |
checked |
| Apr 15, 2026 @ 5:32 PM |
Subhamoy Maiti, Sanusha K B, Dr. Avinash Gowda, and Prasad S Narasimha |
subhamoy_maiti@centurionrobotics.tech |
https://www.linkedin.com/in/subhamoy-maiti-9101381a1/ |
https://centurionrobotics.tech/ |
9483082497 |
|
|
All |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 15, 2026 @ 4:42 PM |
Karan Birpali , Akash Dangee |
karan.birpali@prakarann.com |
https://www.linkedin.com/in/karan-birpali/ |
https://www.pukaarai.com/ |
+918427698641 |
Karan - CEO , backround - ( lead business , anlaytics and growth , for 5 years , scaled a business from 0 to $0.5M ARR ) , Akash Dangee ( CPO ) ( lead b2b team in startups from 0-10 teams to 10-100 teams , deep expertised in product and ops ) |
We have known each other for more than 15 years , prepped for IIT together , He went to IIT Roorkee , i went to IIT Delhi ,
Been in constant touch , started together , been working on this for more than 8 months |
All |
Team which is Very lean and having complimentory skills .
have worked together in the past , understand how to get constructive outcomes even from an arguement
Known each other for more than 15 years
Karan (CEO) , Akash (CPO) , Mishal ( CTO )
very distince roles and verftical expertised in each other's roles and trust |
Pukaar AI |
https://www.pukaarai.com/ |
Gurgaon |
Building the next generation digital coparent evey baby needs |
Parental Anxiety |
We are building an AI-powered digital co-parent that continuously interprets an infant’s signals—such as cry, breathing, and behavior—to generate longitudinal health insights. By creating personalized baselines for each baby, we enable early detection of deviations, provide real-time guidance to parents, and support timely interventions , shifting infant care from reactive consultations to proactive, predictive care. |
We are building a proprietary longitudinal dataset of infant behavioral signals—something no existing system captures at scale along with our foundational cry analysis model where we can decode infant cries |
Users |
Users, Pilots, Signups |
Tech savvy , First time parents living in Tier 1 cities having nuclear families |
3000 cr |
300 cr |
40 cr |
subscription + SaaS |
Babynama |
B2C acquisition , google and meta ads |
Doctor led GTM , B2B2C |
To build a doctor-led digital co-parent that becomes the default interface for infant care.
We aim to capture the longitudinal behavioral and clinical record of every baby’s first 1,000 days ,passively and through trust.
This data layer will power the entire infant ecosystem, becoming foundational infrastructure for healthcare, nutrition, insurance, and beyond |
Private limited |
1.31cr |
Yes |
To deeply integrate with the medical ecosystem early. We want to leverage Avinya’s clinical expertise and network to validate our models, build trust with doctors, and design the right GTM through healthcare channels. |
better GTM strategy |
Both |
Yes |
Core engine |
1. Cry Intelligence (Foundation Model)
A domain-specific audio foundation model that converts infant cries into structured signals using spectrogram-based features and temporal deep learning. It classifies needs and detects anomalies based on each baby’s baseline.
2. Qara (Fine-tuned LLM Layer)
A fine-tuned LLM that combines cry outputs, user inputs, and longitudinal history with medical knowledge (via RAG) to deliver personalized insights, risk triage, and real-time guidance. |
What proprietary data advantage (if any) do you have?
We are building a proprietary dataset of labeled infant cry signals, which forms the foundation for our audio intelligence models.
On top of this, Qara creates a longitudinal memory layer for each baby capturing behavioral patterns and enabling highly personalized insights over time.
Crucially, we close the loop with doctor-verified outcomes, generating high-quality “gold” data that continuously improves both accuracy and trust. |
Our moat is proprietary data + personalization + clinical feedback loops.
We combine labeled cry data, individual baby baselines, and doctor-verified outcomes to continuously improve accuracy.
This creates a compounding system that becomes more personalized, trusted, and hard to replicate over time. |
We evaluate across three layers:
• Model accuracy: Precision, recall, F1-score on labeled cry classification and anomaly detection tasks
• Personalization quality: Deviation detection accuracy vs individual baselines and reduction in false alerts over time
Also still working on this .. |
We follow a privacy-first, security-by-design approach.
All user data is encrypted in transit and at rest, with strict access controls and anonymization of sensitive data.
We ensure compliance with relevant regulations (e.g., HIPAA/GDPR-equivalent standards where applicable), and give users full control over their data, including consent and deletion. |
Yes , once our data includes doctor-labeled outputs, it can be classified as medical-grade data, increasing regulatory scrutiny.
This may require compliance with medical device and clinical decision-support regulations, along with stricter standards for validation, auditability, and usage. |
What will break in your AI system if you scale 10x?
The main stress points are data quality, personalization, and latency.
• Data quality: Noisy, unlabeled, or diverse real-world inputs can degrade model performance at scale
• Personalization: Maintaining accurate per-child baselines becomes complex with rapid user growth
• Latency & infra: Real-time processing of high-frequency audio and multimodal data can strain pipelines
We are addressing this through strong data pipelines, active learning loops. |
Yes—we have an in-house team with strong expertise across AI/ML, data engineering, and applied healthcare AI.
Our experience includes building real-time data pipelines, multimodal ML models, and analytics systems at scale, along with hands-on work in audio processing and predictive modeling.
Mishal - CTO ( built the backbone of UPI in razorpay
Ravinder - AI engineer , ex-founder building an AI native converasational agent
Saish - Fresh graduate working with deep learning |
We use a combination of licensed medical knowledge sources (including IAP guidelines and standard pediatric medical literature) and proprietary data we are generating in-house. |
We use a closed-loop learning system combining real-world data, personalization, and clinical feedback.
New data from cry signals and user interactions is continuously added, with active learning and re-labeling to improve model accuracy. |
We are starting with India as our primary market, given the large population, high unmet need, and strong early adoption potential.
Our immediate focus is to build, validate, and scale within India. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/JKMUxYI2Wb2P-Pukaar-AI-blurb.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/JKMUxYI2Wb2P-Pukaar-AI-blurb.pdf |
https://drive.google.com/file/d/1kVP1sYWJ7-Mvj7p6KbKydsTiGkIwawmg/view?usp=drive_link |
Yes we are deeply mission-driven.
Our goal is to reduce parental anxiety and enable early, accessible healthcare during the most critical first 1,000 days of a child’s life.
By turning everyday infant signals into actionable health insights, we aim to shift care from reactive to proactive, improving outcomes for both children and parents. |
No , we do not identify as part of an underrepresented group. |
Amit Mishra Founder of Dazeoff |
checked |
| Apr 15, 2026 @ 5:02 AM |
Nitin Jagadish , Santhosh kumar and arpit saxena |
nitin@augustinnovate.com |
https://www.linkedin.com/in/nitinjagadishprofile?utm_source=share_via&utm_content=profile&utm_medium=member_ios |
http://www.procesync.ai |
9845375943 |
GTM consultant and management leader |
We met at Zendesk between 2022-2024. I was leading as Senior Director, Santhosh was a Senior Manager, and Arpit was a Senior Solutions Consultant. Over approximately 2 years, we worked closely together on GTM strategy, customer success, and business development for Zendesk's India operations.
During that time, we realized we shared a common vision for helping businesses with their digital transformation journey. That alignment led us to co-found August Innovate and Procesync together starting in mid-2024.
So in total, we've been working together for about 2+ years** (Zendesk collaboration + ~1 year 11 months as co-founders building our own ventures). We're still very much in the early, exciting phase of our entrepreneurial journey, with all three of us based in Bengaluru, driving innovation in business solutions and digital transformation. |
All |
All 3 of us Compliment each other . Nitin brings
in strategic leadership and GTM. Santhosh Product Design and Product Build capabilities and Arpit brings in technical and operational Execution.
Here are the details :
1. Nitin Jagadish* — Strategic Leadership & Go-to-Market
- 10+ years at IBM (sales leadership, regional management)
- 3+ years at Salesforce (Country Head for Platform Cloud)
- 2+ years at Zendesk (Senior Director, India GTM)
- **Brings:** Vision, strategy, business scaling, and executive-level decision making
2. **Santhosh Kumar** — Solutions & Operations Management
- 2+ years at Zendesk (Senior Manager)
- 1+ year at Yellow.ai (Solutions Manager)
- 2+ years at Zendesk earlier (Senior Manager)
- **Brings:** Solution architecture, customer success, operational excellence
3. **Arpit Saxena** — Technical Solutions & Customer Implementation
- 2+ years at Zendesk (Senior Solutions Consultant)
- 1+ year at ServiceNow (Senior Solutions Consultant)
- 2+ years at Oracle (Solutions Consultant)
- **Brings:** Technical depth, customer implementation, solution customization
---
## **Why This Matters:**
End-to-end capability** From C-level strategy (Nitin) → solution design (Santhosh) → technical execution (Arpit)
Proven track record— All three have successfully scaled business at world-class SaaS companies (Zendesk, Salesforce, ServiceNow, Oracle)
Shared mindset— Built together at Zendesk, so they already understand how to work as a high-performing team
Market credibility — Combined enterprise sales, implementation, and leadership experience that clients trust
We are not just co-founders—we are a complete business team with the strategy, solutions, and execution capabilities to build and scale an enterprise software company. |
Augustinnovate.com |
www.procesync.ai |
bangalore |
Procesync is an AI operating system for cash flow — automating AP and AR from invoice to reconciliation, end-to-end |
The average mid-market company takes 15 days to close their books. 80% of that time is spent on repetitive AP and AR tasks that should take zero human effort. That's half a month of delayed decisions, trapped working capital, and overworked finance teams — every single month.hflow We unify both into one intelligent workflow that closes itself |
Procesync is the AI operating system for enterprise cash flow — unifying AP and AR into one intelligent, self-orchestrating workflow.
Today's finance teams are trapped in chaos. Invoices arrive across email, portals, and drives. Approvals stall in Slack threads. Payments get processed in one system, receivables chased in another, and reconciliation happens in spreadsheets at month-end. The result: 15-day close cycles, frozen working capital, and finance teams buried in manual work instead of strategic decisions.
We're eliminating this entirely.
Procesync uses AI to extract invoices with 99% accuracy, auto-route approvals based on policy, match POs and GRNs instantly, trigger payments, chase overdue receivables, and reconcile ledgers — all from a single control plane. No re-keying. No tool-hopping. No month-end fire drills.
The outcome: enterprises close books in 3 days instead of 15, unlock real-time cash visibility, and free finance teams to focus on growth — not data entry.
One workflow. Full cash cycle. Zero friction. |
We're not just automating AP and AR — we're building the financial intelligence model that learns from every transaction, predicts cash flow, and gets smarter with every customer we add |
Revenue |
Revenue |
Enterprise and MSME |
8000 cr |
2000 cr |
20-40 cr |
per page or revenue share |
High radius , growfin |
Targeted outboud, Referrals,service buit pipeline |
Early Founder + services Growth Outbound + SDR Scale Partner-led + inbound |
We are building a layered platform—starting with AI-powered AP/AR automation, adding fraud detection and data visibility, and scaling into intelligent insights across vendors, purchasing, and receivables to drive better cash flow and financial decisions.
2–3 Year Product & Vision Roadmap
Year 1: Strengthen Core & Monetization
Scale AI-powered AP/AR automation across key enterprise workflows
Improve accuracy, speed, and integration depth (ERP, accounting systems)
Expand service-to-product conversion and increase recurring revenue share
Build foundational data layer for financial document standardization
Year 2: Intelligence Layer Expansion
Launch fraud detection engine for document and transaction anomalies
Introduce real-time financial dashboards for visibility into cash flow, payables, and receivables
Roll out vendor intelligence (ranking, risk scoring, pricing insights)
Enable receivables intelligence (payment behavior, delay prediction, term optimization)
Year 3: Decisioning & Platform Play
Evolve into decision intelligence platform for finance teams
Automate recommendations for purchasing, vendor selection, and payment strategies
Introduce predictive cash flow optimization and capital allocation insights
Deepen ecosystem integrations (FPOs, ERPs, financial systems) to become embedded infrastructure |
Private ltd |
na |
Yes |
I’m applying to Avinya to sharpen my thinking as I scale ProcessPay, where we’re solving real cash flow inefficiencies in businesses. At this stage, I value structured feedback, strong peers, and diverse perspectives. I bring execution and early traction, and I’m looking to accelerate both my growth and the company’s long-term impact. |
I want to gain sharper strategic clarity, strengthen my decision-making as a founder, and build a scalable path from early traction to a category-defining company. I aim to learn from peers and mentors, refine our go-to-market and product thinking, and accelerate ProcessPay’s journey toward becoming a ₹100 Cr business. |
Yes |
Yes |
Core engine |
Procesync’s AI architecture is designed as a layered system—starting with document ingestion and preprocessing, followed by multi-model extraction using OCR and deep learning models, and a validation layer that ensures high accuracy through rule-based and AI-driven checks. This structured data then feeds into an intelligence layer that powers fraud detection, real-time financial visibility, and decision-support insights across AP and AR workflows. |
Our advantage lies in the quality, context, and continuous flow of real-world financial data—positioning us to build a strong, defensible data moat as we scale volume, customers, and use cases.
-Processing thousands of financial documents per month across AP/AR workflows
-Achieving 95%+ extraction accuracy, improving continuously with usage
-Delivering 70–90% reduction in manual processing effort
-Data sourced from live enterprise workflows, not static or synthetic datasets
-Service-to-product model accelerates high-quality data acquisition and validation
-Continuous learning loop improves models and insights over time
-Enables context-rich intelligence (vendor ranking, payment behaviour, fraud signals)
-Building a compounding data moat as volume and customers scale |
-Proprietary data loop → trained on real enterprise financial workflows
-Domain-specific models → optimized for AP/AR, not generic OCR
-Continuous learning → improves accuracy and insights with usage
-Workflow embedding → deeply integrated into finance operations
-Switching cost → high due to data, integrations, and process dependency
-Intelligence layer → insights (vendor, payments, fraud) beyond extraction |
We ensure reliability through a hybrid approach—combining AI predictions with rule-based validations and human-in-the-loop review for edge cases, maintaining high accuracy and trust in critical financial workflows.
-Field-level accuracy : precision, recall, F1-score for key data points (invoice number, amount, dates)
-Document-level accuracy: end-to-end correctness of extracted data
-Confidence scoring : triggers human review for low-confidence outputs
-Exception rate: % of documents requiring manual intervention
-Processing time:turnaround time per document
-Human-in-the-loop validation :continuous feedback for model improvement
-Drift monitoring : performance tracking across new document formats/vendors
-Business impact metrics:
70–90% reduction in manual effort
Faster processing cycles
Improved data consistency |
Security and compliance are core to our platform. We are ISO 27001 certified and have successfully completed a recent VAPT assessment, validating our security posture. Our architecture includes end-to-end encryption, role-based access controls, tenant-level data isolation, and full auditability—ensuring enterprise-grade data protection and compliance readiness. |
While data privacy regulations are evolving, we are well-positioned to manage this through our security-first design, ISO 27001 certification, and recent VAPT validation. Additionally, our hybrid deployment model allows us to deploy within a customer’s private cloud, ensuring compliance with strict data residency requirements while preserving our core IP—reducing regulatory friction and accelerating enterprise adoption. |
At 10x scale, we expect pressure on three areas: model generalization across diverse document formats, increased exception handling for edge cases, and infrastructure scalability for real-time processing. We are mitigating this through a modular AI pipeline, active human-in-the-loop validation, and scalable cloud architecture to ensure performance and reliability are maintained as volume grows. |
Yes, we have an in-house team focused on applied machine learning, data engineering, and financial modeling. The team specializes in building behavioral models, time-series forecasting, and risk scoring systems, combined with strong domain understanding of finance workflows, enabling us to translate real-world complexity into practical AI s |
Yes, we have an in-house AI/ML team focused on developing domain-specific models for financial document extraction, validation, and intelligence. Our expertise spans OCR, deep learning, and applied AI in financial workflows, supported by a strong understanding of enterprise systems and data pipelines. This allows us to rapidly iterate, improve model performance through real-world feedback, and build a tightly integrated intelligence layer beyond basic automation. |
Our improvement strategy is built on a continuous learning loop—capturing data from live enterprise workflows, validating outputs through human-in-the-loop processes, and feeding corrections back into model retraining. This is complemented by performance monitoring, drift detection, and domain-specific tuning, enabling consistent improvements in accuracy, reliability, and insights over time. |
Our primary focus is India as a launch and scale market, given the high volume of unstructured financial workflows and strong demand for automation. At the same time, our product is designed for global applicability, allowing us to expand into international markets with similar use cases once we establish strong product-market fit and operational scale. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/r9kiswZR2wtP-ProceSync-InvestorDeck-v3.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/r9kiswZR2wtP-ProceSync-InvestorDeck-v3.pptx |
https://drive.google.com/file/d/1-Ou1mF4i7yIydAdhRCfJFMwMJdfWyn0N/view?usp=drivesdk |
We are a mission-driven startup focused on transforming how enterprises manage and act on financial data. Today, 60–80% of financial workflows are still manual and unstructured, leading to inefficiencies, errors, and delayed decision-making. We are solving this by building an AI-powered platform that reduces manual effort by 70–90%, improves data accuracy to 95%+, and enables real-time financial visibility.
Our approach is deeply market-driven—we work closely with customers through a hybrid service and product model, ensuring we solve real problems and continuously adapt to evolving enterprise needs. Beyond automation, our mission is to enable intelligent decision-making through insights such as vendor performance, payment behavior, and fraud detection—helping businesses improve cash flow and operational efficiency.
Ultimately, we are building a intelligence layer that shifts finance teams from reactive processing to proactive, data-driven decision-making at scale. |
NA |
AIPOD |
checked |
| Apr 14, 2026 @ 11:19 PM |
Ashutosh Prakash Singh, Neeraj Gupta, Pankaj Gupta |
ashutosh@revrag.ai |
https://www.linkedin.com/in/ashutoshpsingh/ |
https://revrag.ai/ |
+91-9999071459 |
Ashutosh P Singh, Co-Founder & CEO, 9+ years in SaaS GTM & Product Management | Neeraj Gupta, Co-Founder & COO, 8+ years in Product Management | Pankaj Gupta, Co-Founder & CTO, 9+ years in Engineering, AI & Data Science |
We’ve known each other since childhood, as we all studied at the same school in Lucknow. Over the years, that relationship evolved from friendship into collaboration, and we’ve been working together professionally for over 6-7 years now while building Revrag.ai and previously Blance and Hunkaar.
In total, we’ve known each other for over a decade, which brings a strong foundation of trust, understanding, and alignment in how we operate as a team. |
All |
Our biggest strength as a team is deep trust combined with complementary execution strengths. Having known each other since childhood, we operate with high ownership, low ego, and complete transparency, which allows us to make fast, aligned decisions even under pressure.
At the same time, we bring distinct strengths across product, GTM, and operations, enabling us to move quickly without overlap and stay focused on outcomes. |
Revrag AI |
https://revrag.ai |
Bengaluru, Karnataka |
We build in-app AI agents that automate onboarding, sales, and customer journeys for BFSI companies to reduce drop-offs and drive revenue. |
We are solving high drop-offs and lost revenue in onboarding and sales journeys caused by friction, lack of real-time guidance, and inefficient follow-ups. |
We deploy in-app AI agents that actively guide users through onboarding and sales journeys, reducing friction and recovering drop-offs in real time.
Our agents are embedded directly within product workflows and interact with users across voice, chat, and in-app interfaces to assist with tasks like form completion, document uploads, and query resolution. They proactively engage users at critical points in the funnel, ensuring higher completion rates and improved conversion.
Unlike traditional automation tools that focus on outreach, our solution is designed to drive outcomes, helping businesses convert more users, improve onboarding efficiency, and increase revenue without adding headcount. |
Our uniqueness lies in building outcome-driven AI agents that are deeply embedded into product workflows, not just surface-level automation tools. Our agents actively guide users, intervene at drop-off points, and drive measurable outcomes like higher conversions and revenue. Our defensibility comes from deep integrations, continuous learning from real user behaviour, and strong alignment with regulated environments like BFSI. |
Revenue |
Revenue |
Our target customers are BFSI and fintech companies that rely on digital onboarding and high-volume sales workflows, where drop-offs and conversion inefficiencies directly impact revenue. |
$64B – Global BFSI AI market by 2030 |
$4.5B – India BFSI AI market by 2028 |
$1.8B – India Voice AI segment within BFSI |
We follow a usage-based AI SaaS model, charging per minute of AI interaction, along with recurring monthly maintenance fees and one-time implementation fees for deployment and integrations. |
Our main competitors include SquadStack, Gnani AI, Yellow.ai, Haptik, Arrowhead, and Greylabs, along with other players in voice AI, conversational AI, and customer engagement platforms. While these solutions address parts of the problem, most are either voice-first or chat-first tools and are not deeply embedded into BFSI apps and workflows to drive end-to-end onboarding and sales outcomes, which is our core focus. |
We acquire customers primarily through direct outreach and inbound leads driven by our build-in-public marketing approach, along with referrals and ecosystem partnerships that generate warm introductions within the BFSI ecosystem. |
Our go-to-market strategy is centered around enterprise sales for BFSI companies, using a POC-led approach to demonstrate ROI and convert customers into long-term deployments, with a roadmap towards product-led growth for mid-market scale. |
Our long-term vision is to become the operating system for AI agents in BFSI, powering every customer interaction across onboarding, sales, and servicing.
We aim to build a unified platform where AI agents across voice, in-app, and messaging channels continuously learn from every interaction, reduce friction, and drive revenue outcomes at scale, ultimately becoming the default revenue automation layer for global BFSI enterprises. |
Incorporated as a Private Limited Company in India (Revrag AI Technology Private Limited) |
$400K (pre-seed round) |
Yes |
We are applying to Avinya for its operator-led approach and focus on building companies through hands-on support, not just capital. At our current stage, we are looking for partners who can help us refine our go-to-market strategy, scale enterprise deployments, and build a strong foundation for long-term growth. |
During the programme, we aim to refine our go-to-market strategy, strengthen our enterprise sales motion, and build a repeatable and scalable deployment model for BFSI customers. We also aim to improve unit economics visibility and validate additional high-impact use cases to accelerate our path to product-market fit. |
Yes, we are open to both hybrid and virtual participation, and can adapt based on the programme’s requirements. |
Yes |
Core engine |
RevRag's AI architecture is built around one core principle: the intelligence sits in the orchestration, not the model.
We use best-in-class third-party providers at the model layer including OpenAI and Google Gemini for reasoning, Azure, ElevenLabs, and Sarvam for voice output, and Deepgram and Sarvam for speech recognition. Every capability has 2 to 3 alternate providers configured for automatic failover.
What makes us different is what sits on top of that. Three proprietary layers form our core IP:
Our Context Engineering Layer dynamically assembles the most relevant inputs from knowledge bases, CRM data, conversation history, and user memory before anything reaches the LLM, ensuring every interaction is precise and contextualised.
Our Memory System maintains session, episodic, and long-term memory per user, making every interaction increasingly personalised over time.
Our Workflow Framework powers complex multi-step agent behaviours with configurable rule-based, AI-driven, and human-in-the-loop decision points.
We deploy two agent types on this stack. Our Voice AI Calling Agent handles outbound conversations in real time targeting sub-1,000ms latency. Our In-App Embedded Agent is delivered via native SDKs across Flutter, React Native, React, Angular, Android, and iOS handling form filling, screen navigation, generative UI, and backend actions with PII masking built in.
All AI inference happens within India, all LLM calls are routed through our backend, and our infrastructure runs on AWS Mumbai in a single-tenant architecture with database-level isolation per client. |
RevRag's proprietary data asset is a compounding behavioural intelligence corpus built exclusively from real financial app interactions across India's most regulated BFSI workflows.
It has three distinct components.
First, a BFSI-Specific Interaction Graph. Every deployment generates high-signal, outcome-labeled behavioural data covering where users drop off, what questions they ask, what errors they encounter, and what agent responses led to successful conversions across lending, insurance, wealth management, credit cards, and banking. A competitor cannot buy or synthesize this data. It can only be accumulated by operating at production scale inside these apps.
Second, Per-User Longitudinal Memory. We maintain session, episodic, and long-term memory for every end user. The longer a client runs RevRag, the smarter the agent gets for their specific user base. Replacing RevRag means restarting this memory from zero, creating structural switching costs.
Third, RevRag Central Intelligence. Every deployment feeds a platform-wide learning loop that improves agent behaviour across all clients. Each new client makes the platform smarter for all subsequent ones.
This data asset is hard to replicate for five reasons. Access to regulated BFSI apps requires 4 to 6 months of procurement and compliance clearance that we have already completed. Our ISO 27001, SOC 2, and SAR certifications take 12 to 18 months to obtain and are prerequisites for BFSI enterprise access. Our behavioural patterns are specific to Indian BFSI journeys that generic AI data cannot substitute. Our memory layer cannot be ported or recreated by a replacement. And our intelligence loop widens every month, compounding our performance advantage over any new entrant. |
Our defensibility is not a single moat. It is five compounding barriers that widen with every month we operate.
The orchestration layer is the product. Our core IP sits in the Context Engineering Layer, Memory System, and Workflow Framework that sit on top of third-party models. Even if a competitor uses the same underlying models, they cannot replicate what sits between those models and the end user.
Our data asset cannot be bought or synthesized. Every production deployment generates high-signal, outcome-labeled behavioural data from real BFSI workflows. This data can only be accumulated by operating at production scale inside regulated BFSI apps, which requires months of procurement, security review, and compliance clearance that we have already completed with our existing clients.
Memory lock-in creates structural switching costs. Our Memory System builds a user-level intelligence layer over time. Replacing RevRag means restarting this memory from zero, which is an immediate and measurable regression in product quality that no client wants to voluntarily accept.
The cross-deployment learning flywheel compounds over time. Every client we add makes the platform smarter for all subsequent ones. A competitor launching today would need years of equivalent production history to catch up.
Our compliance posture takes 12 to 18 months to build. We hold ISO 27001, SOC 2 Type I and II, and have completed a SAR Data Localization Audit. These are prerequisites for BFSI enterprise access. Any new entrant faces this compliance runway before they can compete for the same clients.
In short, our defensibility compounds with time. The longer we operate, the harder we are to displace and the harder we are to replicate from the outside. |
We evaluate model performance and reliability through two deeply integrated systems that operate at every stage of the agent lifecycle.
Before any agent goes live, it goes through our pre-call simulation system built directly into the Agent Builder platform. Automated test scenarios are run and benchmarked against defined quality criteria. No agent reaches a live client without passing this quality gate. This system is fully reproducible, meaning the same simulation can be run repeatedly to validate consistency across model updates, prompt changes, and configuration adjustments.
Once an agent is live, our post-call evaluation system automatically analyses every conversation. The key metrics we track are call success rate, re-engagement outcomes, task resolution rate, user satisfaction, and interaction quality. These give us and our clients a continuous real-time view of how the agent is performing against the outcomes it was deployed to deliver.
The same evaluation infrastructure powers our A/B testing framework, allowing us to compare model configurations, prompt variants, and agent behaviours both before deployment and in production. Every change is measured against a baseline before it is rolled out at scale.
On the reliability side, every capability including LLM, TTS, and STT has 2 to 3 alternate providers configured with automatic failover. Hallucination detection and output validation are built into the pipeline to catch quality issues before they reach the end user.
Our feedback loop captures downstream outcomes including call success rates, task completion, and user satisfaction and feeds them back into prompt refinement and context engineering improvements, creating a continuous improvement cycle for each client's agents.
Our current P90 end-to-end conversation latency sits at 1,000 to 2,000ms with a roadmap target of sub-1,000ms through edge computing and open-source model integration. |
Data privacy, compliance, and security are architectural decisions we made from day one, specifically because our clients are regulated BFSI enterprises where anything less is a disqualifier.
We operate a single-tenant architecture with database-level isolation. Every client has separate data stores with no shared databases and no cross-tenant data pathways. All data is encrypted at rest using AES-256 via AWS KMS and in transit using TLS 1.2 and above. Access is secured via JWT-based authentication with per-session tokens.
All data is hosted on AWS Mumbai. All AI inference happens within India. All LLM calls are routed through our backend with no direct client-to-model exposure. This meets the data localisation requirements of Indian BFSI clients including RBI and SEBI regulated entities.
We operate as a data processor and practice data minimisation. PII masking is built into our in-app agent SDK at the collection layer. Client data is used exclusively to improve that client's own agents. We do not use customer data for cross-client purposes or model training.
Our compliance posture is one of the strongest in our category. We hold ISO/IEC 27001:2022 certification for both our India and US entities, SOC 2 Type I and II, and have completed a SAR Data Localization Audit. We conduct VAPT semi-annually through a CERT-In empanelled auditor. We are aligned with ISO 42001 for AI governance covering bias detection, fairness evaluation, and explainability. We maintain formal BCP and DR plans with documented drills.
All third-party vendors are configured for zero or minimal data retention. We have a formal organisation-wide Data Retention Policy with log retention of 180 days and above in compliance with CERT-In requirements.
Our track record is clean with no security incidents reported with respect to client data to date. |
We operate in one of the most regulated sectors in India and we think about regulatory risk seriously. Our compliance posture is deliberately built to stay ahead of the regulatory curve rather than react to it.
The primary regulatory areas that impact our product are as follows.
TRAI and Telecom Regulations govern our Voice AI Calling Agent. We ensure full compliance with TCCCPR 2018 and its amendments, the DLT framework, and the Digital Consent Acquisition system. We exclusively use 140 series numbers for promotional communications and 160 series numbers for transactional communications as mandated.
RBI Regulations on AI in financial services could impact how our agents operate in customer facing workflows. We mitigate this by building configurable human-in-the-loop escalation points into every agent workflow and by maintaining close alignment with client compliance teams. Our contracts clearly define the boundaries of AI-driven versus human-driven decision making.
The Digital Personal Data Protection Act 2023 is the most significant data regulation affecting our platform. We operate as a data processor with the client as data fiduciary. All contracts include Data Processing Agreements defining permissible data usage, consent obligations, and data deletion requirements. PII masking is built into our SDK at the collection layer as a technical safeguard.
SEBI Regulations apply to wealth and capital markets clients like Motilal Oswal, imposing requirements on data record retention of up to 8 years, audit access, and disclosures around automated customer interactions. Our single-tenant architecture and formal data retention policy are designed to meet these requirements.
AI Governance Regulations are evolving rapidly globally and domestically. Our alignment with ISO 42001 positions us well for any future mandatory AI governance requirements that Indian regulators may introduce.
In short, regulatory risk is real in our space but it is also a competitive advantage for us. Every new regulatory requirement raises the bar for entry and makes our already compliant platform more valuable to clients who cannot afford to work with vendors who are not ready. |
Scaling 10x will stress specific parts of our system and we know exactly where those are.
Voice infrastructure and real-time latency will face the most immediate pressure. We currently run LiveKit self-hosted for voice targeting a P90 latency of 1,000 to 2,000ms. At 10x call volume, maintaining this latency target will require significant horizontal scaling, more aggressive auto-scaling on EKS, and edge computing to bring inference closer to the end user. Our Q3 2026 roadmap already includes an edge computing POC for TTS, STT, and LLM inference specifically to address this.
LLM inference costs and throughput will become a significant pressure point on unit economics. At 10x volume, the cost per conversation needs to come down materially. We are addressing this through open-source model integration alongside commercial APIs and through our context engineering layer which minimises token usage by assembling only the most relevant inputs for each LLM call.
Our single region infrastructure on AWS Mumbai is the right choice today for data localisation compliance but at 10x scale it creates latency risk for non-India traffic and concentration risk in the event of a regional outage. Our roadmap includes infrastructure readiness for US data residency by Q4 2026.
DevSecOps capacity is a gap we are candid about. Our current security function is overseen at the leadership level. At 10x client scale the volume of security reviews, compliance audits, and incident monitoring will exceed what can be managed without a dedicated function. Building a dedicated DevSecOps team is one of our priority hiring areas.
The Memory System and Context Engineering Layer will face significantly higher read volumes at 10x scale. We will need to invest in caching strategies, read replica infrastructure, and optimised retrieval pipelines to maintain performance.
Onboarding and customer success will not scale at 10x without product investment. This is precisely why Agent Builder self-serve capability and PLG features are central to our 2026 roadmap, reducing human effort required per client deployment.
What will not break is equally important to note. Our provider-agnostic architecture with alternate providers per capability scales horizontally without single points of failure. Our EKS based infrastructure with auto-scaling handles incremental growth without major architectural changes. Our evaluation and simulation systems are automated and scale with deployment volume. Our single-tenant architecture scales predictably because each client environment is independent. |
Yes, we have a fully insourced AI and ML team with no outsourcing of any kind.
The team is led by Pankaj Gupta, our CTO and Co-founder, who holds a BS and MS Dual Degree in Mathematics and Scientific Computing from IIT Kanpur. He brings over 10 years of hands-on experience across AI, ML, data science, and engineering leadership. Prior to RevRag he was a Data Scientist at Adobe working on traditional ML, deep learning, and computer vision models, led recommendation model development at ShareChat, led recommendation and search systems at Udaan, and built the technology team from scratch at Lohum. At RevRag he leads all technology, AI, and security functions and is also the primary AI lead.
Beyond the CTO, our founding team brings additional AI depth. Neeraj Gupta, our COO, leads AI implementation, agent building, and client delivery with hands-on involvement in agent design and deployment at the leadership level.
The broader AI engineering team works on our core AI stack covering voice orchestration, context engineering, the memory system, the workflow framework, and the evaluation and simulation infrastructure. The team spans backend engineering, frontend engineering, and AI with all three co-founders actively involved in day to day execution.
Our priority hiring areas going forward are engineering depth in backend and AI as we scale across more enterprise clients and new channels, and a dedicated DevSecOps function to complement our existing ISO 27001 and SOC 2 compliance posture.
In short, AI is not a bolt-on capability at RevRag. It is the core of what we build and the founding team has the technical depth to lead it from first principles. |
We do not rely on large-scale training datasets and we do not train or fine-tune our own models. Our architecture is built on context engineering and prompt orchestration with third-party foundation models, which means the traditional concept of a training dataset does not apply to us in the conventional sense.
The data assets we work with fall into three categories.
Client-provided data is the primary input to our system. This includes knowledge bases, CRM and CDP data, and conversation history that clients share with us for the purpose of configuring and running their AI agents. This data is governed by Data Processing Agreements that explicitly define permissible usage. It is used exclusively to improve that specific client's agents and never used for cross-client purposes or model training.
Proprietary interaction data is generated through our own production deployments. Every conversation our agents handle generates structured behavioural data covering user journey events, drop-off patterns, resolution outcomes, and interaction quality metrics. This data is owned by RevRag and is the foundation of our BFSI-Specific Interaction Graph and RevRag Central Intelligence layer. It is not licensed from any third party and cannot be replicated by a competitor without equivalent production history.
Third-party foundation model APIs form the base inference layer. We use OpenAI, Google Gemini, Azure, ElevenLabs, Sarvam, and Deepgram through commercial API licenses. All vendors are configured for zero or minimal data retention, meaning client data is not used by these vendors for their own model improvement. All third-party usage is governed by commercial agreements with clearly defined data handling terms.
In short, our data assets are either owned outright through production operations or governed by explicit contractual agreements. We have no dependency on unlicensed or grey-area datasets. |
We have a continuous improvement cycle built into the platform at every layer, from pre-deployment benchmarking to post-deployment monitoring to cross-deployment learning.
Before any agent goes live, our pre-call simulation system runs automated test scenarios benchmarked against defined quality criteria. This gives us a reproducible baseline for every agent and every configuration change. Any prompt update, model swap, or workflow change is validated against this baseline before it reaches a live client.
Once an agent is live, our post-call evaluation system automatically analyses every conversation measuring call success rate, task resolution, re-engagement outcomes, and user satisfaction. These metrics feed directly back into prompt refinement and context engineering improvements, creating a continuous improvement cycle for each client's agents without requiring any model retraining.
Our A/B testing infrastructure allows us to run controlled experiments on prompt variants, model configurations, and agent behaviours in production. Every change is measured against a baseline before it is rolled out at scale, ensuring improvements are validated rather than assumed.
At the platform level, our RevRag Central Intelligence layer captures what worked and what did not across all deployments and feeds it back into improvements that benefit every client on the platform. Each new deployment makes the platform smarter for all subsequent ones.
On the model side, our provider-agnostic architecture means we can swap to better performing models as the underlying foundation model landscape improves, without any code deployment. We are also actively integrating open-source model alternatives to improve cost efficiency and reduce dependency on any single provider.
Looking ahead, our Q3 2026 roadmap includes a small language model POC specifically for RevRag Central Intelligence, which will give us even greater control over the core intelligence layer and reduce our dependency on third-party foundation models for the most critical parts of our platform.
In short, model performance improvement at RevRag is not a periodic exercise. It is a continuous, automated, and compounding process built into the core of how our platform operates. |
We are building for India first and global markets next, in that order and for deliberate reasons.
India is our foundation and our current focus. We have 13 signed clients across India's BFSI ecosystem including PhonePe, Mahindra Financial Services, Motilal Oswal, Razorpay, Finbox, Branch International, and others spanning payments, lending, wealth management, and insurance. Our platform is purpose built for the Indian BFSI context, with native support for Indian languages through Sarvam, compliance with TRAI, RBI, and SEBI regulations, data localisation on AWS Mumbai, and BFSI-specific behavioural data accumulated from live production deployments at scale.
India is not a stepping stone for us. It is a large, complex, and fast-growing market in its own right. India's BFSI sector spans over 1,500 NBFCs, 100 plus banks, and thousands of fintechs, all under mounting pressure to scale customer engagement while reducing cost per interaction. We believe we can build a very large business in India alone.
Global expansion is on our roadmap with the US as the first international market. Our Q4 2026 product roadmap includes US market GTM readiness covering US data residency infrastructure, international voice readiness, and multi-language support. We plan to hire a dedicated US sales team of 5 in FY28 and are building our LATAM partner pipeline via our investor network in parallel.
Our expansion sequencing is intentional. The data moat, compliance posture, and switching costs we are building in India create a template that is repeatable in other markets. The BFSI AI agent problem is not unique to India. Every market with complex regulated financial workflows has the same need and the same gap that RevRag fills.
In short, India first, global next, and the foundation we are building today is designed to travel. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/cEPTrIu7NSjG-Pitch-Deck-RevRag-AI.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/cEPTrIu7NSjG-Pitch-Deck-RevRag-AI.pdf |
https://youtu.be/l6fN9RkfICk?si=Z0fiTqfBElipQ6oX |
Yes, RevRag AI is mission-driven at its core, and the mission is straightforward: make intelligent, personalised financial services accessible to every person in India, regardless of how complex the underlying workflow is.
India has one of the largest and most underserved financial services populations in the world. Hundreds of millions of people interact with banks, NBFCs, insurance companies, and fintechs every day, but the experience is often broken. Onboarding flows are abandoned mid-way. Loan applications get stuck. Insurance renewals are missed. Collections happen through aggressive human calling. The gap between what financial institutions intend to deliver and what their customers actually experience is enormous, and it exists largely because scaling personalised human interaction is prohibitively expensive.
RevRag exists to close that gap using AI. Every agent we deploy makes a financial institution's product more accessible, more responsive, and more human for the end customer. When a first-time borrower gets guided through a loan application without dropping off, when a wealth management client gets a personalised follow-up at exactly the right moment, when a collections interaction happens through a respectful and contextualised conversation rather than a cold call, that is a real improvement in a real person's experience of financial services.
The impact compounds with scale. We are not just building software for enterprises. Every enterprise client we power serves millions of end users. PhonePe alone has hundreds of millions of users in India. Motilal Oswal serves lakhs of investors. Mahindra Financial Services reaches borrowers in semi-urban and rural India who have limited access to branch banking. When our agents work well for these institutions, the downstream impact reaches people who have historically been underserved by the financial system.
We also believe that building responsibly is part of the mission. Our AI governance framework, human-in-the-loop decision points, hallucination controls, bias detection, and data minimisation practices are not just compliance checkboxes. They reflect a genuine belief that AI in financial services must be trustworthy to be useful, and trustworthy to be fair.
In short, we are building RevRag because we believe AI can make financial services work better for the people who need it most, and we think that is worth building seriously and building well. |
No |
Amit Mishra, Founder Dazeinfo |
checked |
| Apr 14, 2026 @ 9:13 PM |
Aiswarya Ravi , Tushar Cherian |
aiswarya@theoramadvisory.com |
https://www.linkedin.com/in/aiswaryaravi10/ |
http://NA |
+918971862233 |
C-Suite and Board Member of fintech for last decade, Treasury and partnership builder specialist |
We worked together in our last role (an NBFC fintech out of Bangalore) for more than 4 years. The mutual respect and synergy in our work and our vision brought us together to co-create this new venture. |
2 |
Our combined team strength is that while we can envision a new idea/venture we are hard core execution specialists who have operated and been part of multiple organizations and scaled them up. We understand multiple target segment and can truly morph a mere idea into practical launch and scale it to its full potential. |
KAI (Built under Theoram Advisory LLP) |
KAIKITCHEN.AI |
Bangalore |
Making meal planning seamless through AI based digitization of your kitchen |
The problem of a blind kitchen which creates a ripple effect of over ordering, wastage and mental load on planning meals |
KAI solves the problem of the “blind kitchen” by creating a real-time, AI-powered digital twin of the household kitchen.
It automatically aggregates grocery data from online platforms, receipts, and physical inputs using computer vision, scanning, and voice—eliminating the need for manual tracking. This creates a live, unified view of what exists in the kitchen, including quantities and expiry timelines.
On top of this data layer, KAI acts as an intelligent decision engine. It recommends meals based on what is already available—prioritizing items nearing expiry—while suggesting only essential purchases, thereby reducing over-ordering and waste. It also coordinates seamlessly with household members or cooks, ensuring execution without friction.
By closing the loop between buying, cooking, and consumption, KAI transforms the kitchen from a reactive, memory-driven system into a proactive, data-driven environment—making everyday food decisions efficient, sustainable, and stress-free |
KAI’s uniqueness lies in owning the “inventory layer” of the kitchen—something no current solution captures effectively. Unlike recipe apps, grocery platforms, or generic AI tools, KAI creates a real-time, unified view of household food by aggregating fragmented data across online and offline sources. This proprietary data layer enables highly contextual, execution-ready recommendations grounded in actual consumption, not assumptions. Over time, KAI builds a defensible data flywheel—more usage leads to better consumption insights, improved AI accuracy, and higher user lock-in. Additionally, features like multi-user coordination and integration with household help create strong behavioral and cultural stickiness, making KAI deeply embedded in daily routines and difficult to replace. |
Idea |
Pilots |
Busy households in Tier 1 and Tier 2 Indian cities. |
$30bn Indian smart households market |
$5Bn Smart kitchen/Food tech |
$250mn (initial foray) |
D2C subscription (primary), Commerce & Affliate Revenue |
Recipe & meal apps like samsung whisk, health apps like healithify, quick commerce like blinkit, zepto |
Community led growth (primary), referral and content led discovery. |
Community led entry through whatsapp/slack/telegram, then D2C scale through media and content, and then B2B2C |
Imagine this: what if you could hold your entire kitchen in your phone—and seamlessly coordinate everything around it?
KAI is building the intelligence layer for food within every household, transforming the kitchen from an invisible, memory-driven space into a fully connected, real-time system. From knowing exactly what’s in your kitchen to planning meals, coordinating with family or cooks, and making smarter purchase decisions—everything flows through one unified platform.
Over time, KAI will evolve beyond the individual household into a closed-community food network—where trusted groups can share recipes, meals, and daily cooking moments, creating a private, Instagram-like layer for food within real-life communities.
At scale, KAI enables intelligent redistribution of surplus food, reducing waste while improving access and sustainability.
The vision is simple but powerful: to make every food decision—what to cook, what to buy, what to share—informed, connected, and efficient, turning food from a fragmented process into a coordinated ecosystem. |
Incorporated |
Bootstrapped |
Yes |
We are applying to Avinya because KAI is building an AI-native product for a real-world, everyday problem that has remained invisible for too long: household food management. KAI uses AI to create a live digital twin of the kitchen, turning fragmented purchase, inventory, and consumption data into actionable decisions around meals, waste reduction, and coordination. Avinya stands out because it is designed specifically for early-stage AI startups, with structured mentorship, access to datasets, deep industry engagement, and seed support to help ventures move from promising products to scalable companies. For KAI, this is especially relevant because our next phase depends on strengthening the AI core of the product: improving receipt parsing, inventory detection, expiry prediction, personalization, and recommendation quality. We believe Avinya can help us sharpen both the technical defensibility and the go-to-market discipline needed to build a category-defining AI company from Karnataka. |
During the program, our goal is to transition KAI from a working product to a robust, AI-native system with clear scalability and defensibility. On the product side, we want to significantly strengthen our AI core—improving inventory detection (receipt parsing, computer vision), expiry prediction, and personalized meal recommendations based on real household behaviour. This is critical to delivering consistently accurate, high-utility outputs that drive daily engagement. On the business side, we aim to validate strong user retention and conversion, targeting consistent weekly usage and early paid adoption across our initial user base. We also want to refine our go-to-market strategy, particularly community-led growth and early B2B pilots. Equally important, we want to build a clear data and model strategy—defining how KAI’s proprietary dataset and feedback loops translate into long-term defensibility. By the end of the program, we aim to have a production-ready AI product, strong early traction, and a clear path to scale and fundraising. |
Open to hybrid or in-person |
Yes |
Core engine |
KAI is built as an inventory-first intelligence system — a real-time digital twin of the household kitchen. At its core is a multimodal ingestion layer that structures messy real-world data from grocery receipts, app confirmations, barcode scans, images, and voice inputs. Computer vision and OCR extract items and quantities, while NLP and entity-resolution models normalize products into a canonical food ontology. This feeds a unified state layer that acts as the single source of truth, enabling real-time visibility across users, cooks, and devices.
On top of this, prediction models estimate expiry, consumption velocity, and replenishment needs. A ranking layer recommends meals optimized for availability, freshness, and preferences—turning waste reduction into a system outcome. An LLM orchestration layer generates contextual outputs (meals, substitutions, instructions) grounded strictly in the inventory state, minimizing hallucination. The architecture is modular and API-first, with a continuous feedback loop where ingestion signals and user actions improve model accuracy over time—creating a compounding data advantage |
KAI’s proprietary data advantage comes from being the first system to unify fragmented household food data into a continuous, real-time consumption dataset. Today, grocery platforms, recipe apps, and delivery services each see only a small slice of user behavior. KAI aggregates data across digital receipts, offline purchases, images, voice inputs, and household interactions—creating a complete view of what exists, what is consumed, and what is wasted inside the kitchen.
This enables us to build a unique dataset around consumption patterns, expiry behavior, substitution choices, and household coordination, which is not available in any existing system. Over time, this forms a compounding data flywheel—more usage improves inventory accuracy, prediction models, and recommendation quality.
Additionally, KAI captures India-specific food and household dynamics, including regional cuisines, unstructured ingredient formats, and multi-user environments (families and cooks), which are significantly underrepresented in existing datasets.
This combination of cross-source aggregation, real-world consumption tracking, and culturally specific data creates a strong and defensible foundation for long-term AI performance and differentiation |
KAI’s AI defensibility comes from the combination of a proprietary data layer, system design, and continuous learning loops.
Unlike competitors that operate on static or partial data (recipes, transactions, or calorie logs), KAI builds a real-time, household-level inventory and consumption dataset by aggregating fragmented inputs across online and offline sources. This gives us access to the post-purchase layer of food behaviour, which is largely invisible to existing platforms.
Our models are tightly coupled with this state layer—predictions (expiry, replenishment) and recommendations improve as the system observes actual consumption, waste patterns, and user decisions over time. This creates a compounding data flywheel, where more usage leads to better accuracy and higher utility.
Additionally, KAI is designed for multi-user, real-world environments (families, cooks, shared kitchens), creating behavioural lock-in and richer interaction data that is difficult to replicate.
Together, this makes KAI not just an AI application, but a learning system that becomes more accurate, personalized, and indispensable with scale. |
We track extraction accuracy, prediction accuracy (expiry/replenishment), and user-driven outcomes like meal adoption and waste reduction, while ensuring reliability through grounded, inventory-based outputs and continuous feedback loops. |
We handle data privacy, compliance, and security through a privacy-by-design approach. All user data is encrypted in transit and at rest, with strict access controls and role-based permissions. We collect only the minimum data required to deliver functionality and ensure clear user consent for all data inputs (receipts, images, voice). Personally identifiable information is segregated and anonymized where possible for model training.
KAI is designed to comply with applicable data protection regulations (such as India’s DPDP Act), with transparent data usage policies and user control over data access and deletion. We do not share or sell user data to third parties without explicit consent. Security best practices, regular audits, and monitoring are implemented to ensure data integrity and system resilience. |
KAI operates as a consumer software platform, so direct regulatory risk is relatively low. However, we are mindful of evolving regulations in data privacy and AI governance, particularly under India’s Digital Personal Data Protection (DPDP) Act. Changes in consent requirements, data storage, or cross-platform data usage could impact how we collect and process user data.
There is also potential dependency on third-party platforms (e.g., email, receipts, or grocery integrations), where changes in access policies or APIs could affect data ingestion.
We mitigate these risks through a privacy-first architecture, explicit user consent flows, minimal data dependency, and a modular system design that allows flexibility in adapting to regulatory or platform changes. |
At 10x scale, the first pressure points would be data quality, latency, and model consistency across highly variable real-world inputs. As user volume grows, ingestion pipelines for receipts, images, and voice will face more edge cases, which can reduce extraction accuracy if not continuously retrained. Recommendation quality can also degrade if the system does not adapt fast enough to new household patterns, cuisines, and usage contexts. On the infrastructure side, real-time inventory updates and orchestration may face latency and cost pressure. We mitigate this through a modular architecture, human-in-the-loop validation where needed, monitoring at each pipeline stage, and progressively fine-tuned models trained on KAI-specific data |
We currently do not have a full in-house AI/ML team. The product has been built in collaboration with an experienced tech partner with prior expertise in developing AI-driven systems and production-grade applications.
From the founder side, I bring hands-on experience in building AI-driven decision models for underwriting in fintech, including designing data pipelines, risk frameworks, and model-led decisioning systems at scale. This provides strong grounding in how to structure, evaluate, and deploy AI systems in real-world environments.
As we move into the next stage, we are actively looking to onboard a founding CTO / AI lead to bring core AI capabilities in-house. Our goal is to build a strong internal AI team over time, particularly around data, modeling, and system architecture, while continuing to leverage external expertise selectively for speed. |
We are currently using a combination of publicly available and general-purpose datasets (e.g., food/recommendation datasets) to build and validate our initial models during the PoC stage. These datasets are either open-source or appropriately licensed for development use.
Our approach is to use external datasets for bootstrapping models, while systematically transitioning to owned, real-world household consumption data as KAI scales. |
We improve model performance through a continuous data and feedback loop. As users interact with KAI—adding inventory, selecting meals, making substitutions, and completing actions—we capture real-world signals on accuracy, preferences, and outcomes. This data is used to retrain and fine-tune models for extraction, prediction, and recommendations.
We also run regular evaluation pipelines to track accuracy (e.g., item extraction, expiry prediction) and optimize ranking models based on user behavior. A modular architecture allows us to upgrade individual components (CV, NLP, recommendation models) without disrupting the system. |
Initially India and then global. Food is global and the problem is transcending geography. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/pkALEj6F5TIm-KAI-Pitch-Deck-vf-April.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/pkALEj6F5TIm-KAI-Pitch-Deck-vf-April.pdf |
https://kaikitchenai.netlify.app/ |
KAI is a mission-driven startup with strong impact outcomes. Our core mission is to make everyday food consumption more efficient, informed, and sustainable by solving the problem of “kitchen blindness.”
While we are building a scalable consumer and enterprise product, the impact is inherent to the model—reducing food waste, improving household-level food utilisation, and lowering unnecessary consumption. This directly contributes to better resource efficiency and climate outcomes without requiring behavior change as a primary barrier.
We believe the strongest impact comes from embedding sustainability into a high-utility product, where users adopt it for personal value, and impact is delivered as a natural outcome at scale. |
NA |
NA |
checked |
| Apr 14, 2026 @ 2:46 PM |
Imtimongba - Founder ; Tiakumzuk - co-founder |
imtimongbaimsong1991@gmail.com |
https://www.linkedin.com/in/imtimongba-36557617b/ |
https://nagademy.in/ |
8088784892 |
Imti - Product Development and operation ; Tia - Technology |
My cofounder is my brother, and we have known each other our entire lives. Over the years, we have developed a strong understanding of each other’s strengths, working styles, and decision-making approaches. While this is our first formal venture together, our long-standing relationship and trust have enabled us to collaborate effectively and move quickly in building Nagademy |
two |
Our biggest strength is clear ownership and decision-making. For product development and customer-facing aspects, I take the final call, ensuring the product aligns closely with user needs and market direction. For technology and backend systems, my cofounder takes the final call, ensuring strong technical architecture and execution. This clarity in roles allows us to move fast, avoid conflicts, and make efficient, high-quality decisions. |
Nagademy AI |
https://nagademy.in/ |
Dimapur |
Nagademy is India’s first AI-powered full-stack preparation platform, building the AI Operating System for government exam preparation |
Government exam aspirants lack a structured, adaptive preparation system—current AI platforms are non-personalized, fragmented across Prelims, Mains, and Interview, and fail to provide behavioral reinforcement or realistic AI-driven interview practice, leading to inefficient preparation and poor outcomes. |
Nagademy is an AI-powered full-stack preparation platform that provides adaptive learning, integrates Prelims, Mains, and Interview into one continuous journey, and uses behavioral reinforcement to improve consistency. It also offers AI-driven mock interviews using audio-based models, delivering personalized feedback and a structured, end-to-end preparation system for aspirants. |
Nagademy’s defensibility comes from its preparation intelligence layer, not just content AI wrapper. We are building a system that continuously learns from user performance, behavior, and preparation patterns to deliver truly adaptive strategies for each aspirant. Over time, this creates a proprietary data moat around what actually drives success in government exams. Additionally, our full-stack approach—integrating Prelims, Mains, and Interview into one continuous system—combined with behavioral reinforcement mechanisms and AI-powered interview evaluation, makes the platform difficult to replicate. As usage grows, the system becomes smarter and more personalized. |
Users |
one |
Our target customers are aspirants preparing for state government job exams, particularly students and graduates from Tier 2 and Tier 3 cities who seek structured, affordable, and technology-driven preparation. These users are typically self-preparing or underserved by traditional coaching systems and are looking for a more personalized, guided, and outcome-focused preparation approach. |
$17.21 Billion |
$1.52 Billion |
$400M+ |
Subscripton |
Super Kalam, Dalvoy AI |
We acquire customers through free mock tests, PYQ analysis, and preparation resources that demonstrate immediate value. we also acquire through Local partnerships: We collaborate with libraries, and student communities to onboard users offline. |
Online - Social media through educational content, Influencer marketing, youtube ads ; Offline - Partnered with 10+ libraries and study space centers |
Our vision is to build the AI Operating System for government job aspirants, where every aspirant gets:
- a personalized AI Mentor
- end-to-end preparation system
- motivation to stay consistent
- intelligent feedback
We also plan to launch an "AI Teacher" which will be a course created by an AI avatar and teach students in different languages, and not just in English & Hindi. Imagine "Khan Sir or Alakh Pandey", teaching a topic in their own style but explaining in Kannada or Marathi or Bengali. This is logically not possible but AI can make this happen and We want to be at the front row making this to reality. |
Incorporated on 19 Jan, 2026 |
10,00,000 as grant |
one |
We are applying to Avinya to accelerate Nagademy’s journey from an early-stage product to a scalable, high-impact platform. Avinya’s focus on building meaningful, technology-driven solutions aligns closely with our vision of transforming government exam preparation through AI. We are particularly looking for mentorship, strategic guidance, and access to the right ecosystem to refine our product, strengthen our go-to-market strategy, and execute large-scale pilots. Additionally, Avinya can help us connect with institutional partners and unlock opportunities that are critical for scaling in the public sector ecosystem. We believe Avinya can play a key role in helping us build a category-defining AI platform in this space. |
During the programme, we aim to achieve three key outcomes: 1. Pilot validation at scale: Secure and execute structured pilot programs with institutional partners to validate real-world impact and improve product-market fit. 2. Product refinement and intelligence layer: Strengthen our core AI systems—especially adaptive preparation and AI mock interviews—by improving accuracy, personalization, and user outcomes. 3. Go-to-market and growth engine: Build a repeatable and scalable user acquisition model, while strengthening retention through behavioral reinforcement and incentives. Overall, our goal is to move from an early-stage product to a validated, scalable platform with strong user traction and measurable outcomes. |
Yes, we are open to both hybrid and virtual participation. We are flexible and willing to adapt to the programme’s format to ensure maximum engagement, collaboration, and learning |
one |
one |
Nagademy’s AI architecture is built as a modular, data-driven system that combines multiple AI models with a centralized intelligence layer.
At the core is our Preparation Intelligence Engine, which collects and analyzes user data such as test performance, response patterns, accuracy, and consistency. This engine uses machine learning models to identify strengths and weaknesses and dynamically generate adaptive preparation strategies for each aspirant.
For intelligence and evaluation, we leverage the OpenAI API, which powers answer evaluation, feedback generation, and overall decision-making across the platform. For content generation, including mock tests and practice questions, we use Google Gemini to ensure diverse and scalable question creation.
Our AI Mock Interview system is currently powered by OpenAI’s speech-to-speech model, enabling real-time voice interaction, speech processing, and feedback on communication, reasoning, and articulation. We are also actively exploring alternative models to further optimize performance, cost, and scalability.
In the next phase, we will integrate multilingual translation models to power our AI Teacher feature, enabling personalized learning experiences across regional languages. |
Nagademy is building a proprietary data advantage around aspirant performance, preparation behavior, and evaluation patterns.
As users interact with the platform, we collect structured data on:
- test performance and accuracy trends
- answer quality across different subjects and formats
- preparation consistency and behavioral patterns
- interview responses and communication signals
This allows us to continuously improve our adaptive preparation engine, identifying what strategies actually lead to better outcomes over time.
Additionally, our AI-based evaluation and mock interview systems generate unique datasets on how aspirants think, write, and respond—data that is not easily available or structured in traditional coaching systems.
Over time, this creates a compounding data moat, enabling more accurate personalization, better predictions, and increasingly effective preparation pathways that are difficult for competitors to replicate |
Our defensibility comes from the intelligence layer built on proprietary user data, not just the use of AI models.
Nagademy continuously learns from aspirant performance, preparation behavior, answer patterns, and interview responses, creating a unique dataset that improves our adaptive preparation engine over time. This allows us to deliver increasingly accurate personalization and strategy recommendations that are difficult to replicate without similar data.
In addition, our full-stack integration across preparation stages and our AI-driven evaluation and interview systems generate structured insights that compound with scale. |
We evaluate model performance using a mix of accuracy, consistency, and user outcome metrics.
1. For answer evaluation and feedback quality, we check:
- Accuracy of scoring against human-reviewed benchmarks
- Consistency of evaluations for similar responses
- Relevance of feedback to the actual answer
- Hallucination rate to ensure the model does not generate unsupported feedback
2. For question generation, we measure:
- Topic relevance
- Difficulty balance
- Duplication rate
- Quality of generated questions through internal review and user response quality
3. For the AI mock interview system, we track:
- Speech-to-text accuracy
- Response evaluation consistency
- Latency / response time
- User satisfaction with feedback
4. For overall reliability, we also monitor:
- Uptime and failure rate
- Cost per request
- Human validation against sample outputs
In simple terms, we do not judge the model only by technical accuracy. We also check whether it gives useful, stable, and exam-relevant output for aspirants. |
We follow strict data protection practices by encrypting user data, limiting access through role-based controls, and storing only essential information. All data is processed securely via trusted APIs and HTTPS (TLS) protocol. We do not share user data with third parties without consent, and continuously monitor systems to prevent unauthorized access or misuse. |
Nagademy operates in the education and AI space, which currently has low direct regulatory barriers, but there are a few considerations:
- AI-related regulations may evolve, especially around transparency, bias, and responsible use of AI-generated content.
- Content and exam-related sensitivity, ensuring that generated questions and materials do not infringe on copyrighted or restricted sources.
We actively design our systems to be compliant, privacy-focused, and adaptable to future regulatory changes, minimizing risk while maintaining trust and reliability. |
At 10x scale, the main challenges are around latency, cost, and consistency of outputs.
- Latency & performance: Real-time features like AI mock interviews and evaluations may slow down with high concurrent usage if not properly optimized.
- Cost scaling: Heavy reliance on external APIs (for evaluation, audio, and generation) can significantly increase costs as usage grows.
- Consistency of outputs: With more diverse user inputs, maintaining stable and reliable AI feedback becomes harder without stronger validation layers.
- Data handling & infrastructure: Increased load on databases and processing pipelines may affect system reliability if not scaled properly.
- Quality control: Ensuring accuracy and reducing hallucinations at scale requires better monitoring and feedback loops. |
Yes, we have in-house AI/ML capability led by my cofounder, who has prior experience working as a machine learning engineer in Bangalore. He is responsible for building and managing our AI systems, including model integration, backend architecture, and performance optimization.
Additionally, we receive technical guidance from another brother who is an AI/ML engineer currently working in the Netherlands. He has deep expertise in AI systems and architecture and acts as a technical advisor, providing strategic direction and oversight, although he is not formally part of the company. |
We primarily use a combination of publicly available exam-related data and internally generated datasets.
- Public datasets include previous year questions (PYQs), exam patterns, and syllabus structures that are widely available in the public domain.
- Internally generated data includes user performance data, answer responses, mock test attempts, and AI-evaluated outputs generated through platform usage.
Our question generation is powered by AI models (Gemini), and evaluation is done using AI systems (OpenAI), but the resulting user interaction data and preparation insights are owned by us.
We ensure that all data used is either publicly accessible or generated within our platform, and we do not rely on any unauthorized or scraped proprietary datasets. |
We improve model performance through a combination of data feedback loops, evaluation systems, and iterative optimization.
- User feedback loops: We continuously learn from user interactions—test performance, answer quality, and interview responses—to refine our adaptive systems.
- Human-in-the-loop validation: Sample outputs are reviewed to benchmark accuracy and improve evaluation quality.
- Prompt and system optimization: We regularly refine prompts and workflows to improve consistency, reduce hallucinations, and enhance relevance.
- Performance monitoring: We track key metrics like accuracy, consistency, latency, and user satisfaction to identify gaps.
- Model upgrades and diversification: We evaluate and integrate better models over time to improve quality and scalability.
Over time, our growing proprietary dataset further strengthens the system, making it more accurate and personalized with scale. |
We are currently focused on India, specifically on state government exam aspirants, where we see a large and underserved market.
However, the core technology we are building is globally applicable. In the future, we plan to expand to international markets by adapting our platform for aspirants preparing for exams such as GMAT, GRE, MCAT, along with broader "Interview Preparation" use cases.
Our long-term vision is to build a scalable AI-powered preparation system that can serve both Indian and global aspirants across different high-stakes exams and career pathways. |
|
|
Yes, Nagademy is a mission-driven and impact-focused startup.
Our mission is to make high-quality, structured government exam preparation accessible to every aspirant, regardless of their background or location. Today, many aspirants—especially from Tier 2 and Tier 3 cities—lack access to personalized guidance and rely on fragmented resources or expensive coaching.
By using AI to deliver adaptive preparation, behavioral reinforcement, and interview readiness, we aim to improve outcomes, reduce inefficiencies, and save time for aspirants. Our goal is not just to help students prepare better, but to enable capable individuals to enter public service, ultimately contributing to stronger governance and societal development. |
Yes, we come from a Tier 3 and below region in India, where access to startup ecosystems, funding, and mentorship is relatively limited compared to metro cities. Building from this background gives us a deep understanding of the challenges faced by aspirants in similar regions, which directly shapes our product and mission. |
NA |
checked |
| Apr 14, 2026 @ 2:46 PM |
Jerold Vinod |
jeroldvinod2@gmail.com |
https://www.qoqikylutakyfy.org.uk |
https://www.ligalyva.cm |
09620736706 |
Libero assumenda non |
Ea temporibus amet |
All |
Itaque sed eum quos |
Kylan Benjamin |
https://www.mykuzasegonohoz.org.uk |
Voluptate quia conse |
Qui non amet omnis |
Molestiae aliquid eo |
Magnam repudiandae e |
Consectetur delectu |
Users |
Testimonials |
Culpa fugit quia a |
Aut voluptatem bland |
Dolor sunt nihil aut |
Dolore odio elit qu |
Reiciendis nihil nis |
Vero exercitationem |
Dolore esse in qui |
Possimus ex asperna |
Quisquam eos possim |
Est molestiae perspi |
Aut esse do excepteu |
two |
Molestiae at officia |
Lorem aliquid provid |
Voluptatem quibusdam |
one |
Not-applicable |
Culpa inventore natu |
Aperiam officia ad h |
Nostrud excepteur eo |
Minus in deserunt po |
Pariatur Et laborio |
Ut architecto sint |
Repellendus Eum ess |
Neque beatae dolor n |
Ad in voluptatum vol |
Possimus est sunt e |
Et repudiandae qui p |
|
https://www.zykamohyqejaq.info |
Dolore expedita vero |
Ea dolorem dolore et |
Fugiat ipsum except |
checked |
| Apr 14, 2026 @ 2:22 PM |
Jerold Vinod |
jeroldvinod2@gmail.com |
https://www.hyxoj.cm |
https://www.pyguberiquje.cc |
9620736706 |
Incididunt vitae ull |
Aperiam magna est t |
one |
Fugit reiciendis es |
Neville Ramirez |
https://www.hutyzewixobeq.info |
Sed sit enim adipisi |
Corporis voluptas ei |
Sint a odit iste mag |
Qui fuga Qui fuga |
Quia nihil modi nost |
Users |
two, Signups, Testimonials |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 14, 2026 @ 2:18 PM |
Jerold Vinod |
jeroldvinod2@gmail.com |
https://www.vulimoxamicuvu.ca |
https://www.geroxavexevoxag.org.au |
09620736706 |
Unde labore minima e |
Id labore quis neque |
two |
Consectetur et magna |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 14, 2026 @ 12:19 PM |
Anuj Srivastava, Priyesh Srivastava |
anuj@onfinance.in |
http://linkedin.com/in/anujsrivastava02 |
https://www.onfinance.ai/ |
9811070401 |
Anuj Srivastava — CEO Leads business strategy, product vision, and overall growth Former experience across global financial institutions including HSBC, Barclays, Credit Suisse, and WorldQuant Deep expertise in financial markets, quantitative finance, and regulatory ecosystems Recognized in Forbes 30 Under 30 Asia BITS Pilani, Goa alumnus | Priyesh Srivastava — CTO Leads technology, AI/ML architecture, and platform development Previously worked at Nutanix, Pixxel, Saavn, and DemandMatrix Strong background in scalable systems, AI engineering, and data-driven platforms Recognized in Forbes 30 Under 30 Asia BITS Pilani, Goa alumnus |
We met at BITS Pilani, Goa, where we both studied. We’ve known each other since our university days and have been working together closely since founding OnFinance AI. |
2 |
Our biggest strength is the combination of deep domain expertise in financial services and strong technical execution. With experience across global banks and leading tech companies, we’re able to both understand regulatory complexity and build scalable AI systems to solve it effectively. |
OnFinance AI |
https://www.onfinance.ai/ |
Bangalore |
AI-powered regulatory compliance platform that helps financial institutions manage their regulatory compliance end-to-end |
Financial institutions struggle with managing regulatory compliance due to fragmented processes, manual workflows, and constantly changing regulations. This leads to inefficiencies, higher risk, and significant operational overhead. |
We’ve built ComplianceOS, an AI-powered platform that enables financial institutions to manage regulatory compliance end-to-end, from automatically fetching and interpreting regulations to generating actionable tasks, assigning ownership, tracking progress, and ensuring audit readiness. |
Our defensibility comes from the combination of proprietary AI (NeoGPT) and deep domain expertise in financial regulations. NeoGPT is trained to understand complex regulatory language and convert it into structured, actionable workflows—something generic LLMs cannot do reliably. Over time, our system gets stronger with proprietary compliance data, workflows, and institutional integrations, creating a strong data and product moat. |
Product-Market Fit |
Users, Revenue, Pilots, Signups, Testimonials |
Our target customers are financial institutions—including banks, stock exchanges, and other regulated entities—that need to manage complex regulatory compliance workflows. |
$25B+ |
$8–10B |
$200–500M |
B2B |
NA |
Our go-to-market is focused and relationship-driven. We run targeted outbound campaigns to compliance, risk, and audit leaders at financial institutions, supported by warm introductions from our existing network and clients. We also leverage partnerships with industry bodies and ecosystems. Most conversions happen through structured demos and pilot deployments, which expand into enterprise-wide rollouts. |
We follow a focused, enterprise-led go-to-market strategy targeting compliance, risk, and audit teams at financial institutions. We start with targeted outbound and warm introductions, run demos and pilot deployments to prove value, and then expand into enterprise-wide rollouts. |
We follow a focused, enterprise-led go-to-market strategy targeting compliance, risk, and audit teams at financial institutions. We combine targeted outbound with warm introductions, run demos and pilot deployments to quickly demonstrate value, and then expand into enterprise-wide rollouts. |
Incorporated in 2023 |
5.3 Million USD |
Yes |
We’re applying to Avinya to accelerate our growth with the right mentorship, network, and ecosystem support. As we scale ComplianceOS across financial institutions, we believe Avinya can help us refine our go-to-market strategy, unlock enterprise connections, and strengthen our positioning as a category-defining AI platform in regulatory compliance. |
During the programme, we aim to accelerate enterprise adoption, particularly across India and the Middle East, by strengthening our sales motion and partnerships. We also want to refine our product positioning, deepen our AI capabilities (NeoGPT), and build strong relationships with industry stakeholders and potential customers. |
Yes, we are open to both hybrid and virtual participation. |
Yes |
Core engine |
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/nc1ZNrYL1aFu-OnFinanceAI_PitchDeck_2026.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/nc1ZNrYL1aFu-OnFinanceAI_PitchDeck_2026.pdf |
https://www.youtube.com/watch?v=fjM93dvK9wU |
Yes, we are building a mission-driven company. Our goal is to make regulatory compliance more efficient, transparent, and reliable for financial institutions. By reducing manual effort and improving accuracy, we help institutions operate more responsibly, reduce systemic risk, and build greater trust in financial systems. |
No |
Email |
checked |
| Apr 14, 2026 @ 12:14 PM |
H R TAJUDDIN BASHA |
tajuddin@aroushtech.in |
https://www.linkedin.com/company/103663927/admin/dashboard/ |
https://aroushtech.in/ |
8867692074 |
Founder and Director |
as long as the any work or project will work together |
1 |
well technical skill team. we belive out skills as per the IT Standard |
aroush solutions (opc) private limited |
https://aroushtech.in/ |
Bangalore Urban |
IT consultancy and IT Training company |
fuillfill the technical gaps, make Industry Ready, and implementation of IT. |
Fullfill the technical gaps, make industry ready, and implement IT. |
we use AI to deployment of any IT in a companies, this makes us unique |
Users |
Signups |
all corporate companies, Government entities, Univesities students |
none |
none |
none |
all corporate companies, which is highly paid. |
local companies with no technical team and consultants |
by all social media |
we have our own unique stategy to enter market |
being IT Trainnig company, every technical people need to upgrade thier skill, we are looking forward |
startup |
500000 |
Yes |
i have got email |
to support our society with all IT awareness and generate good reneve |
yes |
No |
Automation layer |
Key Architectural Components
Tool use / Function calling — I can invoke specialized tools (web search, image search, code execution, URL fetching, memory retrieval, LinkedIn search) to gather external data before responding
Persistent code sandbox — A Jupyter-like Python environment where I can run computations, generate charts, and process data across a conversation session
Memory layer — A personalized memory bank that stores your preferences and past interactions (like your role at Aroush Solutions, your IT consulting interests, and your location in Bengaluru) to tailor responses
Citation engine — Every factual claim is grounded and cited inline to a source, ensuring transparency and verifiability |
Unlike simply relying on third-party search engines, Perplexity's infrastructure includes an AI-powered content understanding module that dynamically generates parsing logic to handle the complexity of the open web. This indexing system continuously self-improves through an iterative AI process, powered by real-time signals from millions of user queries processed every hour.
Premium & Professional Data Sources
Beyond the open web, Perplexity's Computer agent can access paywalled professional datasets typically reserved for VC firms and Fortune 500 strategy teams:
PitchBook — full investment lifecycle data on private companies, VC/PE, M&A activity
CB Insights — 8,000+ research reports on the startup and innovation landscape
Statista — market research and statistics
Crunchbase & FactSet — available for Enterprise Pro customers |
Proprietary Data
Unique, high-quality data is the strongest moat in the AI era. If your product generates data through user interactions that continuously improves your AI models, competitors face a compounding disadvantage — they can't catch up just by writing better code. Real-time or domain-specific data (e.g., from a niche vertical like legal, medical, or IT operations) amplifies this further.
Deep Workflow Embedding
AI that sits inside a user's core operational processes becomes infrastructure, not software. When removing the system introduces real risk — audit failures, compliance gaps, or operational breakdowns — it becomes nearly impossible to dislodge. This is why tools like Cursor became defensible: they learned team-specific codebases and conventions, not just general coding patterns
Network Effects & Memory
Every additional user making the product smarter for all users creates a self-reinforcing loop. Persistent memory — where the AI learns a user's or team's context over time — turns a generic tool into a personalized utility that a competitor's fresh install simply cannot replicate. |
Core Performance Metrics
These are the foundational quantitative measures used across most ML models:
Accuracy — Correct predictions ÷ total predictions; simple but misleading on imbalanced datasets
Precision — Of all predicted positives, how many were actually positive (minimizes false alarms)
Recall (Sensitivity) — Of all actual positives, how many did the model catch (minimizes missed cases)
F1 Score — Harmonic mean of precision and recall; best when you need a balance between the two
AUC-ROC — Area Under the ROC Curve; measures the model's ability to distinguish between classes; a perfect model scores 1.0, a random guess scores 0.5 |
What Data Is Collected
Perplexity collects standard account and usage information including your prompts and queries, uploaded files, device info, and interaction logs. Payment details are handled through third-party processors, not stored directly by Perplexity. Perplexity's privacy policy is aligned with major global regulations including GDPR and CCPA, meaning your personal information is not sold or traded.
Regulatory Compliance
Perplexity complies with GDPR, CCPA, HIPAA, and holds SOC 2 Type II certification. GDPR rights such as data access and deletion can be exercised by eligible users. Regular security audits and assessments are conducted to maintain these compliance standards.
Enterprise & API Protections
Enterprise users get the strongest privacy controls available on the platform:
No AI training on enterprise data — Perplexity explicitly guarantees enterprise data is never used to train or fine-tune its models
Third-party provider agreements — Contracts with providers like OpenAI and Anthropic prevent your data from being used in their model training
Workspace-level controls — Enterprise admins can limit data retention, restrict access, and manage sharing permissions
Zero Data Retention for Sonar API — No prompts or responses sent through the API are retained after processing
Infrastructure Security
Access to production data is strictly controlled through authenticated internal tools. Perplexity's Trust Center outlines infrastructure-level protections including unique production database authentication, restricted encryption key access, and production inventory management. Third-party sub-processors and vendors are evaluated through a risk-based security framework. |
Data Privacy & Cybersecurity Regulations
India's Digital Personal Data Protection (DPDP) Act and evolving cybersecurity frameworks create compliance obligations for IT service providers. Failing to adhere to data governance rules can result in fines, legal consequences, or lost business — and these rules are increasingly enforced even when a company is "technically compliant."
Evolving Licensing & Certification Requirements
Regulators can withdraw operating licenses or apply new conditions that adversely affect business value. For IT training businesses specifically, this includes changes to authorized testing center requirements (e.g., Pearson VUE, Prometric) or government-mandated skill certification frameworks in India such as NSDC/NASSCOM standards.
Cross-Jurisdictional Compliance
Regulations vary significantly across regions, and keeping up with changes at every stage of service delivery is a major challenge. If you offer training or consulting services to clients across multiple states or internationally, misalignment with local regulations can block market access.
Trade Policy & Tariff Risks
Geopolitical shifts — such as changes in US-India trade policy — can impact software licensing costs, cloud service pricing, and vendor agreements. In fact, 58% of executives surveyed in 2025 said they were diversifying suppliers in response to trade-related pressures.
Key Mitigation Strategies
Monitor regulatory changes continuously through compliance management tools
Maintain accurate documentation — incomplete certifications or records can block service delivery
Conduct internal audits regularly to identify gaps before regulators do
Train staff on applicable laws, especially around data privacy and cybersecurity
Screen vendors and partners for their own compliance posture, as third-party violations can be attributed to your business |
Scaling an AI system 10x doesn't just stress the obvious components — it creates cascading failures across infrastructure, data pipelines, human processes, and model behavior simultaneously. Here's a breakdown of what actually breaks:
Inference Costs Explode
Inference workloads already consume 10x more electricity than traditional IT operations at current scale. Multiply traffic by 10x and costs scale linearly, quickly becoming a financial drain that outpaces revenue growth. This is often the first wall organizations hit — not a technical failure, but an economic one.
Data & Context Pipelines Collapse
Pre-training data is finite, and at 10x scale you begin hitting the ceiling of available high-quality training data. AI models already require roughly 100,000x more data than humans to learn the same concept, so scaling demand without new data sources degrades model quality. Retrieval pipelines, vector databases, and context windows that work at 1x often have unoptimized query patterns that become catastrophically slow under 10x load.
Infrastructure Bottlenecks: Compute & Networking
GPU interconnect bandwidth becomes the binding constraint — scale-up architectures require at least 10x the bandwidth of current network switches
The latency wall in training (sequential batch processing) caps how much compute you can throw at a problem within a fixed timeframe
Power availability is the first physical constraint to bind, followed by chip manufacturing capacity
The Human Review Bottleneck
One of the most overlooked failure points is human oversight. One engineer cannot review 10x the code output, meaning you also ship 10x the untested assumptions. Bugs don't disappear — they arrive later in production, in a worse context, at higher cost (2 AM on-call incidents). AI as a force multiplier also means mistakes happen 10x faster. |
NO |
Perplexity, Office 365, copilot. |
Iterative Model Tuning
Model improvement is never a one-time event — it follows a structured cycle. The widely used Plan-Do-Check-Act (PDCA) framework guides this:
Plan – Identify the performance gap or opportunity
Do – Implement changes on a small scale
Check – Analyze results against metrics
Act – Scale successful changes and repeat
Hyperparameter tuning and careful feature engineering are central to this loop. Importantly, combining multiple optimization techniques — such as prompt tuning, retrieval-augmented generation (RAG), and targeted fine-tuning — yields better results than relying on any single method.
Continuous Monitoring in Production
Deployment is not the finish line. After going live, models must be monitored against real-world performance, since data distributions shift over time (known as data drift). Metrics should always be correlated with business KPIs, not just model accuracy, because a metric improvement doesn't always translate to business value. Continuous feedback loops and regular audits help catch degradation early. |
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Pc1DLYKKuvhw-AROUSH-SOLUTIONS-PVT-LTD-PITCH-DECK-_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Pc1DLYKKuvhw-AROUSH-SOLUTIONS-PVT-LTD-PITCH-DECK-_compressed.pdf |
https://www.youtube.com/@AROUSHSOLUTIONS |
Not yet, but we want to start |
NA |
Tasneem Firdous |
checked |
| Apr 13, 2026 @ 9:45 PM |
Akash gupta |
farmspherica.innovations@gmail.com |
https://www.linkedin.com/in/akash-gupta-96bb48261/ |
http://www.aavrt.com |
7459805391 |
Akash Gupta - Founder & CEO - 4th year B.Tech BKBIET Pilani student. Leading vision, prototype development, fundraising (₹12.4L grants secured), partnerships (MoU Meerut Udyami). Drives NanoPAW hydroponics commercialization. |
Met at BKBIET Pilani during agritech project in 2024. Working together 18 months on NanoPAW prototype. |
1 |
Technical expertise + execution speed
Strengths: CEERI Pilani professors (nanobubbles/plasma) + BKBIET engineering talent. Secured ₹12.4L grants in 6 months. Built working NanoPAW prototype while planning field trials. Perfect blend of hardcore science + startup hustle. |
Farmspherica innovations pvt. ltd. |
www.aavrt.com |
pilani, jhunjhunu, rajasthan |
NanoPAW hydroponics: 90% water savings + 3X crop growth using plasma water & nanobubbles for arid farms. |
Indian arid farms lose 70-80% irrigation water to evaporation + 40% crops to soil diseases, crippling yields in Rajasthan's desert belts. NanoPAW solves this with plasma-activated water + nanobubbles: cuts water use 90%, disinfects roots naturally, triples growth rates. Proven in prototype trials; ready for Pilani field tests. |
We are currently working to solve the challenges of traditional agriculture and low agricultural productivity in semi-arid and arid regions of India, currently Rajasthan. This is a deep and complex problem which has several causes, geographical, economic and social. To tackle the technical aspects, we are developing a hydroponics system. Hydroponics is a viable technology which is suitable for arid agriculture, however it remains a semi-mature industry due to difficulty of breakthrough and high initial cost to establish farms. However, the core of our innovation lies in the usage of specific nanotechnologies to increase the growth rate of crops, thus increasing yields and reducing the ROI period and allowing farmers to make more money for each investment they make. We enable farmers grow high value crops in semi-arid and arid regions and generate wealth and employment through our use of hydroponics and nanotechnology to revive agriculture in semi-arid and arid regions of India. |
Plasma-activated water + nanobubbles combo delivers 90% water savings + 3X growth unmatched by drip irrigation or chemical disinfectants alone. Defensibility: Patented plasma reactor design by CSIR-CEERI advisor Dr. Udit Narayana Pal (Pilani labs), ₹12.4L grants secured, working prototype with field trial data. Competitors lack this science-startup fusion. |
Idea |
Revenue |
The primary beneficiaries of our idea are farmers in arid and semi arid regions who are struggling to make ends with traditional soil-based agricultural solutions. With Nanopaw hydroponics they can earn higher revenue and sell higher value crops. |
NA |
NA |
NA |
The startup will generate revenue primarily through two distinct channels- B2B and B2C. In our B2B section, we enable farmers conduct high value agriculture by establishing these high productivity NANOPAW hydroponic farms. In our B2C section, we establish the farm and operate it to grow high value crops aimed at premium markets in urban areas and hospitality industry. |
Kryzen Biotech, AgriJoy, Eeeki foods, Rise hydroponics |
Pilani field trials + 10 demo-days to showcase 90% water savings/3X yields to Rajasthan arid farmers, leveraging Pusa Krishi/PIEDS networks and local agri-retailers for B2B trust-building; B2C via www.aavrt.com organic produce sales and hyper-local chaupals/micro-entrepreneurs for word-of-mouth adoption. |
B2B arid farmers (Rajasthan districts) + B2C urban consumers via hydroponic farms and www.aavrt.com. Pilani field trials + 10 demo-days post-prototype; establish farms (15% CAPEX recovery + 5-10% OPEX repairs/SaaS). Phase 2: Scale via Pusa Krishi/PIEDS networks, partner local agri-retailers for distribution |
Scale NanoPAW to every arid farm in Rajasthan, enabling 1M+ farmers with 90% water savings and chemical-free 3X yields.
Build a national hydroponics network—leveraging Pusa Krishi/PIEDS incubation to expand B2B farms + B2C organic produce via www.aavrt.com, becoming the sustainable agritech leader for arid India. |
U01199J2025PTC104957 |
₹12.4 Lakhs in non-dilutive grants raised to date. |
Yes |
We are applying for Avinya to gain mentorship, funding support, and access to advanced infrastructure that can help us scale our NanoPAW Hydroponics solution. The program offers a valuable platform to connect with industry experts, validate our technology, and accelerate our journey towards building a sustainable and impactful agritech startup. |
Secure pilot funding + complete Pilani field trials to validate 90% water savings/3X yields, attract first 10 Rajasthan farmers via demo-days, and build commercial partnerships through programme mentorship/networks. Transition from prototype to revenue-generating arid farm installations by programme end. |
yes |
No |
Automation layer |
Not an AI-powered core product. AI used only as supporting automation layer for image-based plant health monitoring in premium IoT systems." |
Proprietary dataset from Pilani field trials: PAW + nanobubbles performance (90% water savings, 3X growth) across arid crops, IoT sensor logs (pH/EC/DO), and image-based disease patterns.
CSIR-CEERI validated, Rajasthan-specific—competitors lack this field data moat for dosing optimization and yield prediction. |
NA |
NA |
Data stays local on Pilani servers - no cloud. IoT sensors encrypt everything. No personal farmer info stored. CEERI checks security regularly. Follows Indian data laws. |
No major risks—plasma tech & IoT sensors follow standard FSSAI/AGMARK approvals (routine for agritech). DPIIT-recognized status helps.
DPDP easy since no farmer personal data stored—just anonymized trial data on local Pilani servers. |
NA |
No in-house AI/ML team. |
Owned proprietary datasets from Pilani prototype trials - PAW + nanobubble performance data, IoT sensor logs (pH/EC/DO), crop growth metrics.
No external licensed data used - all Farmspherica-generated and controlled. |
NA |
Primarily India (Rajasthan arid farms first), with global expansion potential to other water-scarce regions.
Focus: National hydroponics network via Pusa Krishi/PIEDS scaling. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/H0Y2ydVND9Xl-NanoPaw-Farmspherica-Pitch.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/H0Y2ydVND9Xl-NanoPaw-Farmspherica-Pitch.pptx |
https://youtube.com/shorts/gdl1oJJjZx8?si=byF8jELrVhE8vwZY |
Yes, mission-driven startup focused on sustainable farming for arid India.
Solving 70-80% irrigation water loss + 40% crop failures in Rajasthan deserts through NanoPAW—90% water savings, chemical-free 3X growth for 1M+ small farmers. |
NO |
Pusa Krishi incubator / DPIIT network / CEERI Pilani contacts. |
checked |
| Apr 13, 2026 @ 3:06 PM |
Orion AI |
orionaiacademy@gmail.com |
https://www.linkedin.com/in/priyaroshini?utm_source=share_via&utm_content=profile&utm_medium=member_android |
https://www.orionai.co.in/ |
+91 93631 97584 |
Founder |
Nil |
2 |
We share the same mission but bring opposite strengths. That mix of tech + business lets us both build and scale.Trust and speed. We know how each other thinks, so we can make hard calls quickly and stay aligned under pressure. |
Orion AI |
https://www.orionai.co.in/ |
Chennai |
Orion AI is an AI-powered platform that helps businesses automate workflows and make data-driven decisions. |
Startups and small businesses struggle with manual processes, scattered data, and lack of access to affordable AI tools. |
Orion AI provides a simple, scalable platform that automates operations and delivers actionable insights using AI. |
It combines automation, analytics, and decision intelligence in one platform, designed specifically for non-technical users with affordability and scalability. |
MVP |
Revenue |
Startups, small and medium enterprises (SMEs), and growing businesses seeking affordable AI solutions for automation and data-driven decision-making. |
₹1.5–2 lakh crore (Indian AI and digital transformation market) |
₹30,000–50,000 crore (AI solutions market for startups and SMEs in India) |
₹100–200 crore (initial target market capture in early growth stage) |
Subscription-based (SaaS) model with monthly/yearly plans, along with additional revenue from customization and enterprise solutions. |
LatentView Analytics, Tiger Analytics, Cimplify AI, Rubixe, Rytsense Technologies |
Through digital marketing, startup communities, partnerships, networking events, and direct outreach to businesses. |
Focus on early adoption through startups and SMEs, pilot testing, building strong customer relationships, and scaling through partnerships and online channels. |
To become a leading AI platform enabling businesses globally to automate operations and make intelligent, data-driven decisions. |
Incorporated |
NA |
Yes |
We are applying to Avinya to gain mentorship, technical guidance, and access to a strong startup ecosystem that can help us refine our AI solution and scale effectively. |
We aim to complete MVP development, validate our product with real users, achieve early traction, and prepare for scaling and funding. |
Yes |
Yes |
Core engine |
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/v4eQb5e9zqAN-Orioverse-by-Orion-AI-pvt-ltd-compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/v4eQb5e9zqAN-Orioverse-by-Orion-AI-pvt-ltd-compressed.pdf |
http://NA |
Yes, Orion AI is a mission-driven startup focused on making artificial intelligence accessible and affordable for startups and small businesses. Our goal is to empower businesses to improve efficiency, reduce costs, and make data-driven decisions, contributing to economic growth and digital transformation. |
NA |
NA |
checked |
| Apr 13, 2026 @ 1:04 PM |
Manik Sehgal and Surbhi Sehgal |
manik.sehgal@raasakarts.com |
https://www.linkedin.com/in/manik-sehgal/%20 |
https://raasakarts.com/ |
9811259603 |
CEO |
We have been married for 13 years and have been working together as entrepreneurial partners for several years. Prior to Raasa Karts, we ran an artificial jewellery business, where we successfully operated an eBay store for nearly three years, gaining hands-on experience in e-commerce, customer handling, and operations.
At Raasa Karts, Manik Sehgal (CEO) leads overall strategy and technology, while Surbhi Sehgal heads cataloging, vendor onboarding. |
2 |
Our biggest strength is our complementary skill set and deep on-ground execution. While Manik drives strategy and technology, Surbhi leads vendor onboarding and operations, allowing us to move fast while staying closely connected to the real needs of street vendors. |
Raasa Karts India Private Limited |
https://raasakarts.com/ |
Ghaziabad |
Digitizing India’s Street Food Ecosystem through E-Commerce and Hyperlocal Delivery |
Raasa Karts is India’s first street food–focused e-commerce platform, connecting local street food vendors with online customers. India has nearly 1 million street food vendors generating over USD 40 billion (₹3.3 lakh crore) in annual economic activity, yet 90% remain unorganized, offline, and excluded from the digital economy. They struggle with no digital presence, unstable income, lack of dignity or recognition, and limited access to licenses and finance. By digitizing these vendors, Raasa Karts aims to empower them to sell online like big restaurants do—ensuring fair income, daily payouts, and financial inclusion—while formalizing a massive hidden economy. Even a 5% GST inclusion from this sector could add nearly USD 2 billion (₹16,000 crore) annually to India’s tax revenue, proving that empowering street vendors is not only a social cause but a direct boost to the national economy. What we are building has never been done before in India—a mission to turn street food vendors into the next generation of digital entrepreneurs of Bharat. |
Raasa Karts is India’s first street food delivery and vendor empowerment platform, officially launched in April 2024. In just one year, we’ve built a working ecosystem for India’s local street-food sellers — helping them go digital, increase sales, and earn with dignity.
The platform includes live mobile apps for both customers and vendors.
Through the Raasa Karts app, customers can discover and order food from verified street vendors at the same price as offline, while vendors receive complete digital support — we create digital menus, enable UPI payments, assist in licensing, and integrate with delivery partners like Ola and Rapido for last-mile fulfillment.
Our platform is fully functional, tested by real users, and generating daily transactions.
So far, we have onboarded 4,000+ vendors, completed over 40,000 orders, and processed ₹50+ lakh in total order value, validating that the model works and the market is ready.
Currently at the MVP (Minimum Viable Product) stage, we are now entering the stability and scale phase — focusing on expanding to new cities, increasing daily orders, and strengthening our field and catalog teams.
As we scale, Raasa Karts will integrate AI-driven automation to make vendor onboarding and operations faster and smarter — our goal is to enable a new vendor to go live within 10 minutes, unlocking rapid expansion and nationwide inclusion. |
Raasa Karts delivers authentic street food at offline prices while empowering local vendors—especially women—with zero commission, daily payouts, digital visibility, and ONDC-enabled delivery. Unlike traditional food apps that benefit restaurants, Raasa Karts transforms unorganized street vendors into digital micro-entrepreneurs, creating income stability for them and affordable access for customers. It’s the only platform that combines low-cost food access + social impact + tech-driven vendor empowerment, making it uniquely valuable to both users and India’s grassroots economy. |
Revenue |
Users, Revenue, Signups |
Our target customers are urban consumers in Tier 1, 2, and 3 cities who want affordable, authentic, and hyperlocal street food, along with street-food vendors who need digital visibility, stable income, and access to online customers. |
50 Billion Dollar |
20 Billion Dollar |
2 Billion Dollar |
Currently, Raasa Karts generates revenue through a small platform fee charged to customers on each order. As we scale, we will introduce a vendor-side subscription/SaaS model, where vendors pay a nominal fee of ₹10–₹20 per order based on order value. This ensures affordability for vendors while creating a sustainable and scalable revenue stream for the platform without heavy commissions. |
Raasa Karts operates in the broader food delivery and digital commerce space, but with a unique focus on street food vendors. Our primary competitors include food delivery platforms such as Zomato and Swiggy, which serve urban consumers but mainly cater to organized restaurants and are not optimized for informal street vendors |
We acquire customers primarily through hyperlocal social media campaigns, targeting specific neighborhoods and high-density clusters where vendors are active. We use geo-targeted ads. |
Raasa Karts follows a hyperlocal, cluster-based go-to-market strategy. We first build strong supply by onboarding street food vendors in a focused area through on-ground teams, ensuring high availability and variety. Once a cluster is ready, we drive demand through targeted social media campaigns. |
Our long-term vision is to become the one-stop digital and physical infrastructure for India’s entire street food ecosystem—covering online ordering, offline cluster markets, corporate catering, vendor supplies, and advertising—so that whenever anyone thinks of street food, they think of Raasa Karts. |
Yes |
1.5 crore |
Yes |
We are applying to Avinya to access strategic funding and ecosystem support to accelerate our growth. Avinya’s focus on impact-driven ventures aligns with our mission to formalize India’s street food economy, and their support will help us strengthen operations, expand faster, and build a sustainable, scalable business. |
We want to leverage Avinya’s hybrid programme (virtual + incubation) to build and scale Raasa Karts in a structured and accelerated way. |
Yes |
Yes |
Automation layer |
Our AI architecture is designed as a lightweight, modular layer on top of the Raasa Karts platform, focused on vendor onboarding, catalog digitization, and operational intelligence.
At the input layer, we collect unstructured vendor data such as menu images, text notes, and catalog photos through our field teams and vendor workflows. This data is processed through AI-powered extraction and classification modules that convert raw menu inputs into structured digital catalogs with item names, categories, tags, and pricing suggestions.
At the application layer, these outputs flow into our backend systems for catalog creation, search optimization, and vendor listing management. We also plan to use AI models for demand prediction, item popularity analysis, and customer-side recommendations based on location, time, and ordering patterns.
The architecture is supported by our core product stack, with AI acting as an intelligence layer rather than a standalone system. Today, it is mainly a supporting feature to reduce manual effort and onboarding time; over time, it will become a stronger scale-enabler across onboarding, discovery, and decision-making. |
Raasa Karts is building a strong proprietary data advantage around India’s street food ecosystem, a segment that is largely undocumented and offline.
We capture unique, ground-level data that is not available on any other platform, including vendor menus, pricing at offline parity, location-specific demand patterns, peak ordering times, and customer preferences for local street food. Since we onboard vendors directly through field operations, we generate high-quality, structured data from unorganized sources.
Over time, this creates a rich dataset of hyperlocal consumption behavior—what people eat, where they eat, and at what price points—across different neighborhoods. This data enables better recommendations, targeted promotions, demand forecasting, and efficient vendor placement.
Additionally, as we scale vendor onboarding and transactions, our dataset becomes increasingly difficult to replicate, creating a defensible moat in the street food segment. |
Our AI component is defensible because it is built on proprietary, ground-level data from India’s street food ecosystem—an unstructured and largely untapped segment. Unlike generic platforms, we are continuously collecting unique data on vendor menus, pricing, demand patterns, and hyperlocal consumption behavior through our own onboarding and transactions.
We are also building in-house AI-driven tools tailored specifically for this segment, such as automated menu digitization, catalog standardization, and onboarding workflows. These tools are designed for low digital literacy users and improve over time with our data, making them difficult to replicate.
As our network grows, the combination of proprietary data + in-house AI tools creates a strong feedback loop, where better data leads to better models, and better models improve operations and user experience—building a scalable and defensible advantage. |
We evaluate AI model performance and reliability using a mix of accuracy, operational efficiency, and business impact metrics.
For onboarding and menu digitization, we track data extraction accuracy (correct item names, prices, categories) and manual correction rate, ensuring outputs are reliable for real-world use. We also measure time to onboard a vendor, aiming to reduce it significantly through automation.
For discovery and recommendations, we track click-through rate (CTR), conversion rate, and order uplift, ensuring that suggestions are relevant and useful to customers.
On the reliability side, we monitor error rates, fallback/manual override frequency, and system uptime to ensure consistent performance. We also use continuous feedback from field teams and vendors to validate outputs and improve models.
Overall, our focus is on practical performance—how well AI reduces manual effort, improves accuracy, and drives real business outcomes. |
We take data privacy, compliance, and security seriously across our platform. Raasa Karts operates on secure, cloud-based infrastructure hosted on AWS, with data stored on well-protected and encrypted servers aligned with Indian data security standards.
All sensitive data, including user and vendor information, is encrypted in transit and at rest. Access to data is strictly controlled through role-based permissions, ensuring only authorized personnel can access critical information. We follow best practices for secure authentication, API protection, and regular monitoring to prevent unauthorized access.
We also ensure compliance with applicable data protection guidelines and maintain transparency in how data is collected and used. As we scale, we are committed to continuously strengthening our security frameworks and adopting industry standards to safeguard our users and vendors. |
Yes, there are certain regulatory considerations, but they are manageable and aligned with our model.
Street food vendors often operate in an informal setup, so requirements such as FSSAI licensing, local municipal permissions, and hygiene compliance can impact onboarding and operations. Additionally, evolving regulations around food safety, e-commerce, and digital transactions may require ongoing compliance.
We address this by actively supporting vendors with basic guidance on licensing, hygiene standards, and documentation. Our model is also aligned with government initiatives like ONDC, which promotes inclusion of small businesses into the digital economy.
Overall, while regulatory factors exist, they also create an opportunity for Raasa Karts to play a role in formalizing the sector and enabling compliant, scalable growth. |
Challenge would be human validation capacity. Today, AI works as a support layer, but at 10x scale we would need stronger review workflows, confidence scoring, and exception handling so low-confidence outputs are routed for manual checks instead of affecting live catalogs. |
Currently, we do not have a dedicated in-house AI/ML team. Our technology development is led by our CTO, along with an in-house tech resource who is actively working on building AI-driven features such as catalog digitization and onboarding automation.
We are leveraging existing AI tools and frameworks while developing customized, in-house solutions tailored to the street food ecosystem. As we scale, we plan to strengthen our AI/ML capabilities by building a focused team to enhance automation, data intelligence, and platform efficiency. |
We primarily use proprietary, first-party datasets generated through our platform operations. This includes vendor menus, pricing, catalog images, location data, order transactions, and customer behavior patterns collected during onboarding and daily usage.
Most of our data is owned by us, as it is created through our direct vendor onboarding, field operations, and platform activity. In addition, we may use publicly available or pre-trained AI models (for tasks like image and text processing), but we do not rely on any sensitive third-party licensed datasets for core operations.
As we scale, our proprietary dataset continues to grow, becoming more structured and valuable, and forms the foundation for improving our AI-driven features and platform intelligence. |
We improve model performance through a continuous feedback and iteration loop built on real platform data.
First, we leverage our growing proprietary dataset—vendor menus, images, orders, and user behavior—to regularly retrain and refine models. As more vendors are onboarded and transactions increase, the system becomes more accurate and context-aware.
Second, we use human-in-the-loop validation, where low-confidence outputs (such as menu extraction or categorization) are reviewed and corrected by our team. These corrections are fed back into the system to improve future predictions.
Third, we track key performance metrics such as accuracy, error rates, and manual correction rates, and continuously optimize models based on these insights.
Additionally, we iterate on our in-house tools and workflows, ensuring models are adapted specifically to the street food ecosystem. Over time, this creates a strong feedback loop where better data leads to better models, and better models improve overall platform performance. |
We are currently focused on building for India, given the massive and underserved street food market here. Our model is designed specifically for India’s informal food economy, low digital literacy vendors, and hyperlocal consumption patterns.
In the long term, we see strong potential to expand to global markets with similar informal food ecosystems, but our current priority is to scale deeply within India before exploring international opportunities. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/PYsXnbyYcVOq-Raasa-Karts_Feb-2026.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/PYsXnbyYcVOq-Raasa-Karts_Feb-2026.pdf |
https://drive.google.com/file/d/1sdrrA60PEJBt_sV-Gnq28oFcxT8PRpVR/view?usp=sharing |
Raasa Karts is both a mission-driven and impact-focused startup. Our mission is to digitize and formalize India’s street food economy by bringing underserved vendors online. We aim to improve income stability, enable financial inclusion, and create dignified livelihoods for street food vendors.
Our impact is measurable not only through the number of vendors onboarded, their income growth, and access to digital tools and payments, but also through customers who benefit from convenient access to authentic, affordable local food at the same price as offline. By serving both vendors and consumers, we are building a scalable ecosystem that creates economic and social value on both sides. |
Yes, our founding team includes a woman co-founder actively leading operations and growth, contributing to greater gender diversity in the startup ecosystem. |
LinkedIn |
checked |
| Apr 13, 2026 @ 12:10 PM |
Abhimanyu Rathi & Yogendra Panchal |
ar@renewcred.com |
https://www.linkedin.com/in/abhimanyurathi/ |
https://www.renewcred.com |
+919925629000 |
Abhimanyu Rathi is the Founder and Chief Executive Officer of RenewCred, with a 16-years background as a Climate, Water and Energy Scientist. A serial entrepreneur with a proven track record, Abhimanyu has successfully led multiple ventures, building two communities and non-profits. His work has earned him the prestigious Mohammed bin Rashid Al Maktoum Global Water Award. His previous experience includes roles at build3, SLII, Google, and IICSE. Abhimanyu brings deep expertise in sustainable technologies and science-driven innovation to his leadership at RenewCred. Yogendra Panchal, Co-Founder & VP of Engineering at RenewCred, is a seasoned software professional with extensive expertise in AI/ML and blockchain technologies. He has a distinguished track record of delivering critical projects for organizations like ISRO and Institute for Plasma Research. With over 8 years of experience, Yogendra plays a vital role in driving RenewCred's technological innovations. |
We’ve known each other for close to 10 years now-we started our careers in the same office, which is where we first connected. Over time, we kept collaborating on different technology development projects, so the working relationship built quite naturally. About two years ago, we decided to come together more formally and started working on RenewCred, building on that shared experience and trust. |
2 |
Our biggest strength is how clearly our roles complement each other while still overlapping where it matters. Abhimanyu leads on product, operations, and finance, ensuring execution and business alignment, while Yogendra drives the technology side along with supporting sales. Marketing is something we both actively contribute to. |
Net Zero Initiative Private Limited (RenewCred) |
https://www.renewcred.com |
Bengaluru, Karnataka, India |
Global South's first carbon credits standard and registry - we measure, certify, issue and manage high quality carbon credits aimed at industrial decarbonisation. |
Climate change, driven by industrial emissions, remains an urgent global threat. The voluntary carbon market meant to mitigate this-is plagued by low-quality offsets, poor transparency, and unreliable verification. Manual, corruption-prone verification processes and costly registry commissions (up to 30%) make it inaccessible for smaller project developers. As a result, credible, tech-based carbon credits remain scarce. Moreover, the high capital requirements for sustainable hardware systems limit adoption. Transitioning to project-based financing and de-risked climate investments could unlock scalable impact, but this gap remains largely unaddressed. |
RenewCred is building the governance and data backbone for voluntary carbon markets through a technology-first approach that ensures transparency, reliability, and accessibility. Our solution offers high-quality, non–nature-based carbon offsets that are digitally verified and fully traceable via a transparent registry and standard. Using advanced digital MRV (Monitoring, Reporting, and Verification) systems powered by IoT, proprietary machine learning, and blockchain, we automate the entire data collection and verification process-reducing fraud, human error, and verification time. Each carbon credit is backed by rigorous scientific validation and accompanied by financial performance ratings to attract green financing and de-risk climate investments. Through centralized dashboards, RenewCred enables real-time project insights for both buyers and developers, bridging the trust gap that has long hindered voluntary carbon markets. In essence, we are transforming carbon offsetting into a transparent, tech-enabled, and scalable climate financing ecosystem. |
RenewCred combines deep market expertise, scientific credibility, and cutting-edge technology. We offer a blockchain-enabled, science-backed registry to ensure trusted carbon credits. Our dMRV dashboard provides live verified project data, reducing verification cycles from 7 months to 7 days. Banks gain climate-linked insights for better underwriting, while our team brings 40 years of combined experience and backing from 90+ climate scientists (80% PhDs), ensuring unmatched founder-product and founder-market fit. |
Product-Market Fit |
Pilots |
Our target customers are two key groups. First, corporate carbon credit buyers such as Chief Sustainability Officers and ESG Heads in sectors like manufacturing, energy, and transport, who need high-quality, verifiable credits they can trust. Second, small-to-mid-sized project developers, especially MSMEs in sectors like EV, pharma, and agriculture, who generate carbon reductions but lack access to affordable verification, registration, and market linkages. |
$3.0 billion (₹25000 Crores) per year by 2035 |
$910 million (₹7500 Crores) per year by 2030 |
$120 million (₹1000 Crores) per year by 2027 |
RenewCred operates on a governance plus SaaS-based B2B model, generating revenue from carbon credit transactions, green financing solutions for banks, and registry services. Buyers of carbon credits pay a 10% transaction fee, while suppliers are not charged, ensuring incentives remain aligned with maintaining the highest market standards. Banks pay a fixed one-time fee for access to project data for up to 60 months. Using proprietary digital MRV technology, we provide continuously verified carbon offsets with advanced financial ratings. This is a high-LTV business, with customers typically engaged for at least 10 years-the lifecycle of a carbon sequestration project. Each project generates a minimum of 12,600 credits per year at $24 per credit, resulting in a minimum CLTV of $300,000 per project over 10 years, excluding additional revenue from bank data services or carbon price increases. We also offer an engineering innovations playbook and technical support for MSMEs transitioning to sustainable operations, backed by 250+ climate scientists, ensuring high-quality carbon credits and robust project execution. |
Verra and Gold Standard are well-known registries issuing carbon credits, but they rely on periodic manual reporting, focus primarily on nature-based projects, and have slower, less transparent processes. Verra, which holds 72% of the market, has faced credibility concerns over some projects. Puro.earth offers a blockchain-powered carbon removal marketplace but focuses narrowly on carbon removal rather than a broad range of credits. Climeworks innovates in CO₂ capture technology but operates in a geographically limited, high-cost niche. CarbonX and Patch provide trading platforms but lack continuous MRV, integrated financial ratings, and deep transparency. Sylvera and BeZero offer carbon ratings but no registry or dMRV services. Isometric is a new registry player but focuses only on carbon removal projects. RenewCred differentiates itself through an integrated, technology-driven, and verifiable approach across multiple carbon markets. |
We acquire customers through a mix of direct partnerships and targeted outreach. On the supply side, we work closely with project developers (EV fleets, biochar, MSMEs) through industry networks, referrals, and on-ground partnerships to onboard projects. On the demand side, we engage corporate buyers through direct outreach, existing networks, and sustainability-focused platforms. Accelerators, grants, and ecosystem partnerships have also been a strong channel for introductions, helping us build early traction on both sides of the market. |
RenewCred’s Go-To-Market (GTM) strategy targets both buyers and suppliers to drive growth in the carbon credit and green finance markets. On the buyer side, we focus on sector-agnostic global corporations, using tools like the Carbon Disclosure Project and Net Zero Tracker to identify companies with decarbonization goals, particularly the Top-150 corporations with active net-zero commitments and significant engagement in the Voluntary Carbon Markets (VCM). Our product awareness is built through white papers and LinkedIn Sales Navigator, highlighting our methodologies in EV Fleets and Biochar, which align with emerging sustainable sectors. We offer these buyers high-quality, verifiable carbon credits, leveraging digital MRV technology for transparency, traceability, and reduction of Scope 3 emissions. On the supplier side, we target MSMEs and project developers in tech-based sectors, focusing on those transitioning to sustainable practices. We provide a zero-cost, tech-driven platform for carbon credit registration and verification, reducing entry barriers and simplifying the process. Our “engineering innovations playbook” and community of 250 scientists offer technical expertise, helping suppliers generate high-quality carbon credits. Through partnerships with banks, we also facilitate green financing, enabling suppliers to scale their projects and access capital. This dual approach ensures that RenewCred builds trust, credibility, and efficiency across both buyers and suppliers in the carbon credit ecosystem. |
At RenewCred, our BHAG (Big Hairy Audacious Goal) is to avoid, reduce, or remove 2 gigatons of CO2e by 2040. We aim to achieve this impact by facilitating the generation and trading of high-quality, verifiable carbon credits, enabling industries to meet their climate targets and contribute to a sustainable, net-zero future. |
Incorporated, Registered as Net Zero Initiative Private Limited (RenewCred). |
₹4.13 Crores ($458.9K) raised |
Yes |
We’re applying to Avinya at a stage where we’ve built the core product and validated early demand, but now need the right ecosystem to scale faster and more sharply. What stands out to us is the combination of hands-on mentorship, strong operator networks, and access to capital-all at a point where these inputs can directly shape how we grow. |
We want to use Avinya to sharpen our product and scale it faster by learning from founders and mentors who’ve already built in this space. Access to IIT alumni, domain experts, and the wider network can help us close the right partnerships and avoid early mistakes, while the funding support will go into strengthening our dMRV system and moving faster on deployments. |
Virtual |
Yes |
Core engine |
Our AI architecture is built around our dMRV system, where we collect real-time data directly from climate assets such as EV fleets and other emission-reducing infrastructure through IoT integrations and API calls. Each asset continuously streams data, which forms the base layer of our system.
On top of this, we’ve built an AI-led fraud detection layer that analyzes data patterns to identify inconsistencies, manipulation, or improper device setups. This ensures that only clean, reliable data moves forward. Once the data passes these checks, it is processed using our scientific algorithms to calculate carbon credits, which are then issued on-chain through our registry. The overall system is designed to ensure that only high-quality, verifiable, and trustworthy carbon credits are generated. |
Our core advantage comes from being able to collect and process real-time, ground-level data directly from climate assets like EV fleets and other emission-reducing systems. This is not static or reported data—we capture continuous operational data through IoT integrations, which gives us a much deeper and more reliable dataset than what is typically available in carbon markets.
Over time, this creates a strong data flywheel. We’re not just collecting data, we’re cleaning it, structuring it, and building predictive layers on top of it. This allows us to move from just measuring emissions to forecasting performance and credit generation. As this dataset grows across projects and sectors, it becomes increasingly difficult to replicate, giving us a long-term data advantage. |
Our AI is not a standalone layer-it works in tandem with scientific methodologies. The models are built on top of algorithms designed with a network of 90+ scientists across multiple countries, which ensures that outputs are not just data-driven but also scientifically valid. This combination is hard to replicate because it requires both deep domain expertise and strong technical execution.
Second, we have a strong data advantage. We process high-frequency, real-time data directly from assets, sometimes at very granular intervals. Over time, this creates a dataset that improves model accuracy and is difficult for new entrants to match.
Third, our fraud detection architecture is tightly integrated into the system. It continuously checks for inconsistencies or manipulation in data before credits are issued, making trust a built-in feature rather than an afterthought. |
We evaluate model performance by benchmarking it against ground-truth data collected directly from assets. This gives us a real-world reference to check accuracy rather than relying only on simulated outputs. We also cross-verify model outputs with alternate AI models and external datasets where available, to ensure consistency.
In addition, we validate results through our scientific network, where domain experts review whether outputs align with expected physical and environmental behavior. Reliability is measured through consistency of outputs over time, anomaly detection accuracy, and the system’s ability to flag incorrect or manipulated data without impacting valid data flows. |
We handle data privacy and security at both the system and compliance level. All data coming from assets is encrypted in transit and at rest, with strict access controls to ensure only authorized systems and users can interact with it. Since we deal with operational and sensitive project data, maintaining integrity and preventing tampering is a core part of the architecture.
On the compliance side, we align with global standards such as ISO 27001, GDPR, and SOC 2. This ensures that data is collected, stored, and processed in a way that meets regulatory requirements, while also building trust with project developers, enterprises, and institutional partners. |
No |
If we scale 10x, two things will get stressed first. One is data infrastructure-we’ll be handling a much larger volume of high-frequency data coming from assets, so storage, processing, and real-time handling need to keep up. The second is on the blockchain side-issuing credits at scale and at speed can become a bottleneck if the system isn’t optimized for higher throughput. Those are the areas we’re already thinking about as we scale. |
Yes, we have an in-house AI/ML team. Kaushik Gupta, our Principal Engineer, leads the core technology stack, working across backend systems, data infrastructure, and AI-driven product development. He’s built scalable systems both in startups and at organizations like Ericsson, and has experience developing platforms at scale.
Yogendra Panchal, Co-Founder & VP of Engineering, brings deep expertise in AI/ML and blockchain. He has worked on critical systems with organizations like ISRO and the Institute for Plasma Research, and drives the overall technical architecture at RenewCred. Along with them, there are 6 other engineers working across AI/ML, data systems, and product development. |
We primarily generate our own datasets by collecting real-time data directly from climate assets through our dMRV system. This data is then processed and structured based on our carbon models, so over time we’re building proprietary datasets that are owned by us. |
We improve the model continuously using the data coming from real projects on the ground. As more assets get onboarded, we get better, high-frequency data, which helps us refine accuracy and detect patterns more reliably over time.
At the same time, our models are not just data-driven-they are built alongside scientific methodologies. We keep validating and updating them with inputs from our science network to make sure the outputs stay aligned with real-world behavior. So it’s a mix of better data over time and constant scientific validation that keeps improving performance. |
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/bFCUg3sncyGV-RenewCred-Pitch-Deck-January-2026.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/bFCUg3sncyGV-RenewCred-Pitch-Deck-January-2026.pdf |
https://youtu.be/JZlHBmRkGjU |
We are building impact-focused startup. At RenewCred, our goal is to make carbon markets actually work so that real climate action gets funded at scale. We focus on enabling the generation and trading of high-quality, verifiable carbon credits, so industries can meet their climate targets with confidence. Our long-term goal is clear: we aim to avoid, reduce, or remove 2 gigatons of CO2e by 2040. We plan to do this by building reliable infrastructure that connects credible projects with serious buyers, making it easier for capital to flow into real emission reduction and removal efforts. |
NA |
NA |
checked |
| Apr 13, 2026 @ 9:50 AM |
Akshay Kumar Srivastava & Neha Srivastava |
office@lcbfertilizers.com |
https://www.linkedin.com/in/akshay-shrivastav-b82506161/ |
http://www.lcbfertilizers.com |
7905987520 |
I am the founder of LCB Fertilizers Private Limited, where I lead the company’s overall vision, strategy, and operations. My role involves overseeing product development, managing business growth, and building strong relationships with farmers and partners. I come from a background that is closely connected to agriculture and rural development, which has given me a deep understanding of farmers’ needs and challenges. With this experience, I aim to provide high-quality, affordable fertilizer solutions that improve soil health and increase crop productivity. Through LCB Fertilizers, I am committed to supporting sustainable farming practices and contributing to the growth of the agricultural sector. |
As the founder of LCB Fertilizers Private Limited, I brought together the core team through a combination of professional networks, local connections, and shared interest in agriculture and rural development. Some of us have known each other from earlier work experiences and community interactions, while others joined through referrals based on their expertise and commitment to the sector.
We have been working together for the past few years, building the company from the ground up. Over this time, we have developed strong trust, clear communication, and a shared vision to support farmers with better and more sustainable fertilizer solutions. |
2 |
Our biggest strength as a team is our shared commitment to solving real problems faced by farmers, combined with our complementary skills. Each member brings different expertise—whether in agriculture, operations, or business development—but we are all aligned with the same vision.
We have built a culture of trust, open communication, and quick decision-making, which helps us adapt to challenges and move forward efficiently. Being closely connected to the ground realities of the agricultural sector also allows us to create practical and impactful solutions. |
LCB Fertilizers Private Limited |
www.lcbfertilizers.com |
Gorakhpur |
LCB Fertilizers is a deep-tech agri startup that converts agricultural waste into customized, high-performance bio-nano fertilizers using microbial, nano, and polymer technologies to improve soil health and farmer profitability. |
We are solving soil degradation and low farm productivity by turning underutilized agricultural waste into high-quality, sustainable fertilizers that improve soil health and farmer income. |
At LCB Fertilizers, we have developed a technology-driven solution that converts agricultural waste such as slurry, dung, and biomass into high-performance, crop-specific fertilizers. Our approach combines microbial science, nanotechnology, and polymer technology to create fertilizers that not only provide essential nutrients but also improve soil structure, water retention, and overall soil health.
We use a consortium of beneficial microorganisms that naturally generate key nutrients like nitrogen, phosphorus, and potassium in the soil. Along with this, our nano-enhanced components improve nutrient efficiency and microbial survival, while our super-absorbent polymer helps retain moisture in the soil for longer periods, reducing the need for frequent irrigation.
What makes our solution unique is its customization—we analyze soil, crop, and climatic conditions to create targeted formulations that deliver better results for farmers. In addition, we enable decentralized production by setting up low-cost bio-manufacturing units, allowing farmers, FPOs, and local organizations to convert locally available waste into valuable fertilizer products.
Overall, our solution creates a sustainable “waste-to-wealth” ecosystem that reduces input costs, increases yield, improves soil health, and supports long-term agricultural sustainability. |
Our uniqueness lies in combining microbial, nano, and polymer technologies into one integrated, crop-specific solution. Unlike traditional fertilizers, we customize our products based on soil, crop, and climate conditions for better results. We also convert local agricultural waste into high-value fertilizers through our proprietary processes and decentralized manufacturing model, making our solution scalable, cost-effective, and difficult to replicate while delivering long-term benefits to farmers. |
Product-Market Fit |
Revenue |
Our target customers include farmers, FPOs, and agri-based industries like sugar mills, bioenergy plants, and gaushalas, along with government and global markets focused on sustainable agriculture. |
770 Billion USD |
35–50 Billion USD |
150–250 Million USD |
We generate revenue through sales of crop-specific fertilizers, microbial concentrates, setting up decentralized manufacturing units, and technology licensing/co-development with partners. |
Our main competitors include traditional chemical fertilizer companies, organic fertilizer producers, and emerging biofertilizer and agri-biotech startups. |
We acquire customers through on-ground farmer engagement, partnerships with FPOs and agri-organizations, collaborations with industries, and government and institutional networks. |
e follow a hybrid go-to-market strategy by combining direct farmer outreach with partnerships through FPOs, agri-industries, and decentralized manufacturing units to enable local production, distribution, ande follow a hybrid go-to-market strategy by combining direct farmer outreach with partnerships through FPOs, agri-industries, and decentralized manufacturing units to enable local production, distribution, and |
Our long-term vision is to build a global, technology-driven agricultural ecosystem that transforms how fertilizers are produced and used. We aim to lead the shift from chemical-dependent farming to sustainable, biology-driven solutions that restore soil health, improve farmer income, and reduce environmental impact.
We envision scaling our decentralized “waste-to-wealth” model across India and internationally, enabling thousands of local production units that convert agricultural and industrial waste into high-value fertilizers. By integrating advanced microbial, nano, and data-driven technologies, we want to create a system where farming becomes more efficient, climate-resilient, and profitable for millions of farmers.
Ultimately, our goal is to position LCB Fertilizers as a global leader in agri-biotech innovation, driving sustainable agriculture at scale while contributing to food security and environmental sustainability. |
Incorporated as a Private Limited Company on 18 November 2020. |
e are a bootstrapped company and have not raised any external funding so far. |
No |
We are applying to Avinya to gain strong market connections, access valuable mentorship, and learn from experienced founders and experts. We also see it as a great platform to grow, collaborate, and enjoy the journey of building together. |
We aim to build strong market connections, refine our strategy, and prepare for scalable growth and future fundraising. |
Yes, we are open to virtual participation. |
Yes |
Supporting feature |
Our AI layer acts as a supporting system that analyzes soil, crop, and environmental data to recommend customized fertilizer formulations. We use data inputs like soil parameters, location-based insights, and crop requirements to generate actionable recommendations.
The system is designed to assist our core biological technology by improving precision, optimizing input usage, and enabling better decision-making for farmers and production units, while remaining scalable and easy to integrate with our decentralized model. |
Consider yourself as founder of lcb fertilizers private Limited mai ek form fill kar raha hun usme question ha usko jawab dena mujhe in humanized way in english 1-Role(s) and Background
Here’s a natural, humanized answer you can use:
Role(s) and Background:
I am the founder of LCB Fertilizers Private Limited, where I lead the company’s overall vision, strategy, and operations. My role involves overseeing product development, managing business growth, and building strong relationships with farmers and partners.
I come from a background that is closely connected to agriculture and rural development, which has given me a deep understanding of farmers’ needs and challenges. With this experience, I aim to provide high-quality, affordable fertilizer solutions that improve soil health and increase crop productivity.
Through LCB Fertilizers, I am committed to supporting sustainable farming practices and contributing to the growth of the agricultural sector.
How did you meet and how long have you worked together?
Here’s a natural, humanized answer you can use:
How did you meet and how long have you worked together?
As the founder of LCB Fertilizers Private Limited, I brought together the core team through a combination of professional networks, local connections, and shared interest in agriculture and rural development. Some of us have known each other from earlier work experiences and community interactions, while others joined through referrals based on their expertise and commitment to the sector.
We have been working together for the past few years, building the company from the ground up. Over this time, we have developed strong trust, clear communication, and a shared vision to support farmers with better and more sustainable fertilizer solutions.
What’s your biggest strength as a team?
Here’s a natural, humanized answer you can use:
What’s your biggest strength as a team?
Our biggest strength as a team is our shared commitment to solving real problems faced by farmers, combined with our complementary skills. Each member brings different expertise—whether in agriculture, operations, or business development—but we are all aligned with the same vision.
We have built a culture of trust, open communication, and quick decision-making, which helps us adapt to challenges and move forward efficiently. Being closely connected to the ground realities of the agricultural sector also allows us to create practical and impactful solutions.
LCB Pitch Deck 25-26.pdf
PDF
read this pitch deck and answer we carefully in humanized way in english 1= One-liner description of your startup
Here’s a clear, humanized one-liner description of your startup based on your pitch deck:
One-liner description of your startup:
LCB Fertilizers is a deep-tech agri startup that converts agricultural waste into customized, high-performance bio-nano fertilizers using microbial, nano, and polymer technologies to improve soil health and farmer profitability.
What problem are you solving in one lineer?
Here’s a crisp, humanized one-liner you can use:
What problem are you solving?
We are solving soil degradation and low farm productivity by turning underutilized agricultural waste into high-quality, sustainable fertilizers that improve soil health and farmer income.
Describe your solution (this not one liner)
Here’s a clear, humanized answer you can use:
Describe your solution:
At LCB Fertilizers, we have developed a technology-driven solution that converts agricultural waste such as slurry, dung, and biomass into high-performance, crop-specific fertilizers. Our approach combines microbial science, nanotechnology, and polymer technology to create fertilizers that not only provide essential nutrients but also improve soil structure, water retention, and overall soil health.
We use a consortium of beneficial microorganisms that naturally generate key nutrients like nitrogen, phosphorus, and potassium in the soil. Along with this, our nano-enhanced components improve nutrient efficiency and microbial survival, while our super-absorbent polymer helps retain moisture in the soil for longer periods, reducing the need for frequent irrigation.
What makes our solution unique is its customization—we analyze soil, crop, and climatic conditions to create targeted formulations that deliver better results for farmers. In addition, we enable decentralized production by setting up low-cost bio-manufacturing units, allowing farmers, FPOs, and local organizations to convert locally available waste into valuable fertilizer products.
Overall, our solution creates a sustainable “waste-to-wealth” ecosystem that reduces input costs, increases yield, improves soil health, and supports long-term agricultural sustainability.
make little bit shorter and make this attractive
Here’s a shorter and more attractive version:
Describe your solution:
At LCB Fertilizers, we transform agricultural waste like slurry and biomass into high-performance, crop-specific fertilizers using a unique combination of microbial, nano, and polymer technologies. Our solution enriches soil with essential nutrients, improves water retention, and boosts crop productivity.
We also enable decentralized production by setting up low-cost bio-manufacturing units, allowing farmers and local organizations to convert waste into valuable inputs. This creates a sustainable “waste-to-wealth” model that reduces farming costs, improves soil health, and increases farmer income.
What makes your solution unique or defensible?
Here’s a strong, humanized answer:
What makes your solution unique or defensible?
What makes our solution unique is our integrated deep-tech approach that combines microbial science, nanotechnology, and polymer innovation into a single, customized fertilizer solution. Unlike traditional fertilizers, our products are tailored to specific soil, crop, and climate conditions, delivering better efficiency and results.
We have developed proprietary microbial formulations and processes to convert local agricultural waste into high-value fertilizers, creating a strong “waste-to-wealth” model. Our decentralized bio-manufacturing setup further strengthens our defensibility by enabling scalable, low-cost production at the local level.
Additionally, our technology not only improves yield but also enhances soil health, reduces water usage, and lowers input costs—creating long-term value for farmers, which is difficult to replicate.
make little bit shorter
Here’s a shorter, sharper version:
What makes your solution unique or defensible?
Our uniqueness lies in combining microbial, nano, and polymer technologies into one integrated, crop-specific solution. Unlike traditional fertilizers, we customize our products based on soil, crop, and climate conditions for better results.
We also convert local agricultural waste into high-value fertilizers through our proprietary processes and decentralized manufacturing model, making our solution scalable, cost-effective, and difficult to replicate while delivering long-term benefits to farmers.
Who are your target customers?
Here’s a clear, humanized answer:
Who are your target customers?
Our target customers include farmers, Farmer Producer Organizations (FPOs), and rural livelihood groups who are looking for cost-effective and sustainable fertilizer solutions.
We also serve agri-based industries such as sugar mills, bioenergy and bio-CNG plants, gaushalas, and corporate farming companies that generate agricultural waste and can benefit from our waste-to-wealth model.
In addition, we work with government bodies and international markets that are focused on sustainable agriculture and soil health improvement.
make itno 1-2 liner
Who are your target customers?
Our target customers include farmers, FPOs, and agri-based industries like sugar mills, bioenergy plants, and gaushalas, along with government and global markets focused on sustainable agriculture.
TAM
TAM (Total Addressable Market):
Our total addressable market is the global fertilizer, biofertilizer, and waste-to-wealth market, estimated at around $770 billion, covering farmers, agri-industries, and sustainable agriculture ecosystems worldwide.
IN NUMBER ONLY
770 Billion USD
SAM AND SOM
SAM: 35–50 Billion USD
SOM: 150–250 Million USD
What is your revenue model? (IN ONE LINER ONLY)
We generate revenue through sales of crop-specific fertilizers, microbial concentrates, setting up decentralized manufacturing units, and technology licensing/co-development with partners.
Who are your main competitors? (IN ONE LINE ONLY)
Our main competitors include traditional chemical fertilizer companies, organic fertilizer producers, and emerging biofertilizer and agri-biotech startups.
How do you acquire customers?
We acquire customers through on-ground farmer engagement, partnerships with FPOs and agri-organizations, collaborations with industries, and government and institutional networks.
What is your go-to-market strategy?
We follow a hybrid go-to-market strategy by combining direct farmer outreach with partnerships through FPOs, agri-industries, and decentralized manufacturing units to enable local production, distribution, and rapid scale.
What’s your long-term vision? (Not in one liner)
What’s your long-term vision?
Our long-term vision is to build a global, technology-driven agricultural ecosystem that transforms how fertilizers are produced and used. We aim to lead the shift from chemical-dependent farming to sustainable, biology-driven solutions that restore soil health, improve farmer income, and reduce environmental impact.
We envision scaling our decentralized “waste-to-wealth” model across India and internationally, enabling thousands of local production units that convert agricultural and industrial waste into high-value fertilizers. By integrating advanced microbial, nano, and data-driven technologies, we want to create a system where farming becomes more efficient, climate-resilient, and profitable for millions of farmers.
Ultimately, our goal is to position LCB Fertilizers as a global leader in agri-biotech innovation, driving sustainable agriculture at scale while contributing to food security and environmental sustainability.
Incorporation status (we are incorporate in 18-11-2020)
Incorporation Status:
Incorporated as a Private Limited Company on 18 November 2020.
How much funding have you raised (no fund raised in bootstrapped comnpany )
How much funding have you raised?
We are a bootstrapped company and have not raised any external funding so far.
Why are you applying to Avinya? (we applying to avinya markets connections mil sake aor mentorship and little bit fun mil sake )
Why are you applying to Avinya?
We are applying to Avinya to gain strong market connections, access valuable mentorship, and learn from experienced founders and experts. We also see it as a great platform to grow, collaborate, and enjoy the journey of building together.
What do you want to achieve during the programme?
What do you want to achieve during the programme?
During the programme, we aim to strengthen our market presence, build meaningful partnerships, and refine our business and go-to-market strategy. We also want to learn from mentors, improve our product positioning, and prepare ourselves for scalable growth and future fundraising.
in one liner
We aim to build strong market connections, refine our strategy, and prepare for scalable growth and future fundraising.
Are you open to hybrid or virtual participation? (virtual)
Are you open to hybrid or virtual participation?
Yes, we are open to virtual participation.
How is AI integrated into your product? *
Core engine
Supporting feature
Automation layer
Not applicable
Supporting feature
Describe your AI architecture
Describe your AI architecture:
Our AI layer acts as a supporting system that analyzes soil, crop, and environmental data to recommend customized fertilizer formulations. We use data inputs like soil parameters, location-based insights, and crop requirements to generate actionable recommendations.
The system is designed to assist our core biological technology by improving precision, optimizing input usage, and enabling better decision-making for farmers and production units, while remaining scalable and easy to integrate with our decentralized model.
What proprietary data advantage (if any) do you have?
What proprietary data advantage (if any) do you have?
We are building a proprietary dataset based on soil health parameters, crop performance, microbial behavior, and field trial results across different regions. This includes location-specific insights from our on-ground deployments and decentralized production units.
Over time, this data helps us improve our formulations, optimize recommendations, and create a strong feedback loop that is difficult for others to replicate. |
Our AI is defensible because it is built on proprietary, real-world data collected from our field deployments, soil analyses, and decentralized production units. This creates a strong feedback loop between our technology and on-ground results.
Additionally, our AI is deeply integrated with our unique microbial and fertilizer formulations, making it not just a standalone tool but part of a complete solution that is difficult for competitors to replicate. |
We evaluate performance based on real-world outcomes such as improvement in crop yield, reduction in input costs, soil health indicators, and accuracy of fertilizer recommendations. We also track consistency of results across different locations and seasons, along with user feedback from farmers and partners to ensure reliability. |
We ensure data privacy and security by collecting only necessary data, storing it securely, and restricting access to authorized personnel. We follow standard data protection practices and comply with applicable regulations, while anonymizing sensitive data wherever possible.
Additionally, we maintain transparency with users about how their data is used and ensure it is only utilized to improve our services and recommendations. |
Yes, as an agri-input and biotechnology-based company, we are subject to regulations around fertilizer approvals, product certifications, and environmental compliance. Changes in government policies or delays in approvals could impact market entry and scaling.
However, we actively work with regulatory bodies, follow required standards, and align our processes to ensure compliance and minimize such risks. |
At 10x scale, the biggest challenge would be maintaining data quality and consistency across diverse geographies, as variations in soil, climate, and farming practices increase. This can impact the accuracy of recommendations if not properly managed.
We may also face challenges in handling larger data volumes and ensuring real-time processing. However, we are addressing this by building scalable data pipelines, continuous model retraining, and strong validation systems to maintain performance at scale. |
Currently, we do not have an in-house AI/ML team, but we are leveraging external expertise and plan to build a dedicated team as we scale our technology. |
We primarily use a combination of our own field data (soil parameters, crop performance, and trial results) along with publicly available agricultural and environmental datasets.
Our core datasets are increasingly owned and built in-house through our on-ground operations, while external data sources are used as supportive inputs. |
We plan to continuously improve our model through ongoing data collection from field deployments, regular model retraining, and incorporating feedback from farmers and partners.
As our dataset grows across regions, crops, and seasons, we refine our recommendations and validate them through real-world outcomes to ensure consistent improvement in accuracy and reliability. |
We are building for both India and global markets, with a strong focus on scaling our solution across diverse agricultural regions. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/wjFdBpQ17Gp8-LCB-Pitch-Deck-25-26_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/wjFdBpQ17Gp8-LCB-Pitch-Deck-25-26_compressed.pdf |
https://drive.google.com/file/d/1zH-XR8CcjVVgIOqlcDv3csQlm0VAg5oO/view |
**Are you building a mission-driven or impact-focused startup? Tell us more.**
Yes, we are a mission-driven and impact-focused startup. Our core mission is to improve soil health, increase farmer income, and promote sustainable agriculture by reducing dependence on chemical fertilizers.
Through our waste-to-wealth approach, we aim to convert agricultural and industrial waste into valuable inputs, creating both environmental and economic impact. We are focused on empowering farmers, strengthening rural livelihoods, and building a more resilient and sustainable agricultural ecosystem at scale. |
**Are you part of an underrepresented group in tech/startups?** Yes, we represent an underrepresented segment in the startup ecosystem, coming from a rural and agriculture-focused background, which brings unique on-ground insights but is often less represented in tech-driven ventures. |
Not referred by anyone. |
checked |
| Apr 13, 2026 @ 2:46 AM |
arghya pratim ghosh, Rahul Kumar Ghosh |
arghyapratimghosh50@gmail.com |
https://www.linkedin.com/in/arghyapratimghosh/ |
https://maatriva.vercel.app |
9475080383 |
Founder and CEO |
Our partnership was forged during our undergraduate years, where Rahul and I were not only classmates but also comrades in the National Cadet Corps (NCC). The discipline and shared challenges of our bachelors and NCC training created a resilient bond that has become the bedrock of our collaboration. After graduation, our paths diverged momentarily—Rahul transitioned into the professional workforce while I pursued my Master’s degree. It was during this time, two years ago, that I began the intensive process of ideating and designing the early blueprints for this venture.
Driven by a vision to create the best possible version of this technology, I moved from conceptualization to implementation. Upon the completion of my Master’s last year, I shared the detailed project description with Rahul. His immediate enthusiasm for the venture’s potential led him to rejoin me, enrolling in the same course to align our expertise. Since then, we have been working side-by-side, leveraging our long-standing friendship and professional synergy to bring this venture to life. |
1 |
Our greatest strength lies in a unique culture where **unwavering discipline meets deep-rooted camaraderie**. Building on the foundation of trust established during our NCC days, we have fostered an environment where professional collaboration and genuine friendship are indistinguishable. Every new team member is seamlessly integrated into our close-knit circle, creating an immediate sense of belonging that transcends the traditional workplace. This high-trust environment, combined with meticulous strategic planning, empowers our team to consistently deliver **110% effort**, as every member feels personally invested in the entity’s mission and collective success. |
MAATRIVA |
https://maatriva.vercel.app |
New Delhi |
We are building a pre diagnostic tool for early detection of congenital and genetic disease in the process helping parenting easier. |
The Post-NICU Gap: High-risk infants lose medical-grade oversight immediately after discharge, leading to preventable readmissions. Hospital Overload: Overburdened nursing staff and "Alarm Fatigue" caused by inaccurate, single-sensor monitors. Parental Anxiety: Modern monitors rely on invasive wearables or radiation-emitting sensors that cause skin irritation and false alarms. |
MAATRIVA is a Multimodal Edge-AI ecosystem that provides a "Hospital-at-Home" safety net for high-risk infants. Unlike traditional monitors that rely on invasive wearables, our solution uses passive sensor fusion to automate clinical-grade tracking without touching the baby.
The Solution in Brief
Contactless Clinical Monitoring: Using a 60GHz mmWave sensor, the system tracks Heart Rate (HR), Breathing Rate (BR), and sleep posture with medical precision—no wires, no radiation, and no skin irritation.
Intelligent Thermal Regulation: An integrated 6-zone autonomous heating array maintains the infant’s body temperature, specifically designed to prevent cold stress in preterm or low-birth-weight babies.
Automated Triage & Privacy: By processing all data locally on the Edge (STM32MP1), we eliminate "Alarm Fatigue" for nurses and parents through cross-modal verification, while ensuring 100% data privacy.
The Clinical Bridge: We serve as the vital link between NICU discharge and home recovery, integrating with programs like Mission ANMOL to ensure that infants with congenital or genetic disorders are never left unmonitored. |
MAATRIVA is unique because it transforms the smart cradle from a simple "soother" into a clinical-grade diagnostic hub. The Innovation: We are the first to use Passive Sensor Fusion (Sight, Sound, Smell, Heat, and Touch) to monitor high-risk infants without a single wire or wearable touching their skin. The Moat: Our defensibility lies in a Hardware-Software Lock; our proprietary AI is custom-tuned to the specific physics of our piezoelectric fabric and thermal grid, making it impossible to replicate with software alone. The Strategic Edge: By integrating with government initiatives like Mission ANMOL, we secure a distribution channel in public health that consumer brands like Owlet or Nanit cannot access. In short: We win by providing "Hospital-at-Home" security that eliminates false alarms and bridges the dangerous gap between NICU discharge and long-term recovery. |
MVP |
Signups |
Primary Segment (The "High-Risk" Families): Parents of NICU graduates and infants diagnosed with congenital heart defects (CHD), chronic respiratory issues, or genetic disorders who require 24/7 clinical-grade monitoring to prevent life-threatening events. Government Segment (The ANMOL Pipeline): Direct integration with the Delhi Government’s ANMOL program, capturing the high-conversion market of infants screened and diagnosed with life-altering conditions at the grassroots level. Institutional Segment (Healthcare & Childcare): B2B sales to hospitals, maternity wards, and premium daycare centers to automate newborn monitoring, reduce nursing workload, and provide a seamless "safe-discharge" bridge for recovering infants. |
1.56 lakh cr |
7200 cr |
216 cr |
Premium Hardware Sales (Upfront): High-margin sales of the Smart Cradle to NICU graduates (B2C) and bulk leasing/sales to hospitals and maternity wards (B2B) to automate newborn monitoring. "Pediatric-as-a-Service" (Recurring): A monthly subscription for the AI Care-Bot, advanced health analytics, and automated clinical reports that parents share with their doctors. Public Health & Data (Strategic): Service contracts through the ANMOL program and licensing anonymized datasets to clinical researchers studying infant health and SIDS. |
1. The Smart Cradle & Furniture Leaders (Soothing-First) These companies focus on the nursery environment and automated sleep cycles, but lack clinical vital sensors. Cradlewise: A smart crib that uses AI cameras to detect early wake-up signs and bounces the baby back to sleep. SNOO (Happiest Baby): A high-end robotic bassinet that uses motion and white noise to soothe newborns, but tracks zero health vitals. Maxi-Cosi Starling: A newer 2026 competitor using AI cry analysis to trigger specific rocking modes. The Gap: They are lifestyle products for healthy babies. They cannot monitor heart rate, fever, or respiratory distress for a high-risk NICU graduate. 2. The Consumer "Smart" Monitors (Surveillance-First) These are the market giants, but they rely on invasive wearables or simple video analytics. Owlet (Dream Duo): Uses a pulse-oximeter sock. While popular, it causes skin irritation and frequent "false alarms" if the sock slips. Nanit Pro: Uses "Breathing Wear" (patterned fabric) and cameras. It fails if the baby is under a blanket or the room is too dark. Sense-U: A clip-on diaper sensor that tracks abdominal movement and temperature but requires a physical attachment to the baby. The Gap: They are intrusive. Preterm or high-risk infants have sensitive skin and need 100% contactless monitoring, which these brands cannot provide. 3. Contactless & MedTech Startups (Tech-First) These are our closest technical competitors, utilizing radar or clinical-grade AI. Miku Pro: Uses proprietary waveform technology for contactless breathing and sleep tracking. Raybaby: An Indian-origin startup using ultra-wideband radar to track respiration and sleep patterns. Janitri (India): A strong MedTech peer focused on maternal and newborn vitals, often used in hospital settings. CergenX: An emerging 2026 leader in AI-powered brain health monitoring for newborns. The Gap: None of these offer Multimodal Fusion. While they may track breathing, they lack the integrated Thermal Heating Array for temperature regulation and Olfactory sensors for hygiene, which are critical for medical-grade care at home. |
Our customer acquisition strategy is built on a "Trust-First" multi-channel approach, recognizing that for high-risk infant care, credibility is as important as the technology itself. We move away from aggressive "selling" and focus on clinical partnership and authentic support. 1. Community-Led Organic Growth (B2C) Instead of traditional ads, we build a presence where anxious parents look for answers. Educational Content: We create high-value content focused on NICU transition, managing congenital conditions, and infant health milestones. The "Genuine Help" Model: We participate in parenting forums and social media groups not to pitch, but to provide expert advice on infant safety. By positioning MAATRIVA as a thought leader, we turn followers into brand advocates. Influencer Advocacy: Partnering with "Med-moms" and pediatric influencers who have lived through NICU experiences to share authentic testimonials. 2. Institutional Partnerships (B2B) We position the cradle as an operational asset for healthcare providers. Hospitals & Maternity Wards: We offer bulk procurement or leasing models to hospitals. By integrating our system into their wards, we reduce the nursing workload through automated triage, while the hospital provides a "premium safety" experience to new parents. Daycare Centers: We target premium daycares that want to offer parents 100% transparency and "Remote Peace of Mind" via our Dynamic QR access and AI health reports. 3. The Clinical Prescription Path (The "ANMOL" Bridge) Doctors are our most critical gatekeepers, but we maintain a strictly ethical relationship. Doctor Referrals: Rather than "commissions" (which can face ethical/regulatory scrutiny in MedTech), we focus on Clinical Utility. We provide doctors with a "Doctor’s Dashboard," allowing them to view real-time, medical-grade vitals of their high-risk patients. When a doctor sees that our tool helps them save lives and manage cases better, they prescribe MAATRIVA as a post-discharge necessity. Mission ANMOL Integration: As our primary strategic channel, we capture customers at the point of government screening. When an infant is diagnosed with a condition under ANMOL, MAATRIVA is positioned as the recommended "safety bridge" for home care. |
Our customer acquisition strategy focuses on capturing the market at the point of diagnosis, ensuring that MAATRIVA is seen as a medical necessity rather than an optional luxury. 1. The Diagnostic Pipeline (Direct B2C) We capture customers exactly when they realize they need us most: Mission ANMOL Beneficiaries: Infants diagnosed with congenital or genetic disorders through the government's screening program are our primary direct customers. We act as the immediate "prescribed" solution for home monitoring post-diagnosis. External Clinical Diagnoses: We target parents whose babies undergo external private testing (e.g., genetic screening, cardiac echoes, or metabolic panels) and receive high-risk results. For these families, MAATRIVA is the essential bridge between the lab report and daily survival. 2. Institutional & Expert Channels (B2B & Referrals) Hospitals & Maternity Wards: We secure bulk orders and leasing contracts to automate ward monitoring, reducing nursing staff workload while providing a "safety-first" environment for newborns. Daycare Centers: We partner with premium childcare facilities that use our technology to provide parents with real-time health transparency and "Remote Peace of Mind." Doctor-Led Prescriptions: We work with pediatricians and neonatologists, providing them with clinical dashboards. Doctors recommend MAATRIVA to high-risk patients, ensuring their medical oversight continues even after the baby goes home. 3. Organic Advocacy & Social Support Help-First Social Outreach: We grow organically by creating genuine, educational content that helps parents navigate the complexities of high-risk infant care. Community Integration: By actively supporting families in NICU and special-needs forums, we build deep trust. This "human-first" approach turns our social media presence into a supportive community rather than just a sales channel. |
Our long-term vision is to evolve from a specialized infant care startup into a global leader in proactive health intelligence. We aim to revolutionize how the world monitors vulnerable populations through two strategic expansion pillars:
Global Scalability: Leveraging our Edge-AI and mmWave technology, we plan to scale beyond India to international markets where the "aging-in-place" and "NICU-at-home" trends are surging. By adhering to global standards like CE and FDA, MAATRIVA will become a universal safety layer for families worldwide.
The "Smart Bed" Evolution (Elderly Care): We will transition our core technology from the cradle to the adult bedroom. Our vision includes developing Smart Beds for the elderly, using the same 100% contactless, radiation-free sensors to track vitals, fall detection, and sleep apnea. This allows seniors to maintain their independence while providing families and healthcare providers with real-time, clinical-grade oversight without the need for invasive cameras or wearables. |
Not registered yed, but will register soon |
2 lakhs |
Yes |
We are applying to Avinya to bridge the gap between our finalized prototype and market entry. Having successfully completed our incubation and funding tenure at IIT Delhi, we are now seeking new strategic partners to help us navigate the pre-commercialization phase. Avinya’s focus on high-impact, deep-tech social enterprises perfectly aligns with our mission to democratize neonatal care. Alignment & Strategy The program fits our "scheme of things" as we move into Phase 2 (Clinical Pilots & Compliance). We need Avinya’s expertise to: Validate at Scale: Transition from a "lab-ready" prototype to a certified medical device through hospital pilots. Government Integration: Refine our operational model for Mission ANMOL, ensuring our technology reaches diagnosed infants efficiently. Funding Transition: As our initial residency at IIT Delhi has ended, we are actively diversifying our capital base. Avinya is our top choice for the mentorship and runway required to launch by May 2027. |
Our Primary Goals in the Avinya Program Secure Seed Funding for Commercialization: With our initial tenure and funding at IIT Delhi having concluded, we aim to secure Avinya’s initial seed investment (up to ₹10 Lakhs) and qualify for the follow-on investment of up to ₹1 Crore to fuel our manufacturing and compliance costs. Achieve Regulatory & Clinical Readiness: We will leverage ACSEL’s network of clinicians and medical experts to navigate the CDSCO/CE certification process, ensuring we hit our May 2027 pilot deadline. Refine the AI Edge Infrastructure: Utilizing Avinya’s "curated datasets" and AI-focused mentorship, we will optimize our mmWave sensor-fusion and Edge-AI (on the STM32MP1) for zero-latency, high-accuracy medical triage. Validate the GTM Model: We intend to use the program’s industry connects to pilot our B2B hospital leasing model and the Mission ANMOL diagnostic integration. |
Yes we quite prefer virtual participation, actually it would be a headache to transport our whole team to your headquarters. Any way we are open to anything you prefer. |
Yes |
Core engine |
1. The Input Layer: Multimodal Sensor Fusion
We don't rely on a single data point. Our architecture fuses three primary streams to eliminate false alarms:
mmWave Stream (Primary): A 60GHz Frequency-Modulated Continuous Wave (FMCW) radar tracks micro-movements (chest displacement) for Heart Rate and Breathing Rate.
Acoustic Stream: A MEMS microphone array feeds into a Convolutional Neural Network (CNN) for cry classification (distress, hunger, or pain).
Thermal Stream: An IR thermal array monitors body temperature zones, feeding a feedback loop for the autonomous heating system.
2. The Processing Layer: Edge-Inference (No Cloud)
Unlike competitors who upload sensitive baby videos to the cloud, our "Brain" lives inside the cradle:
Hardware: STM32MP157F-DK2 (Dual Cortex-A7 + Cortex-M4).
Signal Processing: Raw radar and audio signals undergo Fast Fourier Transforms (FFT) and filtering on the M4 co-processor to reduce noise.
Deep Learning Models: Compact, quantized models (TensorFlow Lite) run on the A7 cores to perform real-time posture detection and vital sign extraction.
3. The Intelligence Layer: Cross-Modal Verification
This is our "Clinical Logic" that prevents Alarm Fatigue:
The Validation Loop: If the camera sees the baby is "still" but the mmWave sensor detects a normal heart rate, the AI suppresses the "No Movement" alarm.
Autonomous Intervention: If the thermal sensor detects "Cold Stress," the AI triggers the heating array immediately without waiting for a manual toggle.
4. The Output Layer: Secure Access
Local Triage: Real-time alerts are sent via BLE/Local Wi-Fi to the parent’s handset.
Dynamic QR: For hospital/daycare use, the AI generates a time-bound, encrypted QR code, allowing temporary, secure access to the stream without permanent cloud storage. |
Proprietary Data Advantage & Strategy
Currently, there is no existing global dataset that fuses 60GHz mmWave radar, 6-zone thermal arrays, and AI-driven facial/posture cues specifically for high-risk infants.
The Current Gap: We are pioneering a "web of data" that has never been linked before. Because our multimodal approach is novel, a singular, comprehensive data source does not yet exist in the public or medical domain.
The Pilot Advantage: During our upcoming clinical pilot phase, we will collect the world’s first integrated dataset of contactless neonatal vitals mapped against synchronized visual and thermal distress signals.
The Data Moat: As we collect this proprietary data, our models will undergo continuous refinement, increasing accuracy in "False Alarm Suppression" and "Early Distress Detection." This creates a significant barrier to entry; any future competitor would have to spend years replicating the clinical trials required to match our calibrated AI weights. |
1. Multimodal "Sensor Fusion" Logic
Most competitors use a single data source (e.g., just a camera or just a wearable). Our defense lies in the Cross-Modal Verification layer.
The Defense: We are building a "web of data" where the mmWave radar, Thermal Array, and Visual AI constantly validate each other.
Why it's hard to copy: A competitor can buy a radar sensor, but they won't have the proprietary algorithms that determine how to weigh a "breathing signal" from the radar against a "posture cue" from the camera to suppress a false alarm. This "logic-weighting" is our intellectual property.
2. Edge-Native Specialization (STM32MP1)
While many startups rely on "Cloud AI" (sending data to a server), we have optimized our models to run entirely on the Edge.
The Defense: We have undergone the heavy engineering task of quantizing and compressing medical-grade models to run on low-power industrial processors (STM32MP1).
Why it's hard to copy: It takes months of specialized embedded-systems engineering to make a deep-learning model "fit" on a cradle’s processor without losing clinical accuracy. Most consumer brands won't invest in this "hardware-software co-design."
3. The "First-Mover" Clinical Data Moat
As you mentioned, there is currently no public dataset that combines radar, thermal, and facial cues for high-risk infants.
The Defense: By the time a competitor enters, our models will have been trained on thousands of hours of proprietary pilot data from hospitals and the Mission ANMOL program.
Why it's hard to copy: You cannot download our training data from the internet. A competitor would have to start from zero, gaining clinical permissions and running their own year-long pilots just to reach our starting point.
4. Integration Defensibility (The "ANMOL" Lock-in)
Our AI is being designed to plug directly into the Delhi Government’s diagnostic pipeline.
The Defense: Once our AI is calibrated to the specific screening standards of Mission ANMOL, it becomes the "Default Standard of Care."
Why it's hard to copy: A competitor doesn't just need better AI; they need the same institutional trust and diagnostic integration that we have built over years of partnership. |
Evaluation of AI Model Performance & Reliability
We evaluate our 8 integrated models through a three-tier framework: Clinical Precision, Edge Efficiency, and Systemic Reliability.
1. Tier 1: Clinical Precision (Task-Specific Metrics)
Vital Signs (mmWave - HR/BR): We use Mean Absolute Error (MAE) and Bland-Altman Plots to measure correlation against clinical-grade ECG and SpO2 references. Our target is an error margin of 90% using quantized CNNs.
2. Tier 2: Edge Efficiency (Hardware Reliability)
Since our AI runs locally on the STM32MP1, we monitor hardware-level reliability:
Inference Latency: Every model must complete a cycle in <100ms to ensure real-time triage.
Thermal & Resource Stability: We track CPU/RAM utilization to prevent "thermal throttling" of the chip, ensuring the device remains reliable during 24/7 continuous operation.
Quantization Loss: We measure the accuracy drop between our 32-bit training models and our 8-bit quantized Edge models to ensure clinical integrity is maintained.
3. Tier 3: Systemic Reliability (Fusion & Triage)
The ultimate test is our False Alarm Suppression Rate. We evaluate reliability by stressing the system under "Adverse Conditions":
Zero-Lux Testing: Ensuring mmWave and Thermal vitals remain accurate when visual cameras fail at night.
Obstruction Resilience: Testing vital sign extraction when the infant is covered by a light blanket.
Cross-Modal Validation: We calculate the Area Under the ROC Curve (AUC-ROC) for our "Triage AI," ensuring it only alerts parents when multiple sensors (e.g., mmWave + Audio) confirm a distress event. |
Data Privacy & Security Framework
At MAATRIVA, we adhere to a "Privacy-by-Design" philosophy, ensuring that sensitive infant health data remains under the parents' absolute control.
1. Edge-Native Processing (B2C Security)
Unlike standard monitors that stream raw video and vitals to the cloud, MAATRIVA uses Edge-AI on the STM32MP1.
Local Inference: All AI processing (breathing detection, cry analysis, and posture cues) happens locally on the cradle's hardware.
No Raw Data Uploads: We do not store raw video or audio on any external servers. Only processed, encrypted health alerts are sent to the parent’s device, virtually eliminating the risk of large-scale cloud data breaches.
2. Granular Caregiver Access (Dynamic Permissions)
We have developed a "Least Privilege" sharing model for when parents are away:
Time-Bound Access: Parents can grant temporary access to a secondary caregiver (nanny, relative) via a secure, time-expiring digital token.
Limited Data Visibility: Caregivers are restricted from viewing the full medical history or sensitive developmental logs; they only receive real-time safety alerts and a live vitals feed for the duration of their session.
Encrypted Handshakes: All device-to-app communication is secured using End-to-End Encryption (E2EE), ensuring that even our internal systems cannot intercept the live feed.
3. Medical History & PII Protection
Data Minimization: We only collect the minimum Personally Identifiable Information (PII) required for clinical accuracy.
Anonymized Vaults: Any data used for long-term health trends is de-identified and encrypted at rest using AES-256 standards, separating medical vitals from the user's identity.
4. Proactive Vulnerability Management
To ensure institutional-grade security, our roadmap includes:
External Security Audits: We are budgeting for a Third-Party Software Security Expert to perform rigorous penetration testing and vulnerability assessments (VAPT) before our May 2027 launch.
Firmware Integrity: We utilize Secure Boot on our embedded hardware to prevent unauthorized firmware modifications or "man-in-the-middle" attacks on the cradle. |
While MAATRIVA is a non-invasive, contactless device, the fact that it monitors vital signs and regulates temperature for high-risk infants places it under specific medical regulatory categories.
Here is the breakdown of the regulatory landscape and the certifications we are targeting.
1. CDSCO (India) - Our Primary Regulatory Body
In India, medical devices are governed by the Central Drugs Standard Control Organization (CDSCO).
Classification: Based on the Medical Device Rules 2017, MAATRIVA falls under Class B (Moderate Risk). While simple monitors are sometimes Class A, any device that monitors life-critical vitals (Heart Rate/Respiration) or active thermal intervention is classified as Class B.
The Pathway: We will apply for a Manufacturing License (Form MD-9) via the SUGAM portal. Since our initial tenure at IIT Delhi has ended, we are preparing the Device Master File (DMF) and Quality Management System (QMS) documents required for this stage.
2. Essential Global Certifications
To scale globally and ensure "Medical Grade" trust, we are aligning with:
ISO 13485: This is the gold standard for Medical Device Quality Management Systems. It proves our design and manufacturing processes are consistent and safe.
CE Marking (EU MDR): To enter the European market, we will seek Class IIa certification under the Medical Device Regulation (MDR). This requires an audit by a "Notified Body."
FDA (USA): For the US market, we would pursue the 510(k) Clearance or De Novo pathway. The FDA recently issued warnings against "unauthorized" infant monitors; being FDA-cleared will distinguish us from consumer-grade toys.
3. Technical Standards We Must Meet
To pass every audit "without issues," we are designing for:
IEC 60601-1: General requirements for basic safety and essential performance of medical electrical equipment.
IEC 60601-1-2: Electromagnetic compatibility (ensuring our mmWave radar doesn't interfere with other hospital equipment and vice versa).
IEC 62304: Medical device software lifecycle (crucial for our Edge-AI algorithms).
Why we expect to pass "without issues"
Contactless Nature: Because our sensors never touch the baby's skin, we eliminate risks associated with skin irritation, bio-compatibility, or electrical shocks from wearables.
Edge-AI Privacy: Our on-device processing mitigates the massive regulatory "Cybersecurity Risk" associated with cloud-based medical data.
Redundant Logic: Our "Sensor Fusion" (mmWave + Thermal + Visual) provides a fail-safe mechanism. If one sensor fails, the others provide a safety net, reducing the risk of "Device Failure" during clinical trials. |
Scaling 10x (e.g., from 100 to 1,000 units) transitions a startup from "controlled testing" to "real-world stress." While our Edge-AI architecture shields us from server-side meltdowns, scaling introduces physical and operational bottlenecks.
1. What Won’t Break: The Intelligence Layer
Because we use Edge-AI (STM32MP1), a 10x increase in users does not slow down our processing speed.
No Server Lag: Unlike competitors who rely on AWS/Google Cloud for inference, our "brain" is decentralized. 1,000 cradles processing vitals simultaneously put zero additional load on our core infrastructure.
Privacy Scalability: Scaling doesn't increase our data breach surface area significantly because sensitive video/audio stays on the device.
2. What Might Break: The Service & Support Infrastructure
The primary "breaking point" at 10x is the Human & Physical Layer.
Hardware Failures at Scale: At 1,000 units, a 1% failure rate means 10 infants without a safety net. This puts immense pressure on physical repairs.
Firmware Deployment: Pushing "Over-the-Air" (OTA) updates to 1,000 devices simultaneously without "bricking" (crashing) a unit requires a robust DevOps pipeline.
Supply Chain Consistency: Ensuring the 1,000th 6-zone thermal array performs exactly like the 1st requires rigorous Quality Control (QC).
3. Our Defensive Strategy (Scaling 10x)
We are proactively building a "Resilience Layer" to handle this growth:
Cybersecurity & Code Integrity: * Leveraging my background in Ethical Hacking (IIT Kharagpur), we are building security into the kernel level.
Pre-Production Audit: Before mass manufacturing, we will hire a Professional DevOps Engineer and an Ethical Hacker to perform a full-stack code review and penetration test to ensure no backdoors exist in our Edge-to-App communication.
The "Human-in-the-Loop" Service Model: * Recognizing that hardware requires a physical touch, we will scale through a Dedicated Service Network.
We will hire and train specialized service technicians who can visit the site (home or hospital) to perform manual repairs, sensor recalibration, or hardware swaps. This ensures that even if the software is "perfect," the physical cradle remains a reliable medical-grade tool.
Modular Hardware Design: * To make 10x scaling easier, we are designing the cradle with modular components. If a sensor fails, a technician can "plug-and-play" a new module on-site rather than taking the entire unit away, minimizing downtime for the high-risk infant. |
In-House AI/ML Expertise
MAATRIVA is powered by a dedicated in-house AI/ML team that combines academic rigor with hands-on industrial expertise. Our team is specifically structured to handle the complexities of multimodal sensor fusion and Edge-AI optimization.
Rahul Kumar Ghosh (AI Lead): Rahul heads our AI divisions, bringing deep technical insights into model architecture and deployment. His focus is on ensuring that our vital sign extraction (mmWave) and distress classification models meet clinical-grade accuracy. LinkedIn: Rahul Kumar Ghosh
Arman Sharma (AI/ML Engineer): Arman is instrumental in building and fine-tuning our vision and acoustic models. He works closely on the integration of posture detection and facial cue analysis, ensuring these complex models are optimized for our Edge hardware. LinkedIn: Arman Sharma
Founder-Led Integration: I (the Founder) actively drive the AI Pipeline development. With prior experience building multi-model pipelines, I oversee the high-level fusion logic—ensuring the "web of data" (Radar + Vision + Thermal) works synchronously to eliminate false alarms and provide predictive health insights.
Team Growth & Vision
We are currently in an active hiring phase to further strengthen our deep-tech capabilities:
Scaling the Team: We are bringing in new members specialized in Embedded AI and DevOps to ensure our code remains secure and efficient as we move toward clinical trials.
Collaborative Culture: Our internal workflow is highly collaborative; while specialized members head individual divisions, the entire team participates in cross-model validation and ethical AI reviews to maintain our "Privacy-First" architecture. |
Current Data Infrastructure
We are utilizing a hybrid approach to train our initial models while we prepare for clinical data acquisition:
Open-Source Foundations: We use established medical and acoustic datasets from Kaggle and PhysioNet to train our baseline models. This includes:
Cry Classification: Utilizing the Donate-a-Cry and Infant Cry Audio Corpus for initial acoustic feature extraction.
Vitals Baseline: Using human vital sign datasets (HR/BR) to calibrate our mmWave signal processing units.
Governmental Data (ANMOL & NCHS): We leverage statistics from the CDC’s NCHS and published Indian neonatal cohort studies to model the prevalence and specific vital patterns of high-risk infants in India.
Organic Data Collection (Customer Insights)
We have launched a survey-driven outreach on our website to collect organic data directly from parents and caregivers.
The Goal: To analyze customer preferences, pain points in current monitoring solutions (e.g., false alarms with socks/wearables), and the specific monitoring needs of families with high-risk infants.
The Benefit: This provides us with a "qualitative dataset" that informs our AI triage logic—ensuring we prioritize the alerts that matter most to caregivers.
The "Web of Data" Roadmap (Future Proprietary Data)
We recognize that no "singular" dataset exists that fuses mmWave, thermal, and visual data for neonates. Our plan to build this proprietary moat includes:
Phase 1: Pilot Testing: We are setting up clinical pilots in hospital settings to collect synchronized data streams. This will be the first of its kind, linking sub-mm chest displacement (radar) with high-definition visual posture and thermal distress markers.
Phase 2: Labelled Clinical Datasets: Through our partnership with doctors and the Mission ANMOL pipeline, we will have access to "ground truth" data—where our AI’s readings are verified against clinical NICU monitors.
Phase 3: Continuous Learning: Post-launch, our devices will use anonymized Edge-learning to refine their accuracy based on real-world usage patterns, creating a technical barrier that competitors cannot simply replicate with public data. |
Strategy for Continuous Model Improvement
Our roadmap for AI evolution transitions from baseline open-source training to a Proprietary Clinical Feedback Loop. We ensure reliability and accuracy through three distinct phases of data acquisition and refinement:
1. Post-Pilot Proprietary Training
Once our initial clinical pilots are complete, we will move beyond open-source datasets (Kaggle/PhysioNet) to train our models on high-fidelity, synchronized proprietary data.
The "Ground Truth" Advantage: By mapping our mmWave and Thermal readings against hospital-grade bedside monitors during pilots, we can perform "supervised fine-tuning" that is specifically calibrated for the neonatal physiology.
Edge-Refinement: We will use this pilot data to further quantize our models, ensuring that as they get "smarter," they remain efficient enough to run on the STM32MP1 without latency.
2. Institutional Data Partnership (B2B Scaling)
Our primary engine for long-term improvement is the Institutional Sale Model (Hospitals and Daycare Centers).
Clinical Diversity: Data collected from diverse NICU environments and varied demographics through hospital partnerships allows us to solve the "Generalization Problem"—ensuring our AI works across different lighting, crib types, and infant ethnicities.
Anonymized Bulk Learning: In institutional settings, we can collect large volumes of anonymized vital sign trends. This allows us to train our Predictive Triage models to identify early-onset distress patterns (like sepsis or respiratory lag) hours before they become visible.
3. Organic Data from B2C Users
Through our consumer website and app, we are already building a Customer Preference Dataset.
Feedback-Driven Logic: By analyzing user-reported "False Alarms" through our app’s feedback interface, we iteratively tune our Sensor Fusion logic to better distinguish between "Baby is just playing" and "Baby is in distress."
4. Security-First Code Review
To ensure this continuous improvement never compromises safety:
Pre-Production Audit: As a founder with an IIT Kharagpur Ethical Hacking background, I will oversee a rigorous security review.
Expert Validation: Before scaling, we will hire a Professional DevOps Engineer and an Ethical Hacker to review our entire AI pipeline, ensuring that the "OTA" (Over-the-Air) model updates are secure and that the local database remains encrypted against any on-site vulnerabilities. |
Market Strategy: India First, Global Expansion Next
1. Initial Market: The Indian Landscape (Phase 1)
Our immediate focus is the Indian market, specifically targeting urban parents and the public healthcare sector through initiatives like Mission ANMOL.
The Need: India has a high volume of high-risk neonatal cases where affordable, contactless monitoring is currently non-existent.
The Opportunity: By proving our technology in India’s diverse and challenging clinical environments, we establish a level of technical robustness that makes the product "world-ready."
2. Future Necessity: Western Market Expansion (Phase 2)
While our roots are in India, the Global North (North America and Europe) represents an essential strategic expansion for MAATRIVA.
The Consumer Base: In Western markets, the "Smart Cradle" and "Smart Nursery" culture is already deeply ingrained. Parents in these regions are early adopters who use monitoring technology as a standard daily utility rather than just an emergency tool.
High Margins & Adoption: The higher willingness to pay in Western markets allows us to scale our B2C premium features, while the established regulatory pathways (FDA/CE) align with our goal of becoming a globally recognized medical standard.
3. Building a "World-Grade" Product
We are designing for Global Standards from Day 1:
Compliance: Our focus on ISO 13485 and IEC 60601-1 is specifically intended to lower the friction for international entry.
Scalability: Our Edge-AI architecture is language-agnostic and privacy-compliant (GDPR/HIPAA ready), making the transition from a Delhi hospital to a London nursery seamless at the software level. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/3Uzqjb1N5ZKS-GRANT-Copy.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/3Uzqjb1N5ZKS-GRANT-Copy.pdf |
https://drive.google.com/file/d/18-kB6C1FIVoZPp-E6g6aGawkofH49U5g/view?usp=sharing |
Our Mission: Impact-Driven Innovation
MAATRIVA is not just a technology company; we are building an ecosystem of safety for the most vulnerable members of society. Our mission is to democratize high-quality neonatal care through three core impact pillars:
1. Empowering Parents & Early-Stage Childcare
The transition to parenthood is often defined by extreme anxiety and "alarm fatigue." By providing a reliable, contactless monitoring system, we reduce the psychological pressure on parents. We give them the confidence that their baby is being monitored by medical-grade AI, allowing them to rest and recover during the critical early months.
2. Economic Empowerment for Mothers
A significant barrier for women re-entering the workforce is the fear for their infant’s safety at home or in daycare.
Faster Return to Work: Our system acts as a "digital guardian," allowing mothers to sustain their babies’ monitoring for brief periods while they focus on their careers.
Bridging the Gap: By integrating features like Remote Caregiver Access, we provide mothers with real-time peace of mind, directly supporting workforce gender parity and economic independence.
3. Strengthening Healthcare Infrastructure
We aim to solve the critical shortage of nursing staff in hospitals and daycare centers.
Hospital Efficiency: In busy NICUs or pediatric wards, our AI-triage system alerts nurses only to true emergencies, reducing the manual workload of constant vitals checking.
Daycare Scalability: For daycare centers, MAATRIVA provides a "Safety-as-a-Service" model, allowing them to provide a higher standard of care for more infants simultaneously without compromising safety. |
MAATRIVA represents the under-represented category of Deep-Tech Hardware-AI founders in India. Transitioning from academic research at IIT Delhi to a social-impact venture, we are filling a critical gap in neonatal HealthTech. Our focus on Inclusive AI—designed to support working mothers and the Indian public health system—aligns with the mission-driven diversity that the Avinya program seeks to foster. |
Dr. Akash Sondhi, IHFC, IIT D |
checked |
| Apr 12, 2026 @ 1:39 PM |
Dr Himanshu Borase, Dr Ashwini Borase |
borasehimanshu@gmail.com |
http://www.linkedin.com/in/himanshu-borase-a62510a5 |
https://naree.health/ |
00447873892245 |
Dr Himanshu Borase, Founder and CEO, 25 yrs of experience in obstetrics and gynaecology , Dr Ashwini Borase, co-founder 15 yrs of experience as a dentist |
Ashwini is my sister-in-law, and we have worked together for last 5 yrs on this project |
2 |
Our core strength lies in a uniquely complementary, clinician-led founding team with deep domain expertise and real-world problem ownership. As a reproductive medicine consultant with over two decades of experience, I bring frontline insight into fertility care pathways, diagnostic delays, and patient pain points. My co-founder, a dentist with 15+ years of clinical experience, contributes strong patient-centric thinking, preventive care perspectives, and operational discipline from running clinical practice.
This combination enables us to design solutions that are not only technologically robust but clinically relevant and scalable. Unlike typical health tech startups, our product is built from lived clinical challenges, ensuring high adoption potential among both patients and providers.
Additionally, we have already demonstrated execution capability by building and deploying the Naree app, EPR system, and multilingual consent platform in real-world settings. This ability to translate insight into working products positions us strongly to deliver impactful, scalable innovation in women’s health. |
Nareeshakti Global Private Limited |
https://naree.health/ |
Mumbai, Maharashtra |
Naree is a full-stack digital women’s health platform combining a patient app, clinical EPR, and AI-driven insights to improve reproductive health outcomes. |
Women’s healthcare remains fragmented, reactive, and poorly digitised—leading to delayed diagnosis of conditions like endometriosis and PCOS, inefficient clinical workflows, and poor patient understanding. There is no unified platform that connects patient-reported data, clinical records, and decision support tools. Naree addresses this gap by creating an integrated, data-driven ecosystem that enables early detection, informed decision-making, and seamless care delivery across the entire women’s health journey. |
Naree is a full-stack digital women’s health platform that integrates a patient-facing app, a clinician-focused electronic patient record (EPR) system, and a multilingual digital consent platform into a single ecosystem.
The platform enables women to longitudinally track menstrual cycles, symptoms, and health data, while clinicians access structured, real-time patient information through an integrated EPR. This creates a continuous data loop between patients and providers, improving clinical decision-making and care coordination.
On top of this infrastructure, Naree is developing AI-driven risk prediction models for conditions such as endometriosis and PCOS. By analysing longitudinal symptom patterns, investigations, and clinical data, the platform aims to enable earlier detection, reduce diagnostic delays, and personalise care pathways.
Additionally, the multilingual consent module improves patient understanding and informed decision-making, particularly in diverse populations, reducing reliance on interpreters and improving clinical efficiency.
Together, Naree transforms fragmented, episodic care into a connected, data-driven, and patient-centric healthcare ecosystem scalable across both urban and underserved settings. |
Naree’s defensibility lies in its full-stack, clinically grounded ecosystem combined with proprietary longitudinal data and AI capability. First, unlike point solutions (e.g., tracking apps or standalone EPRs), Naree integrates patient app + EPR + multilingual consent platform into a single workflow. This creates high switching costs and deep embeds into both patient journeys and clinical operations. Second, the platform is designed to generate rich, longitudinal, multimodal datasets (symptoms, cycles, investigations, clinical outcomes). This forms a proprietary data moat, which is critical for training and continuously improving AI-driven risk prediction models—particularly in underrepresented populations. Third, the solution is clinician-led and built from real-world workflows, ensuring high relevance, trust, and adoption—key barriers where many digital health products fail. Fourth, our multilingual infrastructure and accessibility-first design uniquely position Naree for diverse and underserved populations, creating a scalable advantage across global markets. Finally, early traction (live deployments, active users, and institutional engagement) demonstrates execution capability, further strengthening our competitive moat beyond just technology. |
Users |
Users |
Naree operates a dual B2C and B2B model targeting both patients and healthcare providers. 1. Patients (B2C): Women aged 12–65 across the reproductive health spectrum—from adolescence to menopause—particularly those seeking support for menstrual health, fertility, PCOS, endometriosis, and general gynaecological care. A key focus is digitally engaged users as well as underserved populations with limited access to structured healthcare and education. 2. Healthcare Providers (B2B): Fertility clinics, gynaecology practices, and mid-sized hospitals that require efficient digital tools for patient management, documentation, and consent workflows. This includes clinics transitioning from paper-based or fragmented systems to integrated digital solutions. 3. Institutional & Ecosystem Partners: Healthcare networks, diagnostic centres, NGOs, and public health programmes aiming to improve women’s health outcomes at scale through digital infrastructure and data-driven insights. This dual approach enables Naree to build a connected ecosystem, driving both user adoption and recurring revenue through SaaS and subscription models. |
The global women’s digital health (FemTech) market is estimated at $60–70 billion+, with India representing a rapidly growing segment driven by ~350 million women of reproductive age. Including digital health, fertility services, and women’s health diagnostics, the India TAM is ~₹50,000–60,000 Cr. |
Naree targets digitally addressable users and healthcare providers in India: • ~120–150 million smartphone-using women (urban + semi-urban) • ~70,000+ gynaecology clinics, fertility centres, and hospitals This translates to a SAM of ~₹8,000–10,000 Cr, combining B2C subscriptions and B2B SaaS revenue. |
In the next 3–5 years, Naree aims to capture: • 1–2 million users on the platform • 1,000–2,000 clinics/hospitals This represents a realistic SOM of ~₹150–300 Cr, driven by subscription, SaaS licensing, and data-driven premium services. |
Naree operates a hybrid B2C + B2B SaaS revenue model, designed for scalability and recurring income. 1. B2C (Patients): • Freemium model with free access to basic tracking and educational content • Subscription-based premium features (AI-driven insights, personalised risk scores, advanced analytics) • Affiliate and marketplace revenue (diagnostics, home test kits, wellness products) 2. B2B (Healthcare Providers): • SaaS licensing fees for clinics and hospitals using the Naree EPR system • Subscription or per-use pricing for the multilingual digital consent platform • Onboarding and integration fees for new clinics 3. Enterprise & Partnerships: • Collaborations with hospitals, NGOs, and public health programmes • Potential partnerships with pharma and diagnostics companies for targeted outreach and research This diversified model ensures multiple revenue streams, strong unit economics, and long-term scalability, while leveraging both patient engagement and provider adoption. |
Naree operates across multiple segments, with competitors including consumer women’s health apps such as Flo and Clue, which focus on tracking and content but lack clinical integration; healthcare IT platforms like Practo and Athenahealth, which digitise clinical workflows but are not women’s health-specific and have limited patient engagement; and various niche point solutions in fertility, telehealth, and consent management that address isolated needs without offering an integrated ecosystem. In contrast, Naree differentiates itself as a full-stack platform combining a patient app, clinician EPR, and multilingual consent system, powered by longitudinal data and AI-driven insights, positioning it as a category-defining solution rather than a direct competitor to any single player. |
Naree uses a multi-channel, capital-efficient acquisition strategy across both B2C and B2B segments. For patients, growth is driven through content-led marketing (doctor-validated educational videos, social media, and YouTube), organic app store discovery, community outreach (schools, colleges, NGOs), and referral loops within the app. Strategic partnerships with influencers, diagnostics providers, and wellness brands further expand reach. For healthcare providers, acquisition is driven through direct outreach, demos, and pilot deployments with clinics and hospitals, supported by strong clinical credibility and workflow integration benefits. Early adopters are converted through onboarding support and case-based value demonstration. Additionally, partnerships with institutions and public health programmes enable scalable distribution. This dual approach ensures low customer acquisition cost on the B2C side and high lifetime value on the B2B side, creating a sustainable growth engine. |
Naree follows a dual-sided, bottom-up + top-down go-to-market strategy designed to build a connected patient–provider ecosystem. On the B2C side, we drive user acquisition through content-led growth (doctor-led education via YouTube and social media), app store optimisation, and community outreach (schools, colleges, NGOs). This builds trust, organic adoption, and a strong base of engaged users generating longitudinal health data. In parallel, on the B2B side, we target fertility clinics, gynaecology practices, and mid-sized hospitals through direct outreach, demos, and pilot deployments. We focus initially on clinics with high patient volumes and inefficiencies (paper-based or fragmented systems), offering immediate value through EPR digitisation and multilingual consent workflows. Our key strategy is B2C → B2B pull, where patient demand and engagement drive clinic onboarding, complemented by B2B → B2C push, where partner clinics onboard patients onto the platform. We prioritise tier 2 and tier 3 markets in India, where digital penetration is increasing but healthcare infrastructure remains fragmented, enabling faster adoption and lower competition. This is complemented by selective expansion in the UK through targeted clinical partnerships. This integrated approach enables rapid scaling, strong network effects, and recurring revenue across both sides of the ecosystem. |
Naree’s long-term vision is to become the default digital infrastructure for women’s health globally, spanning the entire lifecycle from adolescence to menopause. We aim to build a unified, data-driven ecosystem that seamlessly connects patients, clinicians, and healthcare systems, replacing fragmented and reactive care with continuous, personalised, and preventive healthcare.
At the core of this vision is the development of AI-powered predictive and decision-support systems, using longitudinal, real-world data to enable early detection of conditions such as endometriosis, PCOS, infertility, and other chronic women’s health disorders. Over time, Naree will evolve into a clinical intelligence platform, supporting not only individual care but also population health insights, research, and policy-making.
We also aim to create a globally scalable, multilingual platform that improves accessibility and health equity, particularly in underserved populations. By integrating education, diagnostics, care pathways, and digital consent into one ecosystem, Naree aspires to redefine how women engage with healthcare—making it proactive, personalised, and universally accessible. |
Incorporated in Mumbai, Maharashtra |
n/a |
Yes |
We are applying to Avinya to accelerate Naree’s transition from a clinically validated, deployed platform to a scalable, AI-driven women’s health solution with strong product–market fit in India. Avinya’s focus on innovation, mentorship, and ecosystem access aligns closely with our current stage—where we need structured support to refine our AI risk models, strengthen clinical validation, and optimise our go-to-market strategy across diverse Indian healthcare settings. Specifically, we are seeking strategic mentorship, access to healthcare networks, and guidance on regulatory and commercial pathways, particularly for scaling across tier 2 and tier 3 cities where the need is greatest. Avinya’s ecosystem can also help us build partnerships with hospitals, public health programmes, and industry stakeholders, which are critical for large-scale deployment. Additionally, we value Avinya’s ability to support founder development, investor readiness, and capital access, which will be instrumental as we prepare for our next phase of growth and fundraising. Overall, we see Avinya not just as an incubator, but as a strategic partner to help us build a category-defining, globally scalable women’s health platform from India. |
Through the Avinya program, we aim to achieve three core outcomes: First, strengthen product–market fit in India by refining Naree’s platform based on real-world clinical and patient feedback, particularly in tier 2 and tier 3 settings where healthcare delivery gaps are most pronounced. Second, advance our AI capabilities and clinical validation, including improving our risk prediction models for conditions like endometriosis and PCOS, and generating robust evidence to support adoption by clinicians and institutions. Third, accelerate go-to-market and scale, by securing strategic partnerships with hospitals, public health programmes, and ecosystem stakeholders, while building a structured pathway for national rollout. In parallel, we aim to enhance investor readiness and fundraising capability, positioning Naree for its next phase of growth. Overall, the program will help us transition from a strong early-stage platform to a scalable, validated, and investment-ready digital health company. |
Yes |
Yes |
Supporting feature |
Naree’s AI architecture is designed as a secure, modular clinical intelligence stack built on top of our patient app and clinician-facing EPR. At the data layer, we capture longitudinal structured and semi-structured inputs including menstrual cycle data, symptoms, pain patterns, fertility history, investigation results such as blood tests and scans, treatment history, and where available, operative findings. This data is ingested through the app and EPR into a secure cloud-based pipeline with consented data governance, cleaning, normalization, and feature engineering. On top of this sits the analytics and AI layer, where rule-based clinical logic is combined with machine learning models to identify patterns, stratify risk, and generate personalised risk scores for conditions such as endometriosis and PCOS. The architecture is modular, allowing different models to be trained, validated, and updated independently as more data becomes available. A clinician oversight layer ensures outputs are interpretable and used as decision-support rather than autonomous diagnosis. The final application layer delivers patient-facing insights, clinician dashboards, and workflow-triggered recommendations within the Naree ecosystem. This architecture allows us to continuously improve model performance as longitudinal real-world data grows, while maintaining privacy, explainability, and clinical relevance. |
Naree’s primary data advantage lies in its ability to generate longitudinal, multimodal, and clinically linked real-world data within a closed-loop ecosystem.
Unlike standalone apps or hospital systems, Naree captures continuous patient-reported data (menstrual cycles, symptoms, lifestyle inputs) and links it with clinical data from the EPR (consultations, investigations such as blood tests and scans, treatment decisions, and outcomes). This creates a high-resolution, end-to-end dataset across the patient journey, which is currently fragmented or unavailable in most healthcare systems.
Additionally, our multilingual platform and deployment across diverse populations (including underserved and tier 2/3 settings) enable us to build datasets that are more representative and less biased than existing global datasets, which are often skewed toward Western populations.
Over time, this structured and continuously growing dataset forms a proprietary data moat, enabling the development of more accurate and context-specific AI models for early detection and personalised care in women’s health—an advantage that is difficult to replicate without a similar integrated ecosystem and real-world clinical access. |
Naree’s AI is defensible due to the combination of proprietary data, ecosystem integration, and clinically grounded design.
First, our models are trained on longitudinal, multimodal datasets that link patient-reported symptoms with real clinical data (investigations, treatments, outcomes) within a single platform. This type of closed-loop dataset is rare and difficult to replicate, creating a strong data moat that improves model performance over time.
Second, the AI is embedded within a full-stack ecosystem (app + EPR + consent workflows), enabling continuous data capture and real-world validation. Competitors operating as standalone apps or tools lack this depth of integration, limiting both data quality and clinical utility.
Third, our approach is clinician-led with built-in interpretability and decision-support, ensuring outputs are trusted and usable in real clinical workflows—critical for adoption and regulatory acceptance.
Finally, the system is designed to be modular and continuously learning, allowing rapid iteration, localisation across geographies, and expansion into multiple conditions, strengthening both scalability and long-term defensibility. |
We evaluate Naree’s AI models using a combination of statistical performance metrics, clinical validation measures, and real-world reliability indicators.
1. Model Performance (Technical Metrics):
• AUC-ROC (Area Under Curve): Overall discrimination ability
• Sensitivity (Recall): Ability to correctly identify high-risk patients (critical for early detection)
• Specificity: Avoiding false positives
• Precision (PPV) & F1-score: Balance between accuracy and false alarms
• Calibration (e.g., Brier score): Alignment between predicted risk and actual outcomes
2. Clinical Validation Metrics:
• Concordance with clinician diagnosis and gold standards (e.g., imaging, laparoscopy where applicable)
• Reduction in time to diagnosis
• Improvement in appropriate referrals and clinical decision-making
3. Real-World Performance & Reliability:
• Model performance across different populations (age groups, geographies) to assess bias and generalisability
• Stability over time (performance drift monitoring)
• Missing data robustness and input variability testing
4. Safety & Trust Indicators:
• Explainability of outputs (feature contribution/clinical reasoning)
• Clinician acceptance and usability feedback
• False negative monitoring for high-risk conditions
This multi-layered evaluation ensures that our AI is not only statistically robust, but also clinically meaningful, reliable in real-world settings, and safe for decision-support use. |
Naree follows a privacy-by-design and security-first architecture, ensuring that patient data is protected across the entire lifecycle.
1. Data Privacy & Governance:
• Explicit, informed digital consent for data collection and usage
• Data minimisation and purpose limitation principles
• Role-based access control (RBAC) ensuring only authorised users access relevant data
• Anonymisation/pseudonymisation for analytics and AI model training
2. Security Infrastructure:
• End-to-end encryption (data in transit via TLS and at rest using industry-standard encryption)
• Secure cloud infrastructure with access logging, audit trails, and intrusion monitoring
• Regular vulnerability assessments and penetration testing
• Backup, disaster recovery, and business continuity planning
3. Regulatory Compliance:
• Designed to be compliant with GDPR principles (UK/EU)
• Alignment with Indian data protection frameworks (DPDP Act) for local deployment
• Preparation for healthcare-specific standards (e.g., ISO 27001, NHS DTAC, future medical device/AI regulations as applicable)
4. AI-Specific Safeguards:
• Use of de-identified datasets for model training
• Transparency and explainability in outputs (decision-support, not autonomous diagnosis)
• Ongoing monitoring for bias, drift, and unintended outcomes
This approach ensures that Naree maintains high standards of patient confidentiality, system security, and regulatory readiness, while enabling responsible use of data for improving healthcare outcomes. |
Yes—Naree operates in a regulated domain, and we proactively manage several key regulatory risks:
1. Medical Device / AI Classification:
Our AI risk-scoring modules may be classified as Software as a Medical Device (SaMD), requiring regulatory approval (e.g., CDSCO in India, UKCA/CE in Europe). This can impact timelines, validation requirements, and market entry.
2. Data Protection & Privacy Regulations:
Compliance with frameworks such as GDPR (UK/EU) and India’s DPDP Act is critical. Any changes in data localisation, consent, or cross-border data transfer rules could affect operations.
3. Clinical Validation Requirements:
Increasing expectations for clinical evidence and real-world validation of AI tools may require structured studies, ethical approvals, and ongoing monitoring before large-scale deployment.
4. Interoperability & Health IT Standards:
Integration with hospital systems and adherence to evolving standards (e.g., NHS DTAC in the UK) may introduce compliance and technical complexity.
5. Liability & Clinical Use Risk:
Ensuring the platform is positioned as decision-support rather than diagnostic authority is essential to mitigate medico-legal risks.
Mitigation Strategy:
We are building Naree with a compliance-first approach, including early regulatory planning, clinician-in-the-loop design, robust documentation (QMS), and phased validation to ensure safe and scalable deployment. |
If we scale 10x, the main stress points in Naree’s AI system would not be the core model alone, but the data quality, infrastructure, and operational reliability around it. First, a much larger and more diverse user base can increase noise, missing data, inconsistent symptom reporting, and variation across geographies, languages, and clinical settings, which can reduce model accuracy and calibration if not actively monitored. Second, infrastructure pressure rises significantly: ingestion pipelines, cloud storage, API response times, dashboard performance, and real-time scoring workflows must all handle far higher volume without latency or downtime. Third, model drift becomes a bigger risk as the incoming population mix changes, meaning a model trained on one cohort may perform differently at larger scale unless continuously revalidated. Fourth, clinical workflow integration can become a bottleneck, because more partner clinics mean more heterogeneous systems, onboarding complexity, and variable data standards. Finally, governance becomes harder at scale: privacy, access control, auditability, and compliance monitoring all become more complex. Our mitigation is to build for scale with modular architecture, strong data validation layers, monitoring for drift and bias, phased rollouts, and clinician-in-the-loop oversight so that performance, trust, and safety are maintained as adoption grows. |
Yes, Naree has an in-house AI/ML capability supported by a multidisciplinary team combining clinical expertise and technical development.
Our team includes AI/ML engineers and full-stack developers responsible for building the data pipelines, model development, and integration of AI into the app and EPR system. Their expertise spans data engineering (data ingestion, cleaning, and structuring), machine learning model development, and deployment within secure cloud environments.
This is complemented by strong clinical leadership, with domain expertise in obstetrics, gynaecology, and fertility care, ensuring that model design, feature selection, and outputs are clinically relevant and aligned with real-world workflows. This clinician-in-the-loop approach is critical for developing trustworthy and usable AI in healthcare.
Additionally, we leverage external advisors and collaborators when needed for specialised areas such as advanced modelling, regulatory strategy, and validation studies.
This combination of in-house technical capability + deep clinical expertise enables us to rapidly iterate, validate, and deploy AI features within a real-world healthcare setting. |
Naree primarily uses a combination of proprietary (owned) real-world data and curated clinical knowledge sources.
1. Proprietary / Owned Data:
Our core dataset is generated through the Naree platform itself, including:
• Longitudinal patient-reported data (menstrual cycles, symptoms, lifestyle inputs)
• Clinical data from the EPR (consultations, investigations such as blood tests and scans, treatments, outcomes)
• Consent-driven, real-world usage data from deployed environments
This data is owned or controlled by Naree under appropriate user consent and data governance frameworks, and forms our primary competitive advantage.
2. Licensed / Public / Reference Data:
• Clinical guidelines and evidence-based medical literature (used to inform rule-based logic and feature design)
• Publicly available datasets and research studies (where required for benchmarking and initial model development)
We do not rely on third-party proprietary datasets as a core dependency, which reduces licensing risk and increases long-term scalability.
3. Data Governance:
All data used for AI development is consented, de-identified/pseudonymised, and handled in compliance with applicable regulations, ensuring both ethical use and regulatory readiness.
This hybrid approach ensures that our models are built on high-quality, real-world, and continuously evolving datasets, while maintaining full control over our core data assets. |
We plan to improve model performance through a continuous learning and validation loop built into the Naree ecosystem. As more patients and clinics use the platform, we capture larger volumes of longitudinal, real-world data across symptoms, cycles, investigations, treatments, and outcomes, which allows us to retrain and refine models on more diverse and representative cohorts. We combine this with ongoing data cleaning, feature engineering, and periodic recalibration to reduce drift and improve accuracy over time. Performance is monitored using technical metrics such as AUC, sensitivity, specificity, precision, calibration, and subgroup analysis across different populations to detect bias or degradation early. We also incorporate clinician feedback and real-world outcome validation to ensure the models remain clinically relevant, interpretable, and safe. In practice, this means a structured cycle of data collection, monitoring, retraining, validation, and controlled deployment, rather than static one-time model development. |
Naree is building for both India and global markets, with a phased, India-first execution and global scalability by design.
In the near term, our primary focus is India, where there is a large, underserved population, rapidly increasing digital adoption, and significant gaps in women’s healthcare infrastructure—particularly in tier 2 and tier 3 cities. This provides a strong opportunity for rapid deployment, data generation, and product refinement.
At the same time, the platform is being architected to be globally scalable, with key features such as multilingual capability, compliance with international data standards (e.g., GDPR-aligned systems), and adaptability to different clinical workflows. We have already initiated early engagement in the UK market, particularly around our multilingual digital consent platform.
Our strategy is to build, validate, and optimise in India, and then scale internationally, leveraging our technology, data, and clinical insights to address similar gaps in other healthcare systems.
This dual approach allows us to combine India’s scale advantage with global market potential, positioning Naree as a globally relevant women’s health platform. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/6JTvxS7cuuka-Naree-India-1.3Mb.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/6JTvxS7cuuka-Naree-India-1.3Mb.pdf |
https://youtu.be/9WpDikE-wKw |
Naree is fundamentally a mission-driven, impact-focused startup aimed at transforming women’s healthcare from fragmented and reactive to continuous, data-driven, and equitable.
Our core mission is to improve early diagnosis, access to care, and health outcomes for women, particularly for underdiagnosed conditions such as endometriosis, PCOS, and infertility. Millions of women face delayed diagnosis, lack of awareness, and limited access to structured care—especially in underserved and tier 2/3 settings. Naree addresses this by combining digital infrastructure, education, and AI-driven insights to enable timely intervention and informed decision-making.
We are also deeply focused on health equity and accessibility, with multilingual support and scalable digital tools that reduce dependency on physical infrastructure, making quality care more accessible across diverse populations.
Beyond individual care, our platform aims to generate population-level insights that can inform research, public health strategies, and policy-making in women’s health.
While we are building a scalable and sustainable business, our primary driver remains measurable impact—reducing diagnostic delays, improving patient understanding, and enabling better clinical outcomes at scale. |
N/a |
linkedin |
checked |
| Apr 12, 2026 @ 10:56 AM |
safiqul haque |
sqhinnovationprivatlimited@gmail.com |
https://www.linkedin.com/in/safiqul-haque-82365b35 |
https://github.com/sqhinnovationprivatlimited-create |
9864018112 |
Inventor.director graduation |
We met through our shared interest in clean mobility and electric propulsion technologies, connecting during early-stage research and development discussions. We have been working together for over a year, collaborating closely on product design, prototyping, and strategy. |
All |
Our biggest strength as a team is our ability to combine innovative engineering with practical execution. We bring together strong technical expertise in electric mobility and energy systems, along with a shared vision to build scalable, real-world solutions. This allows us to move quickly from idea to prototype while staying focused on market needs and impact. |
Sqh innovation private limited |
WWW. SQHINNOVATION.COM/ |
kamrup |
Sqh innovation private limited is not a jast business innovation is World fast open challenge innovation only sqh innovation private limited |
We are solving the challenge of efficient and sustainable helicopter flight by developing an advanced AC/DC electric power system that reduces fuel dependency, lowers emissions, and enables faster, cleaner, and more reliable aerial mobility. |
An advanced helicopter powered by a hybrid AC/DC electric propulsion system, enabling efficient, low-emission flight. The solution integrates high-density batteries, smart power management, and electric rotors to reduce fuel dependence, lower operating costs, and improve sustainability in aerial mobility. |
Our solution is unique due to its hybrid AC/DC power architecture for aerial mobility, enabling higher efficiency, reduced energy loss, and faster response compared to conventional systems. Unlike traditional fuel or purely battery-based helicopters, our design integrates intelligent power management and scalable electric propulsion, making it cleaner, more cost-effective, and adaptable. This combination of novel energy integration, system optimization, and early-stage innovation creates strong technical defensibility and differentiation in the emerging electric aviation market. |
Idea |
Pilots |
Our target customers include government defense and emergency services, commercial aviation operators (cargo, logistics, and aerial transport companies), and future urban air mobility providers. We also target R&D organizations and aerospace manufacturers seeking electric propulsion solutions. |
2 |
Na |
SOM (Serviceable Obtainable Market): In the near term, our SOM focuses on early adopters in India and select international pilots—EV infrastructure partners, aviation R&D firms, and specialized logistics/drone/helicopter operators. We estimate initial penetration in niche electric aviation and advanced propulsion segments where electric vertical flight and hybrid systems are being tested and adopted first. |
Our revenue model is based on multiple streams: (1) direct sales of electric propulsion/helicopter systems, (2) subscription for maintenance and software monitoring, (3) charging/service network fees, and (4) licensing our AC/DC power and control technology to aviation partners and OEMs. |
Your main competitors depend on the exact stage of your AC/DC electric or hybrid helicopter concept, but they generally fall into these groups: 1. Traditional helicopter manufacturers Airbus Helicopters Bell Helicopter Leonardo Helicopters Sikorsky (Lockheed Martin) 2. Electric / hybrid VTOL & eVTOL companies Joby Aviation Lilium Archer Aviation Volocopter 3. Emerging electric propulsion innovators Vertical Aerospace EHang These companies are working on reducing fuel use, improving efficiency, and developing electric/hybrid vertical flight systems—similar to your AC/DC powered helicopter concept. |
We acquire customers through a mix of direct outreach, strategic partnerships, and pilot deployments. For early adoption, we focus on B2B collaborations (aviation operators, logistics, and infrastructure partners) and government/defense tenders. We also build credibility through demonstrations, pilot projects, and industry events. In parallel, we use digital marketing and technical content to reach engineering-driven buyers, and convert interest into trials, then long-term contracts. |
Our go-to-market strategy starts with pilot deployments in niche industrial and defense-adjacent use cases (survey, inspection, and logistics) to validate performance and safety. We then partner with aviation service providers and OEMs for distribution, followed by scaling into commercial and urban air mobility markets through certifications and strategic government collaborations. |
Our long-term vision is to build a new era of electric aerial mobility—developing safe, efficient AC/DC-powered hybrid electric helicopters and supporting charging ecosystems that make sustainable flight and transportation widely accessible, affordable, and scalable globally. |
Sqh innovation private limited |
Na |
No |
Na |
During the programme, we aim to refine and validate our AC/DC-powered electric helicopter concept, build and test a working prototype, and strengthen system safety and efficiency. We also want mentorship to achieve certification readiness, secure pilot customers, and prepare for early-stage funding and industry partnerships. |
Message Yes, we are open to both hybrid and virtual participation. However, we would prefer in-person engagement when required for prototyping, testing, and key mentorship sessions, as it would help us progress faster. |
Yes |
Supporting feature |
Here’s a clear, startup-ready description of a typical AI architecture suitable for an autonomous/connected system (e.g., EV charging automation or AC/DC-powered aerial systems):
Our AI architecture is built on a **modular |
We are building a proprietary dataset from our prototype testing and simulations, including AC/DC hybrid propulsion performance, energy consumption profiles, and flight stability metrics under varying load and environmental conditions. Over time, this real-world flight and bench-test telemetry becomes a closed-loop dataset that improves our control systems and efficiency optimization models. We also accumulate structured failure-mode data that is not publicly available in aviation electrification systems. |
A strong AI defensibility story usually doesn’t come from the model itself—it comes from the data, system integration, and operational feedback loop around it. For a deep-tech system like autonomous/assisted aviation or energy control, your defensibility can be framed like this:
1. Proprietary real-world operational data (most important)
As your system operates (flight/charging/control), you generate high-value telemetry: load patterns, environmental conditions, failure cases, and performance data. This kind of “real physics + real operations” dataset is extremely hard for competitors to replicate without deploying at scale.
2. Closed feedback loop (AI improves with usage)
Your AI doesn’t stay static—it continuously learns from deployments. Each mission/flight/charge improves prediction accuracy, control stability, and efficiency. This creates a compounding advantage over time.
3. Deep hardware–AI integration
If your AI is tightly coupled with AC/DC power systems, avionics, or control electronics, it becomes part of the system architecture—not just a software layer. Competitors can’t easily swap it out or replicate performance without redesigning hardware.
4. Simulation + real-world calibration engine
A high-fidelity digital twin (flight + power systems) trained and corrected with real data becomes a strong moat. Competitors may simulate, but without real calibration data, their models remain less accurate.
5. Safety, certification, and operational tuning data
In aviation especially, compliance and edge-case safety data is a major barrier. Your AI becomes “certification-shaped” through accumulated validation evidence, which is hard and slow to replicate.
In short:
The defensibility is not “our AI model is better,” but “our AI is trained on exclusive operational reality and embedded into a continuously improving physical system that others can’t easily access or replicate.”
If you want, I can tailor this specifically for your AC/DC powered helicopter concept or compress it into a 300-character application answer. |
1. Core performance metrics
Classification
Accuracy (overall correctness)
Precision / Recall (false positives vs false negatives)
F1-score (balance of precision & recall)
ROC-AUC / PR-AUC (ranking quality, especially for imbalanced data)
Confusion matrix (error breakdown)
Regression
MAE (Mean Absolute Error)
MSE / RMSE (penalizes large errors)
R² (variance explained)
Ranking / Recommendation
NDCG (Normalized Discounted Cumulative Gain)
MAP (Mean Average Precision)
Hit rate / Recall@K
Generative / LLM-style models
BLEU / ROUGE / METEOR (text overlap, older NLP metrics)
Perplexity (language modeling quality)
Human evaluation (fluency, relevance, correctness)
LLM-as-judge scoring (structured rubric-based evaluation)
2. Reliability & calibration
These measure whether predicted probabilities can be trusted:
Calibration error (ECE / MCE) – how close predicted probabilities are to actual outcomes
Brier score – accuracy of probabilistic predictions
Reliability diagrams (visual check of calibration)
3. Robustness & generalization
Performance on holdout / test sets
Cross-validation stability
Out-of-distribution (OOD) testing
Adversarial robustness (noise, perturbations, edge cases)
Slice-based evaluation (performance across subgroups / conditions)
4. Operational reliability (production metrics)
Latency (response time)
Throughput (requests per second)
Uptime / availability
Failure rate / timeout rate
Drift detection (data drift + concept drift over time)
5. Safety & quality (especially for AI systems)
Hallucination rate (for generative models)
Toxicity / policy violation rate
Factual accuracy checks
Human review pass rate
Summary
A strong evaluation setup doesn’t rely on one metric—it combines:
Predictive performance (accuracy-type metrics)
Probability trustworthiness (calibration)
Generalization (robustness/OOD)
Production behavior (latency, drift, failures) |
1. Data privacy by design
We collect only the minimum data required for the specific function (data minimization).
Wherever possible, we anonymize or pseudonymize user and operational data.
Clear separation between personal data, operational data, and analytics data.
2. Secure data storage & encryption
Data is encrypted both in transit (TLS 1.2+) and at rest (AES-256 or equivalent).
Sensitive credentials are stored in secure vaults with restricted access.
Regular key rotation policies are enforced.
3. Access control & internal security
Role-based access control (RBAC) ensures employees only access what they need.
Multi-factor authentication (MFA) for all critical systems.
Audit logs track all data access and changes for traceability.
4. Compliance alignment
We design systems to align with applicable regulations such as India’s DPDP Act, GDPR (where relevant), and industry-specific aviation/IoT safety standards.
Third-party services are evaluated for compliance and security posture before integration.
5. Secure development lifecycle (SDLC)
Security reviews at every stage of development (design, code, deployment).
Regular vulnerability scanning and penetration testing.
Dependency monitoring to prevent supply-chain risks.
6. Monitoring & incident response
Continuous monitoring for anomalies, breaches, and unauthorized access.
Defined incident response plan with containment, notification, and remediation steps.
Regular security drills and backups for disaster recovery. |
y regulatory risks include:
1. Aircraft certification (airworthiness)
In India: DGCA approval is mandatory for design, manufacturing, and operation
You must prove structural safety, redundancy, and flight reliability
For electric propulsion, certification frameworks are still evolving, which can slow approval
2. Battery & electrical safety regulations
High-voltage systems must meet strict fire, thermal runaway, and crash safety standards
Lithium battery certification and aviation-grade compliance (very strict testing)
3. Airspace & operational permissions
Flight permissions for test flights and commercial operations
Restrictions on low-altitude urban flying or VTOL corridors (if applicable)
4. Noise and environmental compliance
Rotor noise limits in populated areas
Environmental impact assessments for commercial operations
5. Pilot licensing & operational rules
Requires DGCA-approved pilot training and licensing for the specific aircraft type
Maintenance personnel certification rules
6. Import/export and component approval
If using foreign avionics, motors, or batteries, each may require approval and documentation
7. Emerging tech uncertainty
Electric aviation / hybrid AC-DC propulsion systems may not fit neatly into existing certification categories
Regulators may require case-by-case approval, increasing timelines |
1. Inference latency & cost blow-up
At 10× traffic, GPU/compute demand becomes the first constraint. Without aggressive batching, quantization, and caching, latency increases and unit economics degrade quickly.
2. Data pipeline saturation
Feature pipelines and real-time ingestion often break before the model does — queue backlogs, delayed features, and stale inputs can degrade output quality even if the model is unchanged.
3. Model serving stability (autoscaling limits)
Kubernetes/autoscaling systems can struggle with sudden spikes, leading to cold starts, uneven load distribution, or GPU scheduling inefficiencies.
4. Memory & context constraints
If the system relies on long context windows or retrieval (RAG), vector DB queries and embedding lookups become a bottleneck under high concurrency.
5. Monitoring blind spots
At 10× scale, edge cases multiply. If observability isn’t strong, you see “everything is working” while actual failure modes (drift, bias, silent degradation) go undetected.
6. Feedback loop degradation
Human feedback or retraining pipelines often can’t keep up. This causes slower adaptation and model staleness.
7. Cost structure collapse risk
Even if everything “works,” margins can break if inference cost per request isn’t tightly controlled.
In short: the model usually holds up better than the surrounding system. The real 10× failure is almost always infrastructure + data freshness + cost control, not raw intelligence. |
s. We have a small in-house AI/ML team focused on applied machine learning for autonomous systems and optimization problems.
Expertise includes:
Deep learning (PyTorch/TensorFlow) for predictive modeling and control systems
Computer vision for navigation, object detection, and environment awareness
Optimization algorithms for energy efficiency and routing
Sensor fusion (combining GPS, IMU, LiDAR/vision data) for real-time decision making
Edge AI deployment for low-latency onboard inference |
types:
1) Proprietary / owned data
Field telemetry (charging sessions, battery performance, usage logs)
Customer/operator data
Hardware sensor data (voltage, current, thermal, fault logs)
Simulation outputs generated in-house
👉 This is usually the most valuable and defensible part, and fully owned by the company.
2) Licensed third-party data
Grid and energy market data (utility APIs, paid providers)
Weather data (aviation-grade or commercial APIs)
Mapping / geospatial datasets (e.g., licensed GIS data) 👉 Usage is governed by contracts (subscription or enterprise license), not owned.
3) Open-source / public datasets
Government aviation or energy datasets
Academic datasets (battery degradation, charging behavior, etc.)
Open simulation environments
👉 These are usually free but come with usage/license terms (e.g., Apache, MIT, research-only restrictions).
4) Synthetic data (increasingly important)
Simulated flight/charging scenarios
Edge-case failure generation
Digital twin environments
👉 Fully owned, generated internally, and often used to reduce reliance on real-world rare events. |
1. Data flywheel (most important)
Continuously collect new real-world data (including edge cases and failures)
Improve data quality via cleaning, deduplication, and labeling checks
Actively mine “hard examples” where the model performs poorly
Expand dataset diversity to reduce bias and improve generalization
2. Strong evaluation system
Maintain fixed benchmark datasets for consistent comparison over time
Add task-specific metrics (accuracy, F1, precision/recall, latency, calibration, etc.)
Include “real-world evaluation sets” that mirror production usage
Track regression tests to ensure new updates don’t break old capabilities
3. Continuous training pipeline
Periodically retrain models with updated datasets (batch or rolling updates)
Use fine-tuning or domain adaptation for new use cases
Apply techniques like active learning to prioritize informative samples
Experiment with architecture improvements and hyperparameter tuning
4. Production feedback loop
Monitor live model performance (drift detection, error rates, user feedback)
Log failures and route them back into training data
Use A/B testing to validate new model versions before full rollout
5. Model robustness & safety improvements
Adversarial testing and stress testing
Regular bias and fairness audits
Calibration improvements so confidence scores reflect reality
6. Infrastructure for iteration speed
Automated MLOps pipeline (training → evaluation → deployment)
Versioning for data, models, and prompts
Fast rollback if performance degrades |
Both.
We are initially focusing on India as the primary launch market due to regulatory alignment opportunities, infrastructure needs, and cost advantages for early deployments. Once validated, we plan to expand to global markets—especially regions with strong demand for electric aviation, short-haul air mobility, and sustainable transport solutions. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/RXlqW6jEcK5o-SQH_Innovation_Halicopter_PitchDeck1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/RXlqW6jEcK5o-SQH_Innovation_Halicopter_PitchDeck1.pdf |
http://www.sqhinnovation.com |
Yes, we are building a mission-driven, impact-focused startup.
Our focus is on developing an AC/DC-powered electric aviation system (starting with helicopters) to reduce fossil-fuel dependence, cut emissions, and make short-range air mobility cleaner, safer, and more affordable. We aim to transform sustainable aviation in India and beyond while improving energy efficiency and accessibility in transport. |
Yes |
Self |
checked |
| Apr 11, 2026 @ 11:41 PM |
Anurag |
sparkyaitech@gmail.com |
|
|
08123059031 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 11, 2026 @ 9:49 PM |
SUREN KUMAR SELVAMANI |
sortyx2024@gmail.com |
https://www.linkedin.com/in/surenkumar-selvamani-pmp-52462918/ |
https://www.sortyx.com/ |
00919962269693 |
CEO AND FOUNDER |
The founding team came together through a shared passion for solving real-world environmental challenges using technology. We connected through professional and academic networks, where our complementary strengths in engineering, AI, and operations aligned around the vision of transforming waste management.
We have been working together for over a year, collaborating closely on product development, pilot deployments, and business strategy. During this time, we have built strong trust, clear role alignment, and a shared commitment to scaling Sortyx into a high-impact, technology-driven venture. |
1 |
Our biggest strength as a team is the combination of deep domain expertise and execution capability. We bring together strong engineering experience, AI and data analytics skills, and real-world operational knowledge in infrastructure and waste management.
This allows us to not only design innovative solutions but also implement them effectively in complex, real-world environments. We move quickly from concept to deployment, validate through pilots, and continuously improve using data-driven insights.
Equally important, we share a clear vision, strong commitment, and complementary skill sets, enabling us to make fast decisions, adapt to challenges, and scale efficiently as a cohesive unit. |
SORTYX |
www.sortyx.com |
CHENNAI |
AI IOT |
Sortyx is solving the inefficiency and lack of intelligence in current waste management systems. Today, waste collection is largely manual, reactive, and poorly optimized—leading to overflowing bins, unnecessary collection trips, high operational costs, and low recycling rates. Additionally, lack of segregation at source results in most waste ending up in landfills, causing environmental damage and lost resource value. Our solution addresses these gaps by bringing real-time visibility, automated waste identification, and predictive analytics into the system. This enables smarter collection, improved segregation, reduced costs, and better sustainability outcomes—transforming waste management from a reactive process into a proactive, data-driven ecosystem. |
Sortyx is an AI-powered smart waste management solution that combines IoT-enabled hardware with advanced data analytics to create an intelligent, end-to-end waste ecosystem.
Our system uses smart bins equipped with sensors and computer vision to monitor fill levels, identify waste types, and enable real-time segregation insights. The data is transmitted to a cloud-based platform where AI algorithms analyze patterns, predict collection needs, and optimize pickup routes—reducing costs and improving efficiency.
A centralized dashboard provides municipalities and enterprises with actionable insights, including waste generation trends, recycling performance, and ESG metrics. The solution is modular, scalable, and easy to deploy across cities, campuses, and commercial spaces.
By integrating AI, IoT, and data intelligence, Sortyx transforms traditional waste management into a proactive, efficient, and sustainable system. |
Sortyx stands out through its deep integration of AI, IoT, and data intelligence into a single, end-to-end platform, rather than offering isolated hardware or software solutions. Our key differentiators include: AI-driven waste identification & predictive analytics – Not just monitoring, but actively improving segregation, forecasting demand, and optimizing operations. Full-stack ecosystem – Proprietary smart bins, cloud platform, and analytics engine working seamlessly together. Data moat – Continuous data collection builds a unique dataset on waste patterns, enabling better models, higher accuracy, and long-term competitive advantage. Scalable & modular design – Easily deployable across cities, enterprises, and global markets with minimal infrastructure changes. ESG & monetization layer – Integrated reporting, carbon tracking, and recycling partnerships create additional value streams. This combination makes Sortyx not just a product, but a data-driven infrastructure platform that becomes more valuable and defensible as it scales. |
MVP |
Revenue |
AIRPORT BUSINESS UBIT |
NA |
NA |
NA |
NA |
BELL |
ONLINE MARKETING |
NA |
Our long-term vision is to build Sortyx into a global, AI-driven resource management platform that transforms how waste is perceived, managed, and monetized. We aim to move beyond waste collection and become a key enabler of the circular economy—where waste is intelligently tracked, segregated, and converted into valuable resources.
We envision creating a connected ecosystem across cities, industries, and supply chains, powered by real-time data, automation, and predictive intelligence. Over time, Sortyx will evolve into a digital infrastructure layer for sustainability, integrating carbon tracking, recycling networks, and smart resource optimization.
Ultimately, our goal is to reduce landfill dependency to near zero, drive measurable environmental impact, and position Sortyx as a global leader in smart, sustainable, and data-driven waste and resource management solutions. |
INCOPORATED |
100000 |
Yes |
We are applying to Avinya to accelerate Sortyx’s growth through strategic mentorship, ecosystem access, and early-stage support tailored for impact-driven startups. Avinya’s focus on innovation, sustainability, and scalable solutions aligns strongly with our mission to transform waste management using AI and data intelligence. We believe Avinya can help us refine our go-to-market strategy, strengthen partnerships with municipalities and enterprises, and access a network of investors and industry experts. The program will also support us in validating and scaling our technology across new markets. Beyond funding, we value the guidance and collaborative environment Avinya offers, which will be critical in helping us evolve Sortyx into a globally scalable, high-impact solution for smart and sustainable waste management. |
During the programme, we aim to achieve three key outcomes: Pilot & Market Validation – Secure and execute multiple paid pilot deployments with municipalities, metro stations, and enterprise campuses to validate product-market fit and demonstrate measurable impact. Product & Technology Enhancement – Strengthen our AI models for waste identification and predictive analytics, while refining hardware reliability and scalability for large-scale deployments. Go-to-Market & Fundraising Readiness – Build a strong sales pipeline, establish strategic partnerships, and prepare for our next funding round with a clear growth roadmap and validated traction. Overall, we want to transition from an early-stage solution to a proven, scalable business with strong commercial traction and investor readiness. |
Yes, we are fully open to hybrid or virtual participation. While we value in-person engagement for collaboration and networking, we are equally comfortable working in a virtual setup and can actively participate, contribute, and deliver outcomes remotely. Our team is adaptable and equipped to ensure seamless communication and progress in either format. |
Yes |
Automation layer |
Sortyx’s AI architecture is designed as a scalable, edge-to-cloud intelligence system that enables real-time decision-making and continuous learning.
1. Edge Layer (On-Device Intelligence)
Smart bins are equipped with cameras and sensors. Lightweight computer vision models run on edge devices to classify waste (wet, dry, recyclable) and detect fill levels in real time, reducing latency and bandwidth usage.
2. Data Ingestion & IoT Layer
Sensor and image data are transmitted via secure IoT protocols to the cloud. A streaming pipeline ensures real-time data flow, device management, and fault monitoring.
3. Cloud AI & Analytics Layer
Centralized ML models perform advanced analytics—predictive fill-level forecasting, route optimization, anomaly detection, and waste pattern analysis. Models are continuously retrained using aggregated data.
4. Application Layer
A cloud dashboard provides insights, ESG metrics, alerts, and operational recommendations to users.
5. Feedback Loop
Continuous data collection improves model accuracy, creating a self-learning, adaptive system and a strong data-driven moat. |
Sortyx is building a strong proprietary data advantage through continuous, real-world waste intelligence collected from our deployed systems.
Our smart bins generate high-frequency, multi-modal data—including images (waste type), sensor data (fill levels, weight, usage patterns), geolocation, and time-series behavior. This creates a unique dataset on waste generation, segregation behavior, and disposal patterns across different environments (metros, campuses, urban zones).
Over time, this data enables:
More accurate AI models for waste classification and prediction
Location-specific optimization (custom collection schedules, route efficiency)
Behavioral insights to improve segregation at source
Benchmarking & ESG reporting datasets for enterprises and cities
Since this data is difficult to replicate without physical deployments and long-term usage, it creates a compounding data moat, improving our models and strengthening our competitive advantage as we scale. |
Our AI component is defensible because it is tightly coupled with proprietary data, real-world deployment, and continuous learning loops, making it difficult to replicate.
Proprietary, real-world dataset – Our models are trained on high-frequency, multi-modal data (images, fill levels, usage patterns) collected from actual deployments. Competitors without hardware presence lack this depth and diversity of data.
Edge + Cloud co-optimization – We run optimized models on-device for real-time decisions while leveraging cloud AI for deeper analytics, creating a performance advantage in both speed and accuracy.
Continuous learning feedback loop – Every interaction improves model accuracy through retraining, leading to compounding intelligence over time.
Context-aware intelligence – Our models adapt to location-specific waste behavior, enabling better predictions and operational efficiency than generic solutions.
Full-stack integration – Hardware, AI models, and analytics platform are designed together, creating high switching costs and strong system-level defensibility.
This combination builds a data-driven moat that strengthens as we scale deployments and usage. |
We evaluate model performance and reliability across AI accuracy, system performance, and operational impact to ensure real-world effectiveness.
1. Computer Vision (Waste Classification)
Accuracy, Precision, Recall, F1-score
Confusion matrix (to track misclassification across waste types)
Latency (edge inference time)
2. Fill-Level Prediction & Forecasting
MAE (Mean Absolute Error), RMSE (Root Mean Square Error)
Forecast accuracy vs actual collection patterns
3. Route Optimization Models
Reduction in distance/time per route (%)
Fuel cost savings and service efficiency improvements
4. System Reliability (IoT + AI)
Uptime (%), data transmission success rate
Edge device performance and failure rates
5. Business Impact Metrics
Reduction in overflow incidents (%)
Improvement in segregation rate (%)
Increase in recycling yield (%)
We also implement continuous monitoring, anomaly detection, and periodic model retraining, ensuring models remain accurate, robust, and reliable in dynamic real-world environments. |
We follow a privacy-by-design and security-first approach across our entire platform to ensure data protection and regulatory compliance.
1. Data Minimization & Anonymization
We only collect essential operational data. Image data is processed at the edge, with sensitive or identifiable information either not stored or anonymized before transmission.
2. Secure Data Transmission & Storage
All data is encrypted in transit (TLS) and at rest (AES-256). We use secure cloud infrastructure with role-based access control (RBAC) and strict authentication mechanisms.
3. Compliance & Governance
We align with global best practices such as GDPR principles and relevant Indian data protection regulations, ensuring transparency, consent, and proper data handling.
4. Access Control & Monitoring
Granular access permissions, audit logs, and continuous monitoring help prevent unauthorized access and detect anomalies.
5. Device & IoT Security
Secure firmware, device authentication, and regular updates protect edge devices from vulnerabilities.
This ensures Sortyx maintains high standards of privacy, trust, and enterprise-grade security. |
Yes, there are some regulatory risks, but they are manageable and, in many cases, actually favor adoption of solutions like Sortyx.
1. Data Privacy Regulations
Laws such as India’s Digital Personal Data Protection Act (DPDP) and global frameworks like GDPR may impact how image and sensor data are collected and processed. We mitigate this through anonymization, edge processing, and strict data governance.
2. Municipal & Public Procurement Policies
Government tendering processes can be complex and time-consuming, potentially slowing deployments. However, compliance and early partnerships help us navigate this effectively.
3. Environmental & Waste Management Regulations
Changes in waste segregation rules, recycling mandates, or ESG reporting standards may require system updates—but these also create demand for our solution.
4. IoT & Device Compliance
Hardware must meet safety, telecom, and certification standards in different regions.
Overall, regulatory trends are moving in our favor, as governments push for smarter, more sustainable waste management systems. |
At 10× scale, the main risks are not the core AI itself, but data quality, system latency, and operational complexity. Key failure points include:
1. Data Drift & Model Degradation
Different cities and waste patterns can reduce model accuracy if not continuously retrained with localized data.
2. Edge Device Constraints
Higher deployments mean variability in hardware performance, leading to inconsistent inference speed and reliability.
3. Data Pipeline Bottlenecks
Massive increases in image and sensor data can strain ingestion, storage, and processing pipelines, causing delays or data loss.
4. Real-Time Decision Latency
As scale grows, maintaining low-latency predictions and alerts becomes challenging without optimized architecture.
5. Model Deployment & Versioning
Managing updates across thousands of distributed devices can create inconsistencies if not properly orchestrated.
Mitigation Strategy:
We address this through modular architecture, automated MLOps pipelines, edge-cloud balancing, continuous retraining, and scalable cloud infrastructure—ensuring the system remains robust and adaptive at scale. |
Yes, we have an in-house AI/ML team with strong cross-functional expertise in building and deploying real-world intelligent systems.
Computer Vision & Deep Learning – Designing and training models for waste classification using CNNs, object detection, and image segmentation techniques optimized for edge devices.
Machine Learning & Predictive Analytics – Developing models for fill-level forecasting, route optimization, and anomaly detection using time-series analysis and optimization algorithms.
Edge AI & Embedded Systems – Deploying lightweight, low-latency models on IoT hardware with focus on efficiency, reliability, and power optimization.
Data Engineering & MLOps – Building scalable data pipelines, model versioning, continuous training, and deployment frameworks to ensure consistent performance at scale.
Cloud & Analytics – Creating robust backend systems and dashboards for real-time monitoring, insights, and decision support.
The team is led by experienced engineers and supported by specialists like Ranjith (AI lead), enabling rapid innovation, deployment, and continuous improvement. |
1. Proprietary (Owned) Data
Data collected from our smart bin deployments: images of waste, fill-level sensor data, usage patterns, and geospatial/time-series data
This is our primary and most valuable dataset, continuously growing and fully owned by Sortyx
2. Curated Internal Datasets
Manually labeled waste images and classification datasets built by our team
Augmented datasets to improve model robustness across lighting, angles, and waste variations
3. Public & Open Datasets (for Pretraining/Benchmarking)
Waste classification and general image datasets used for initial model training and validation
These are used under appropriate open-source or academic licenses
4. Synthetic & Augmented Data
Generated variations to handle edge cases and improve model generalization
Overall, our competitive advantage comes from proprietary, real-world data, while public datasets are used to accelerate development and benchmarking. |
We improve model performance through a closed-loop, data-driven learning system that continuously adapts to real-world conditions.
1. Continuous Data Collection
Smart bins generate high-frequency image and sensor data, expanding our proprietary dataset across locations and use cases.
2. Active Learning & Human-in-the-Loop
Low-confidence predictions are flagged for manual review and labeling, improving dataset quality and model accuracy over time.
3. Automated Retraining Pipelines (MLOps)
We use scheduled and trigger-based retraining with version control, ensuring models stay updated with new patterns and data drift.
4. Edge-Cloud Feedback Loop
Edge devices capture real-time variations, while cloud models aggregate and learn globally—balancing speed and intelligence.
5. A/B Testing & Monitoring
We deploy model variants, track performance metrics (accuracy, latency, error rates), and continuously optimize.
6. Localization & Context Adaptation
Models are fine-tuned for region-specific waste patterns, improving accuracy in diverse environments.
This creates a self-improving AI system that becomes more accurate, efficient, and defensible as it scales. |
We are building for both India and global markets, with a phased expansion strategy.
India as the primary market – India offers a large and urgent opportunity due to high waste volumes, low segregation rates, and strong government push through initiatives like Smart Cities and Swachh Bharat. It is our core market for validation, pilot deployments, and scaling.
Global expansion (Southeast Asia, Middle East, and beyond) – Many emerging and developed markets face similar challenges and are actively investing in smart, sustainable infrastructure. Our solution is designed to be modular and adaptable to different regulatory and operational environments.
By proving scale and impact in India first, we aim to expand globally with a replicable, data-driven, and scalable model, positioning Sortyx as a global leader in intelligent waste management. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qGqRdXS5RXyi-SortyX-Smart-Dust-Bin-Pitch-Desk-_INDIA-_April-26_Social-alpha.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qGqRdXS5RXyi-SortyX-Smart-Dust-Bin-Pitch-Desk-_INDIA-_April-26_Social-alpha.pdf |
https://drive.google.com/file/d/1oswOhHFoVibp4AiPSMys6fXYsPiCu6AE/view?usp=drive_link |
Yes, Sortyx is a mission-driven, impact-focused startup at its core.
Our mission is to transform waste management into a smart, data-driven, and sustainable system, reducing landfill dependency and enabling a circular economy. We are addressing a critical environmental challenge—inefficient waste handling, poor segregation, and rising pollution—by using AI and IoT to create measurable impact at scale.
Our impact focus includes:
Environmental – Increasing recycling rates, reducing landfill waste, and lowering carbon emissions
Social – Supporting cleaner cities, improving public health, and enabling better working conditions in waste ecosystems
Economic – Reducing operational costs for municipalities and creating value from waste streams
We track impact through ESG metrics and data insights, ensuring transparency and accountability.
Ultimately, Sortyx is not just a business—we are building sustainable infrastructure for the future, where waste is managed intelligently and responsibly. |
As a team, we come from diverse backgrounds and bring varied experiences across engineering, operations, and innovation. While we may not strictly fall under a traditionally defined underrepresented category, we are deeply committed to inclusivity, equal opportunity, and diversity within our startup. We actively aim to build a team that includes young leaders, women in leadership (our CEO), and multidisciplinary talent, ensuring broader representation in decision-making and innovation. Our focus is on creating an ecosystem where talent and impact matter most, while also supporting and encouraging participation from underrepresented groups as we grow. |
Not referred (self-applied). |
checked |
| Apr 11, 2026 @ 9:24 PM |
RAKESH |
sangeetachilshetty@gmail.com |
https://www.linkedin.com/company/105285647/admin/dashboard/ |
https://agsync-ai.netlify.app |
9108257239 |
CEO ENGINEERING |
we have been there since our college days as we were doing our best at the college and were the classmates since the beggining of the solution that we have been doing it out there for the company and the culture of operational values to be indicated around the same with the same mindset and the same set of solution indi |
2 |
we are all known from many years since then and know the company based in bangalore we have been working on many of these projects since then for the future of the company and we will be doing it all to the point of view and the point of operational values over the company based so we have combination of AI, GIS based engineering and Agronomy with pathology engineers are on the board of the comapany which deliberately tell about the company based policy and the operational engineering of the company. |
AGSYNC |
https://agsync-ai.netlify.app |
Bengaluru |
SPACE APPLICATIONN IN AGRICULTURE |
1) 85% of farmers are not concern over climate change and climate based agriculture which could make climate resilient. 2) Only 20% of farmers make regenerative agriculture. 3) There is no data pipeline for the agriculture and climate based operation with respect to regenerative agriculture. 4) Every day 100 farmers 300 acre of land in every state quite farming and moves for commercialization of field. 5) 70% of farmers are not concern over land degradation and they over use fertilizer and pesticides no precise use for specific operation. 6) Nil pattern for carbon fixation which increase the soil fertility and crop betterment. 7) 80% of farmers dont diagnose the crop before buying any kind of pesticides its a total unorganized and non diagnotic. 8) The hike in 20% increase in input cost and then 30% of altered market. |
On sulution focused at first farmers and agronomist go round of consultancy by market price sense on CCN neural price, by crop deciding we do take the data of pedigree, location coordinate to the algorithm which is trained with POP LLM model, with interference of satellite sentinal 1 &2 with NDVI and climate API on geospatial technology with a very unique code of CCM ( Continous Climate/ crop monitoring ) which makes visible from seed sowing to till harvesting with respect to real time monitoring on satellite and computer vision analytics of the crop on disease with high resolution bandwidth on hyperspectral analysis and which will be stored at the backend with all the other information as a farmer profilee record which will help in regenerative agriculture carbon restore, will increase the yield and disease management every day farmers will recive the personalized information about the crop ( Hydration Capacity, disease analysis, crop health, Pest infestation management, weedicides contorl methods) from seed sowing to harvesting will be managed, alot of data will be managed at the end which will open up on multiple revenue model and multi dimension analysis open new market. as the revenue model where farmers will be charged on subscription model, D2C on agri input, insurance verification methods, credit access and market linkage and then the quality management which could go with exports for farmers and majorly the data sharing in agriculture, where we do focus on climate change and climate based operation majorly solving the core technology of the farmers and restoring the climate have reached 20000 farmers in the different parts world and testing the algorithm educating the farmers and have increased farmers income by 30% and climate restorartion data company has looking to operate in different parts of world and get access to all other exchange mode for farmers. |
As we do have the whole USP on for the farmer and the data consumer which predominantly executes our solution to the capital.Everyday the farmers will get the customized solution which will indicate the water stress level, crop disease report, yield quality and authorization.The disease is analysed over 3 meter resolution with high multispectral band which predominantly identifies the whole disease and helps in improving the quality and crop loss. The everyday solution also Contains prescribed medicines which solves the whole crop problem with climate as a concern. Farmers will get the solution upon whole customized and also that data will help in insurance pull for the farmers The current market price and also helping in market linkage will wholly help the farmers to I provide solution to their crops |
Revenue |
Revenue |
1) Farmers are our first customer and then we do have a subscriptions based model as B2C 2) e commerce companies and manufacturing units to sell out on the pre and post data on agri input as a cross selling. 3) We do make out through providing the API on banks as a customer who can make the money over it. 4) We do operate as SaaS to the government bodies and other agencies. |
$80 |
$70 |
$65 |
1) Subscription based 1 day 1 rupee per crop for each acre of land. 2) D2C operation based on cross selling the product of the same on the agri input market. 3) API on insurance verification and customer build model. 4) Farmers profile for the financial institution on the performance index of every farmers. DaaS based model over data sharing in the agriculture industry. |
We do have the competitors as Satsurex Satyukt Analytics, few more players in the market but we as a company will have a advantage over 1) Giving every day information and insights for the farmers with 3 meters resolution meanwhile others are with 15 days. 2) We are building a farmers profile index which mainly creates the ecosystem for farm financing on data fetching farmers and field performance in the agriculture ecosystem around the globe. 4) We are building our own satellite which gives a high resolution and a high frequency data which our competator doesn't have with them. 3) we do have a ground data where we have increased the overall quality and quantity if the crop from past 1 year which created a profit of 20%. 4) Our costing is low and margin is high every farmers can adopt the solution and afford it we are giving less than a dollar a month to the farmers costing is the matter. |
Every day we do ahve the certail level of the system where we acquire the set of consumers by lineant approach and marketing based methods |
we do have a propar based GTM that we have mentioned in the pitch deck please reffer the same |
As a long term vision of the company we are going to be one of the farmers first company and also regarding the data based accuracy which we are going to target 2.5 Lack customers in the coming days and get the good revenue and increase the deep tech values of the company over the time. |
yes incroporated |
we have raised the good revenue based on the application that we do to our end consumer |
Yes |
we are looking for the funds and the mentoring |
we are looking for the expansion of the company based on the solution and the environment of the company to be improves as we are young but need some more mentoring and all other details for the follow up |
yes |
Yes |
Supporting feature |
My AI architecture, as explained by Rakesh and team, is based on the Transformer model, a type of deep learning system that helps in understanding and generating human language. It works using a concept called attention, which allows the model to focus on important words in a sentence and understand their relationships. When a user gives input, the text is first divided into smaller parts called tokens and then converted into numerical values so the system can process them. These values pass through multiple layers of the model, where patterns and meanings are learned. The system is trained on large amounts of data to predict the next word in a sentence and is later improved with human feedback to make responses more accurate and useful. Overall, this architecture helps the model to answer questions, generate meaningful text, and assist users effectively, although it does not have human thinking or awareness. |
Any “advantage” I have comes more from the scale, diversity, and quality of training data, along with improvements from techniques like reinforcement learning with human feedback (RLHF), rather than from exclusive ownership of proprietary information. I also don’t retain or learn from individual user conversations in real time, and I don’t have memory of personal data unless it is shared within the session or explicitly stored with permission. |
My AI component isn’t “defensible” because of a single secret advantage, but rather a combination of factors that are difficult for competitors to replicate together. First, the scale and quality of training—managed by organizations like OpenAI—involve large, diverse datasets along with careful curation and alignment using human feedback (RLHF). This produces more reliable, context-aware responses compared to models trained with less refinement.
Second, the underlying architecture based on the Transformer architecture is enhanced through continuous research, optimization, and fine-tuning techniques, which improve reasoning, safety, and coherence over time. These improvements are not just about the model itself, but also about training pipelines, evaluation methods, and safety systems that are built around it. |
Model performance and reliability are evaluated using a combination of quantitative metrics, human evaluation, and real-world testing. There isn’t a single metric—rather, multiple layers of evaluation are used to ensure the model performs well across different tasks.
First, language quality and accuracy are measured using standard NLP metrics such as perplexity (how well the model predicts text), BLEU and ROUGE scores (which compare generated text to reference outputs), and exact match or F1 scores for question-answering tasks. These help assess how fluent and factually correct the responses are.
Second, reasoning and task performance are evaluated using benchmark datasets and standardized tests. Models are tested on problem-solving, coding, and comprehension tasks to measure correctness and consistency. This helps ensure the model can handle complex queries, not just generate fluent text. |
Handling data privacy, compliance, and security involves a combination of system design, policies, and operational safeguards rather than any single mechanism.
First, data privacy is maintained by limiting how data is used and retained. I don’t have independent access to personal databases or private information, and I only process the data that users provide during a conversation. Sensitive data is not stored or remembered beyond the session unless explicitly enabled. Organizations like OpenAI also apply data minimization principles—collecting only what is necessary and avoiding unnecessary retention of personal information.
Second, compliance with regulations (such as GDPR or similar frameworks) is ensured through controlled data handling practices. This includes clear policies on how data is processed, stored, and protected, along with mechanisms for user consent, transparency, and data control. Systems are designed to align with regional and international data protection standards. |
Yes, there are several regulatory risks that could impact AI products like this, especially as governments are rapidly developing new rules for AI systems.
One major risk comes from data protection and privacy laws, such as the GDPR in Europe and similar frameworks emerging globally. These regulations restrict how personal data can be collected, stored, and processed. If an AI system does not comply with these requirements, it could face legal penalties or restrictions on deployment.
Another key area is AI-specific regulation, such as the EU AI Act. This law classifies AI systems based on risk levels and imposes strict requirements on high-risk applications, including transparency, documentation, and human oversight. Similar regulatory approaches are being considered in countries like India and the United States. |
Scaling an AI system 10× doesn’t just make everything bigger—it exposes weak points across infrastructure, model behavior, and operations. Several things are likely to “break” or degrade if not carefully managed.
First, infrastructure and latency become major bottlenecks. Serving many more users simultaneously requires massive compute, memory, and networking capacity. Without optimization, response times increase, costs rise sharply, and system reliability can drop due to overload or failures in distributed systems.
Second, cost efficiency becomes critical. Large models are expensive to run, and scaling usage 10× can make the system economically unsustainable unless there are improvements like model optimization, caching, or more efficient hardware utilization.
Third, model quality and consistency may degrade at scale. With more diverse users and queries, edge cases increase. This can lead to more hallucinations, inconsistent answers, or failures in handling niche or complex prompts that weren’t prominent at smaller scale. |
Yes—systems like me are developed and maintained by dedicated in-house AI/ML teams within organizations such as OpenAI. These teams are highly multidisciplinary, combining expertise across research, engineering, and applied domains.
The core AI research team focuses on advancing model architectures (like transformer-based systems), improving training methods, and developing new capabilities in reasoning, language understanding, and multimodal learning. They typically have strong backgrounds in machine learning, deep learning, mathematics, and computer science.
The applied ML and engineering teams work on turning research into scalable products. Their expertise includes large-scale model training, distributed systems, optimization, and deployment. They ensure the models run efficiently, reliably, and cost-effectively in real-world environments. |
I don’t have visibility into specific, named datasets being used at any given moment, and I don’t actively “use” datasets after training. However, during development, models like me are trained on a mixture of:
Publicly available data (e.g., websites, books, open datasets)
Licensed data (content that organizations have legal permission to use)
Human-created data (annotations, examples, and feedback generated by trainers)
Organizations such as OpenAI design these datasets to be diverse and high-quality, but they are not entirely “owned” in the traditional sense. Instead, they are a combination of:
Licensed content (legally obtained)
Open-source/public domain data
Proprietary internally generated data (like labeled training examples and feedback) |
Continuous improvement of model performance is an ongoing, multi-layered process that combines research, data, evaluation, and real-world feedback.
First, model training is regularly updated. Teams at organizations like OpenAI improve performance by retraining or fine-tuning models with better-quality datasets, more diverse examples, and updated information. This includes incorporating new data sources and refining how the model learns patterns.
Second, human feedback plays a major role. Techniques such as reinforcement learning with human feedback (RLHF) are used to align the model with user expectations—making responses more accurate, helpful, and safe over time.
Third, evaluation systems are continuously enhanced. New benchmarks, test cases, and automated metrics are developed to identify weaknesses (e.g., hallucinations, bias, reasoning errors). These insights guide targeted improvements. |
primarily india as of now |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/PxGZKDEhSbfv-Master-Pitch-5.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/PxGZKDEhSbfv-Master-Pitch-5.pdf |
https://drive.google.com/file/d/1bVPKiOzUeTTijIyvyjmhKgVD8Q4t1M52/view?usp=drivesdk |
Yes—this is best described as a mission-driven, impact-focused AI initiative rather than just a purely commercial product.
The core mission is to make advanced AI systems accessible, useful, and safe for real-world problem solving. Organizations like OpenAI emphasize building AI that benefits a broad set of users—students, developers, businesses, and researchers—rather than serving a narrow use case. The goal is not just technological advancement, but practical impact at scale.
From an impact perspective, the focus is on:
Democratizing access to knowledge – enabling users to learn, solve problems, and make decisions more effectively
Improving productivity – assisting with writing, coding, analysis, and planning
Supporting innovation – helping startups, researchers, and organizations build new solutions faster
Enhancing decision-making – providing insights across domains like agriculture, healthcare, education, and business
There is also a strong emphasis on responsible AI development. This includes prioritizing safety, reducing bias, protecting user privacy, and ensuring compliance with evolving regulations. The aim is to balance capability with accountability. |
no not as of now |
KITS KARNATAKA AND GOVERNMENT OF KARNATAKA |
checked |
| Apr 11, 2026 @ 8:26 PM |
MOHAMMAD JUNAID MOHAMMAD FAROOQUE |
chamtab.fajmas@gmail.com |
https://www.linkedin.com/in/md-junaid-b049b13b2?utm_source=share_via&utm_content=profile&utm_medium=member_android |
http://NA |
8788559816 |
Engeneering |
am the sole founder of this startup and have independently driven the idea, research, and early development. As the venture grows, I am actively looking to onboard aligned co-founders and team members with complementary skills |
1 |
Our biggest strength is the combination of deep problem insight, strong execution capability, and relentless focus. We understand the pain points of our target users at a grassroots level, and we are committed to solving them through continuous iteration and real-world validation |
TITI |
Na |
Mumbai |
LIVE COMMERCE |
We are addressing the lack of customer retention and data ownership for small and local businesses in India. While large platforms dominate customer relationships, local दुकानदार are left without access to their own customer data or tools to re-engage them. As a result, they lose repeat customers, face inconsistent revenue, and remain dependent on third-party ecosystems for growth If you want, next I can help you write |
We are building a digital platform that empowers local businesses to own and manage their customer relationships. Our solution enables shopkeepers to capture customer data, re-engage them through targeted notifications and offers, and drive repeat purchases without relying on third-party platforms. By combining simple tools with powerful insights, we help MSMEs increase customer retention, improve revenue predictability, and grow independently |
What makes us unique is our ground-level understanding of the challenges faced by local businesses and our focused approach to solving customer retention and data ownership—an area ignored by most existing solutions. While others focus on acquiring customers, we enable businesses to retain and re-engage them through owned channels. Our product is built for simplicity, ensuring even small दुकानदार can easily adopt and benefit, driving real and sustainable growth |
MVP |
Pilots |
Pan India every buissnes owners |
Na |
Na |
Na |
Na |
No one |
Data calling only I have |
Multiple when final mvp is ready |
“Our vision is to build a nationwide ecosystem that redefines how local businesses connect with customers—where every transaction leads to a long-term relationship owned by the business itself. We aim to empower millions of MSMEs to become digitally independent, unlock new revenue opportunities, and contribute significantly to India’s economic growth.” |
Na |
Na |
Yes |
Both are needed |
“During the programme, our primary goal is to achieve product-market fit by validating our solution with real local businesses and refining it based on continuous feedback. We aim to onboard and activate our first set of users, build a scalable and repeatable growth model, and strengthen our revenue strategy. Additionally, we want to leverage mentorship and network support to position ourselves strongly for early-stage funding and long-term scale.” |
Solo |
No |
Not applicable |
Na |
Na |
Na |
Na |
Na |
Na |
10 x |
Na |
Na |
Na |
. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/nA4uGoJYQgGO-final-pitch-dack-compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/nA4uGoJYQgGO-final-pitch-dack-compressed.pdf |
http://na |
“Yes, our startup is deeply mission-driven and focused on creating real impact at scale. We are working to solve a fundamental problem faced by millions of MSMEs in India—lack of customer ownership and retention. This directly affects their income stability and long-term survival.
Our goal is to digitally empower these businesses with simple and accessible tools so they can build lasting customer relationships, reduce dependency on large platforms, and grow sustainably. We believe that strengthening local businesses will not only improve livelihoods but also drive inclusive economic growth across the country.” |
Yes |
Friend |
checked |
| Apr 11, 2026 @ 8:15 PM |
raju kumar |
raju0009921@gmail.com |
https://www.linkedin.com/in/raju-kumar-a12910323/ |
https://www.pranagai.com |
8789169939 |
Founder with Engineering |
We met in college and through linkedin we are since 2025 on this project |
2 |
Coordination and understanding |
Petsgo & PetsDoor Pvt Ltd |
https://www.pranagai.com |
jaipur |
We are integrating ai with biotechspace |
We are making a integration of ai with biotech space to solve the problem of agricultural and material discovery through software where our application will be solving the time and cost of organization in cost an time and protect their extra effort and reduce their burden |
We are making a integration of ai with biotech space to solve the problem of agricultural and material discovery through software where our application will be solving the time and cost of organization in cost an time and protect their extra effort and reduce their burden. |
Our software which is capable of doing the research part and our pinn framework which will be our plus point hard to replicate from others |
Idea |
Users |
Organization, university, scholars and university, companies and government |
1Trillion Dollar |
500 Billion Dollar |
300 Billion dollar |
Subscription absed and ip royalty |
Ginkgo Bioworks , Zymergen |
Through advertisement, and freemium services and workshops |
na |
Building world largest research platform from prompt to lab ready prototype |
pvt ltd |
900000 |
Yes |
funding and networking |
Market Ready Product |
both |
Yes |
Automation layer |
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/diI26ge7HZi7-pranag-fl-1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/diI26ge7HZi7-pranag-fl-1.pdf |
http://na |
Impact Focused startup |
na |
na |
checked |
| Apr 11, 2026 @ 6:47 PM |
Anil Pandey |
anilpandey1706@gmail.com |
https://www.linkedin.com/in/anil-pandey-b52b7413?utm_source=share&utm_campaign=share_via&utm_content=profile&utm_medium=ios_app |
http://plexusdigitals.com |
+91 97111 21106 |
Founder, Chairman & Managing Director Engineer and MBA with 30+ years of leadership experience with multinational companies in robotics, digital transformation, and technology innovation. Former CXO with global transformation mandates across industries. |
"We were friends long before Plexus Digitals — shared entrepreneurial mindset and a common vision of building something impactful from India brought us together. Anil and the team collectively bring 80+ combined years of experience across technology, enterprise sales, BFSI, robotics, and digital transformation. That shared belief — that India deserves a world-class, India-originated XR platform — is what made us decide to build this together." |
All |
Based on the pitch deck, here's a strong answer:
"Our biggest strength is the diversity and depth of our combined expertise. Together, we bring 80+ years of cross-functional leadership across technology, robotics, digital transformation, BFSI, real estate, marketing, and enterprise development. Anil Pandey leads with 30+ years in engineering and global CXO mandates, Yogesh Bhatia brings deep entrepreneurial and marketing strategy experience, and Adhaar Bansal adds 25+ years of enterprise and strategic partnership expertise. This means we don't just build great technology — we understand governments, enterprises, and markets deeply. We're not a team of only engineers or only salespeople — we're a complete, battle-tested founding team that can execute across product, sales, finance, and government relations simultaneously." |
Plexus Digitals Private Limited |
plexusdigitals.com |
Lucknow, Uttar Pradesh, India |
Immersive AI & XR platforms built from India — transforming how governments, enterprises, and consumers experience culture, real estate, and digital marketing. |
India lacks a purpose-built, homegrown XR platform at global scale. Businesses suffer from poor digital engagement, ineffective visualisation of pre-construction projects, and wasteful marketing spend with no ROI transparency. |
Three integrated platforms — Divya Drishti (immersive spiritual & cultural XR), ARTE (real-time architectural visualisation for real estate & infrastructure), and DGCC (AI-powered marketing intelligence & lead generation) — forming one unified ecosystem solving real problems across multiple sectors. |
No competitor combines AI, full-stack XR, immersive visualisation, data analytics, and marketing automation in a single platform. Our Indian cultural knowledge layer is a globally unique intellectual asset, and our multi-sector SaaS architecture creates strong recurring revenue with exponential scaling potential. |
Revenue |
Users, Revenue, Pilots |
Central & State Governments (tourism, education, social welfare), Real Estate Developers & Infrastructure Planners, Enterprise & Hospitality (hotels, tourism, brands), Medium & Large Enterprises (AI-driven marketing), and Education & Cultural Tourism (EdTech, museums, cultural organisations). |
$300B — Global XR Market by 2030 (Source: PwC/Statista), within India's $1 Trillion Digital Economy target. |
$170B — Combining Immersive Tourism ($55B), AI Marketing ($105B), Real Estate PropTech ($86B), and India EdTech ($30B) segments. |
₹200 Cr+ ARR target by 2029, starting with Indian government contracts, enterprise SaaS, and real estate deployments. |
Five streams — Subscription SaaS (₹10K–20K/month), Platform Licensing (₹1.5L–3L/year), Project Implementation (₹5L–50L/project), Consulting & Advisory (₹2L–15L/engagement), and future B2C Consumer Digital Products. |
Durlabh Darshan (partial XR, no AI marketing), Generic VR Firms (single sector, no cultural layer), and Marketing Platforms (no XR capability). Plexus Digitals is the only player combining full-stack XR + AI + Indian cultural knowledge layer + multi-sector SaaS architecture. |
Direct engagement with government departments for high-value pilot contracts, enterprise sales force for DGCC subscriptions, ARTE licensing to real estate developers, referral networks, digital campaigns, and strategic partnerships with content creators, educational institutions, and technology integrators |
Four phased approach — Phase 1: Government & enterprise pilot contracts (Divya Drishti & ARTE) Phase 2: Enterprise SaaS rollout (DGCC subscriptions & ARTE licensing) Phase 3: Strategic alliances with content creators, EdTech, and technology integrators Phase 4: Consumer-facing app distribution & international expansion into South-East Asia and Middle East |
To build India's first globally scaled XR ecosystem — where ancient Indian wisdom meets modern AI and immersive technology — empowering billions across education, tourism, culture, and business. Target ₹200 Cr+ ARR by 2029 with IPO readiness evaluation and platform ecosystem lock-in across global markets. |
done |
NA |
Yes |
Plexus Digitals is at a critical inflection point — our platforms are prototyped, partnerships are in discussion, and we are ready to scale. We are applying to Avinya because we believe their ecosystem, mentorship, and network can accelerate our go-to-market execution, strengthen our government and enterprise outreach, and help us close our seed round faster. Avinya's programme aligns perfectly with our vision of building a globally impactful, India-originated technology platform. |
During the programme we aim to — close our first enterprise clients on DGCC, launch the Divya Drishti government pilot, secure at least two ARTE real estate deployments, close our seed funding round, and build a strong advisor and partner network that accelerates our expansion across India and into global markets. |
Yes, we are open to both hybrid and virtual participation and can adapt to programme requirements accordingly. |
Yes |
Core engine |
Our AI architecture combines machine learning models for lead scoring and validation, natural language processing for cultural knowledge delivery, real-time data analytics engines for campaign performance tracking, and computer vision integrated into our XR visualisation pipelines. DGCC operates as an AI-first marketing intelligence layer, while Divya Drishti uses AI to personalise immersive cultural experiences based on user behaviour and preferences. |
As we onboard enterprise and government clients, we accumulate proprietary datasets around lead behaviour, conversion patterns, and immersive engagement metrics — data no competitor currently possesses. Additionally our Indian cultural knowledge framework is a unique structured dataset built in-house, with no equivalent in the market. |
Three layers of defensibility —
Data moat: Proprietary engagement and lead conversion data that compounds with every client onboarded
Cultural knowledge layer: Unique structured Indian knowledge framework embedded into AI-driven experiences — impossible to replicate quickly
Integrated ecosystem: AI powers not just one product but three interconnected platforms, creating cross-platform intelligence that standalone competitors cannot match |
Key metrics include — Lead validation accuracy, Campaign ROI improvement percentage, User engagement duration within XR environments, Conversion rate uplift for enterprise clients, Model drift monitoring through periodic retraining cycles, and A/B testing frameworks for continuous optimisation. |
We follow India's Digital Personal Data Protection Act (DPDPA) 2023 compliance framework. Data is encrypted in transit and at rest, with role-based access controls. Government-grade deployments follow additional security protocols including data localisation requirements. Client data is never shared across accounts and anonymised for model training purposes. |
Potential risks include evolving DPDPA regulations around data localisation, government procurement policy changes affecting XR adoption timelines, and global data privacy regulations (GDPR) for international expansion. We are proactively building compliance-first architecture to mitigate these risks. |
At 10x scale the primary pressure points would be — real-time analytics processing latency in DGCC, cloud infrastructure costs scaling faster than revenue if not optimised early, and model retraining pipelines needing automation. We are building with horizontal scalability in mind from day one, using cloud-native microservices architecture to address these challenges proactively. |
Currently the founding team brings technology and transformation expertise. We are actively hiring XR engineers and AI/ML specialists as a priority use of seed funding — talent acquisition represents 23.4% of our fund utilisation plan. |
We are using a combination of — publicly available cultural and philosophical knowledge datasets (curated and structured in-house), licensed third-party real estate and market data for ARTE, and proprietary engagement data from early pilot interactions. Our cultural knowledge framework dataset is fully owned and built internally. |
Through — continuous client feedback loops, automated retraining pipelines triggered by performance degradation metrics, expanding proprietary datasets with every new client onboarded, regular benchmarking against industry standard models, and dedicated AI/ML team investment post seed funding closure. |
Both — India first, global next. We are establishing deep roots in India through government and enterprise contracts, then expanding to South-East Asia and Middle East by 2028, with full global platform ambition by 2029. Our cultural-tech positioning gives us a uniquely India-originated but globally relevant value proposition. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/ZsMlAsx3CcPj-PLEXUS-DIGITALS-PRIVATE-LIMITED.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/ZsMlAsx3CcPj-PLEXUS-DIGITALS-PRIVATE-LIMITED.pdf |
http://na |
re you building a mission-driven or impact-focused startup? Tell us more. *
This field is required. Please enter text.
Are you part of an underrepresented group in tech/startups? *
"NA" if not applicable
Referred by * |
Yes — we are founders building from Tier-2 India (Lucknow, Uttar Pradesh), a significantly underrepresented geography in India's startup ecosystem which remains heavily concentrated in Bangalore, Mumbai, and Delhi. We believe India's next wave of innovation will come from its heartland and we are proud to represent that movement. |
na |
checked |
| Apr 11, 2026 @ 9:48 AM |
Yogesh Meena, Devesh Meena |
yogesh@oolook.in |
https://www.linkedin.com/in/yogesh-m-8a07851a9/?skipRedirect=true |
http://Oolook.in |
6377528291 |
Yogesh Meena – Founder & CEO. Responsible for product development, technology architecture, and growth. Has experience in performance marketing and has worked with multiple clients across industries. Built Oolook as a full-stack AI product, combining technical development with deep understanding of marketing workflows and user behavior. |
Currently a single-founder startup. The product has been developed independently, with support from early collaborators and users contributing feedback during development and testing phases. |
1 |
Strong combination of technical execution and market understanding. The founder has built the product end-to-end while also having hands-on experience in marketing and user acquisition. This allows rapid iteration, efficient decision-making, and alignment between product development and real user needs. |
Oolook |
oolook.in |
Jaipur |
AI-native marketing platform that automates content creation, customer engagement, and performance optimization using brand-specific intelligence. |
Businesses rely on multiple disconnected tools for content creation, social media management, and analytics, leading to inefficiencies, higher costs, and fragmented workflows. Existing AI tools generate generic content and lack brand-specific understanding, resulting in inconsistent communication and poor marketing performance. |
Oolook is an AI-native platform that integrates content creation, engagement, and analytics into a single system. It builds a persistent “brand memory” using proprietary embeddings trained on business data. This allows the platform to generate personalized content, automate workflows, and continuously improve based on performance signals, enabling scalable and cost-efficient marketing. |
Oolook is built as an AI-native system rather than an add-on to existing tools. It uses proprietary brand embeddings to create a persistent memory of each business, enabling context-aware and personalized outputs. The system continuously improves using feedback loops from user interactions and performance data. Additionally, the platform integrates multiple workflows into one system, reducing fragmentation and creating strong product lock-in. |
Revenue |
Users, Revenue, Pilots, Signups |
Small and medium businesses, digital-first brands, and marketing agencies that rely on online channels for growth. These customers need cost-effective tools to manage content creation, engagement, and analytics without relying on multiple platforms or large marketing teams. |
₹3,50,000 Crore |
₹25,000 Crore |
₹500 Crore |
Subscription-based SaaS model with multiple pricing tiers. Revenue is generated through monthly and annual plans, along with usage-based pricing for advanced AI features. Additional revenue streams include enterprise solutions and white-label offerings for agencies. |
AI content tools such as Jasper, Copy.ai, and Writesonic; design and marketing platforms like Canva; and social media management tools such as Hootsuite and Buffer. These platforms solve parts of the workflow but lack integrated, AI-native systems with brand-specific learning. |
Customer acquisition is driven through product-led growth, organic content, and community-driven channels such as social media platforms and online forums. Early traction has been achieved without paid acquisition, supported by word-of-mouth and creator-driven distribution. |
Initial focus on SMBs, creators, and agencies through a freemium model and organic growth channels. As the product matures, expansion will include partnerships, integrations, and targeted outreach to mid-sized businesses and enterprises. |
To build a fully autonomous AI-driven marketing system that manages end-to-end digital marketing workflows for businesses. The vision is to replace fragmented tools with a unified platform that continuously learns, optimizes, and scales marketing efforts across channels. |
Private Limited |
NA |
Yes |
We are applying to Avinya to accelerate the development and scalability of our AI-native platform through structured mentorship and access to a strong ecosystem. The program’s focus on AI startups aligns well with our product, and we aim to benefit from technical guidance, strategic inputs, and potential collaborations to refine our system and prepare it for larger-scale deployment. |
During the programme, we aim to improve the core AI system, enhance model performance and scalability, and strengthen product-market fit. We also aim to onboard more users, validate key use cases, and prepare the platform for enterprise-level adoption. Additionally, we want to build strong industry connections and position the product for follow-on funding. |
Yes |
Yes |
Core engine |
Oolook is built on a modular AI architecture combining fine-tuned large language models with custom embedding pipelines. We use transformer-based LLMs fine-tuned on domain-specific marketing data. A brand embedding layer stores contextual brand information, enabling personalized generation. The system includes retrieval-augmented generation (RAG), feedback loops from user interactions, and orchestration pipelines for content, analytics, and automation workflows. |
We have built a large-scale dataset of social media content and marketing data used for training and fine-tuning models. Additionally, each user contributes unique brand-specific data, which is transformed into proprietary embeddings. Over time, this creates a compounding data advantage where each brand’s model becomes more accurate and personalized. |
Our defensibility comes from brand-specific embeddings, continuous learning from user interactions, and integration of multiple workflows into a single system. Unlike generic AI tools, our models improve over time with proprietary data and usage patterns, creating switching costs and long-term product lock-in. |
We evaluate performance using a combination of qualitative and quantitative metrics, including engagement rates, content performance (click-through, conversions), and user feedback. Internally, we track generation accuracy, relevance, and consistency with brand guidelines. |
User data is isolated at the account level and used only to improve that specific user’s outputs. We follow standard data security practices including controlled access, secure storage, and anonymization where required. No sensitive data is shared externally without consent. |
Potential risks include data privacy regulations and platform API restrictions. These are mitigated through compliance with data handling standards and by designing systems that do not rely on restricted or sensitive data sources. |
At scale, compute costs and latency for model inference can increase significantly. Additionally, maintaining real-time personalization across a large user base can become complex. We are addressing this through model optimization, caching strategies, and efficient infrastructure design. |
Currently led by the founder with strong technical involvement in building and fine-tuning models. Plans are in place to expand the AI/ML team as the product scales. |
We use a combination of publicly available social media datasets, internally curated datasets, and user-generated data. The proprietary dataset built from aggregated content is used for model fine-tuning and optimization. |
The system improves through feedback loops from user interactions, content performance data, and retraining pipelines. Continuous fine-tuning and optimization are used to improve accuracy, personalization, and efficiency over time. |
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/d8UUB35S3Mt8-Pitch.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/d8UUB35S3Mt8-Pitch.pdf |
https://www.youtube.com/watch?v=gMM7DnXFgLc |
Oolook aims to democratize access to high-quality digital marketing by making advanced AI tools accessible to small and medium businesses. Today, effective marketing is often limited to companies with large budgets and teams. By reducing costs and simplifying workflows, Oolook enables smaller businesses and creators to compete on a more level playing field, improving their ability to grow and participate in the digital economy. |
Founder from a Tier-2 city in India, building for emerging markets and underserved businesses. |
NA |
checked |
| Apr 10, 2026 @ 11:13 PM |
Mohamed Aejaz Ahamed |
aejazahamed57@gmail.com |
https://www.linkedin.com/in/aejaz-ahamed-aa449bb0/ |
http://NA |
9738547792 |
Founder, CEO and CTO, Product Development |
We are brothers.
I work as a solution engineer for a Saas company and brother is a computer science final year student. |
2 |
Our biggest strength is execution speed combined with product clarity.
We focus on identifying real user pain points and quickly turning them into working solutions. DevGuard was built as a functional MVP with a strong emphasis on real-time usability and simplicity, rather than over-engineering.
We also bring a strong product mindset to security, focusing on user behavior and experience. This allows us to build solutions that are not only technically sound but also practical and easy to adopt. |
Devguard |
NA |
Bangalore |
DevGuard is building the control layer for AI, ensuring every interaction is secure, governed, and auditable in real time. |
Organizations are rapidly adopting AI tools without any control over what data is being shared, leading to accidental leaks of sensitive information, lack of visibility, and compliance risks. DevGuard solves this by ensuring every AI interaction is secure, monitored, and governed in real time. |
DevGuard acts as a real-time control layer between users and AI systems, detecting sensitive data, masking or blocking unsafe inputs, and logging every interaction to ensure security, visibility, and compliance. |
Unlike existing solutions that focus only on detection or monitoring, DevGuard operates at the execution layer—actively controlling what data is sent to AI systems in real time. By combining detection, masking, policy enforcement, and auditability into a single layer, it moves beyond passive security to active governance of AI usage. |
MVP |
Pilots |
Our target customers are organizations adopting AI tools, including startups, mid-sized companies, and enterprises that need to ensure secure, compliant, and controlled usage of AI across their teams and applications. |
NA |
NA |
NA |
DevGuard follows a SaaS-based model with subscription plans for organizations, based on the number of users, and advanced security/compliance features. |
Nightfall AI, Lakera, and LayerX |
We plan acquire customers by targeting organizations adopting AI, focusing on security, compliance, and risk teams. We drive awareness through content around AI risks and data protection, combined with direct outreach and pilot programs. Adoption then expands across all employees through a simple, accessible interface that enables safe AI usage. |
NA |
Our vision is to power secure and compliant AI adoption across organizations by becoming the standard layer that enforces policies, manages risk, and provides full visibility into AI usage. |
MVP Stage |
NA |
Yes |
Avinya’s focus on responsible AI strongly aligns with our vision of building a control layer for AI. We are at a stage where guidance on building a secure, compliant, and scalable system is crucial, and we believe Avinya’s mentorship and ecosystem can help us shape both the product and its long-term impact. |
We aim to validate our product through pilots, refine our responsible AI approach, and build early enterprise partnerships. |
Yes |
Yes |
Core engine |
Our architecture consists of an input processing layer that uses AI models for sensitive data detection and classification, followed by a control layer that applies masking, blocking, and policy enforcement. This sits as a middleware between applications and LLM APIs, with an observability layer for logging, auditing, and risk tracking. |
As a security-focused product, we do not store or use raw customer data. Instead, we build a proprietary dataset of anonymized interaction patterns, detection signals, and risk behaviors. This allows us to continuously improve detection accuracy, masking strategies, and policy enforcement while maintaining strict privacy and compliance standards. |
While competitors may have access to similar data signals, DevGuard’s defensibility comes from its position in the execution layer, where it actively controls and enforces how data is sent to AI systems. Over time, this enables us to build deeply integrated policy and control systems within organizational workflows, making the solution more embedded and harder to replace than standalone detection or monitoring tools. |
As an early-stage product, we are currently building our evaluation and feedback mechanisms. We plan to incorporate real-world feedback loops such as user-reported issues, override behaviors, and system logs to identify missed detections and continuously improve reliability in production. |
We follow a privacy-first design where sensitive data is processed in real time and masked before being sent to AI systems, without storing raw inputs. We retain only anonymized metadata for audit and improvement purposes, implement secure data handling practices, and design the system to align with regulations such as GDPR and DPDP. |
DevGuard is not a high-risk AI system but operates within data privacy regulations, which we address through a privacy-first and minimal data retention design. |
The main challenges at scale would be latency, consistency of detection, and handling high concurrency, which we plan to address through scalable infrastructure and optimized pipelines. |
NA |
We currently use synthetic and publicly available datasets for testing, and do not use customer data. |
Our approach to improvement is driven by feedback loops from production usage, including anonymized signals, user feedback, and system logs. We will use these to identify false positives, false negatives, and edge cases, and iteratively refine our detection models, masking strategies, and policy enforcement logic. |
We are building for both India and global markets, beginning with compliance frameworks such as India’s DPDP Act and expanding to global standards like GDPR, ensuring the product is adaptable across regions. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/nUES4aqRMu4E-Devguard-Deck.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/nUES4aqRMu4E-Devguard-Deck.pptx |
https://drive.google.com/file/d/1U5jR9dgiZcH_litZRg8ICaUHT5hvGMLA/view?usp=sharing |
Yes, we are building a mission-driven startup focused on making AI usage safe and trustworthy.
As organizations increasingly adopt AI tools, sensitive data is being unknowingly exposed through prompts, with little to no control once it leaves the system. This creates serious risks around privacy, compliance, and data security.
Our mission is to build a safety layer for the AI-powered world that prevents such risks before they happen. Starting with real-time prompt security, we aim to ensure that individuals and organizations can use AI confidently without compromising sensitive information.
In the long term, we are working towards a broader communication security platform that protects both incoming threats and outgoing data across digital workflows, aligned with global standards like DPDP and GDPR. |
NA |
NA, found on Linkedin Karnataka IT BT Post |
checked |
| Apr 10, 2026 @ 9:27 PM |
Mohammed Saad, Aryan Rajput |
saad@climorisk.com |
https://www.linkedin.com/in/md-saad-mlr/ |
http://www.climorisk.com |
08904492336 |
Saad - CEO, Aryan - CTO |
A mutual friend introduced Aryan and me, recognizing our complementary skills. We quickly aligned on our vision and have been building ClimoRisk together for the past four months. |
2 |
Our biggest strength is our deeply complementary skill set, bridging scalable tech with climate science. Aryan is the technical engine, having built and shipped over 60 products for tech companies. I bring the domain authority with GARP SCR-certification and have executed carbon accounting and impact assessments for over 100 businesses. Aryan knows exactly how to build the software, and I know exactly what the market needs it to do. |
CLIMORISK |
www.climorisk.com |
Udupi, Karnataka |
Climate risk intelligence platform for BFSIs to identify and quantify financial risk from climate change |
Translating climate data into financial risk is too complex and resource-intensive for banks to build in-house, leaving them dependent on costly external consultants and rating agencies. |
We provide a flexible platform where banks can upload and analyze their credit portfolios, whether on a single-asset or bulk basis. To meet strict banking security requirements, we offer two deployment models: a standard SaaS subscription, and a premium, containerized deployment hosted directly within the bank’s own infrastructure. |
ClimoRisk is uniquely positioned by solving the three major flaws of current climate software: We calculate total risk: Competitors ignore transition risks (policy, carbon pricing). We account for both physical and transition risks. We own our data: Instead of packaging open-source data, we build proprietary datasets with 30m resolution, allowing for true asset-level adaptation planning. We speak the language of finance: Incumbents rely on opaque, subjective ESG scores (0-100). We translate climate science into transparent, actionable financial metrics like Climate VaR, EAL, PD, and LGD so banks can make concrete financial decisions. |
MVP |
Pilots |
BFSI, CREs, NBFCs or any industry which has lots of assets under its portfolio |
$4 billion |
$1.8 billion |
$70 million |
Annual Subscription, Infrastructure deployment |
Climate X, Jupiter Intelligence, Big 4 consultants, Bloomberg, MSCI |
Reaching out to CXOs within our network, cold emails |
We initially target mid-market banks and NBFCs (Non-Banking Financial Companies), which have shorter procurement cycles than Tier-1 institutions. |
Our long-term vision is to become the definitive climate risk infrastructure for the Global South. We plan to capture 80% of the Indian market before aggressively expanding into the Middle East and Southeast Asia. Current global tools are heavily biased toward US and European data, leaving Asian markets drastically underserved despite a massive, imminent regulatory push. ClimoRisk is building the localized, high-resolution platform these markets actually need. |
We plan to register Pvt Ltd in May |
50,000 grant from IIT Madras |
Yes |
ClimoRisk sits directly at the intersection of Avinya’s Climate Tech and FinTech tracks. We are specifically applying to leverage Avinya’s pilot testing opportunities and institutional credibility, which will be crucial in helping us bypass the notoriously difficult enterprise sales cycles and secure our first banking clients. |
Our primary objective is to transition from product-ready to enterprise-validated. |
Comfortable with any option |
No |
Not applicable |
|
Our proprietary advantage is our hyper-granular, 30-meter resolution dataset. Unlike global incumbents who simply repackage open-source, regional climate data, we use specialized machine learning models to downscale complex climate raster maps from scratch. |
|
For the ML Data Engine (our 30m resolution raster downscaling), we measure statistical accuracy against historical ground-truth data. |
Our current POC is designed strictly for sandbox validation using synthetic or non-sensitive data to test our financial translation engine. |
We don't face regulatory risk; we are actually riding a massive regulatory tailwind. Governments and central banks are increasingly mandating climate stress tests and carbon disclosures for financial institutions. |
|
Yes, Aryan (CTO) currently handles the ML downscaling. He is a AI & Data Science engineer from IIT Madras. |
|
|
India first, then expand to APAC market |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/znIbzIl7NZdq-Climorisk-Pitch-deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/znIbzIl7NZdq-Climorisk-Pitch-deck.pdf |
https://youtu.be/NZaxg5lqDHg |
We believe that achieving a net-zero future requires fundamentally changing how global capital is allocated, and financial institutions are the ultimate catalysts for this. Currently, banks underestimate the exposure of the 'brown economy' (like oil and gas). By giving them the tools to accurately price the financial losses associated with these high-emission assets, we make the hidden costs undeniable. When banks see the true financial risk, market forces naturally divert capital away from fossil fuels and toward a resilient, green economy. |
We come from Tier 3 city |
NA |
checked |
| Apr 10, 2026 @ 7:30 PM |
Ajay S Kabadi, Vinayak Kabadi |
ajay@docketrun.com |
https://www.linkedin.com/in/ajay-kabadi/ |
http://NA |
9449034387 |
Ajay S. Kabadi – Founder & CEO AI & Computer Vision specialist with 7+ years of experience in building and deploying industrial AI solutions. Former Tech Lead at Deshpande Startups, mentoring startups from PoC to production. At DocketRun, he leads client engagement, partnerships, and drives adoption of AI solutions across industrial enterprises. Vinayak Kabadi, Strategy & Operations. Handles strategy, operations, and enterprise project execution. Experienced in proposal design and go-to-market strategy, working closely with customers to convert industrial challenges into deployable AI solutions. |
We have known each other through family and have been working together closely for over a year, collaborating on strategy, operations, and execution to scale DocketRun’s industrial AI deployments. |
2 |
While one of us focuses on building and scaling AI solutions, the other drives strategy, customer engagement, and deployment. This balance allows us to move quickly from problem identification to real-world implementation in industrial environments. |
Docketrun Tech Private Limited |
https://docketrun.com/ |
Hubballi, Karnataka |
DocketRun is an industrial AI platform enabling real-time safety, process, and operational intelligence through video analytics and automation. |
Industrial operations rely heavily on manual monitoring and reactive systems, leading to safety risks, process inefficiencies, downtime, and lack of real-time visibility. |
We provide an AI-powered video analytics platform that integrates with existing cameras and industrial systems to detect, analyze, and trigger real-time actions (alerts, hooters, PLC signals), enabling proactive safety and process control. |
Our platform goes beyond detection to enable real-time intervention through integration with industrial systems (PLC, IoT), combined with context-aware AI models tailored for complex industrial environments. |
Revenue |
Revenue |
Large and mid-sized industrial enterprises such as steel, manufacturing, logistics, and refineries. Companies looking to improve safety, compliance, and operational efficiency.. |
$277.8M (2024) → $887.9M (2033) CAGR: ~13.78% (IMARC) |
Industrial Video Analytics (~30% of TAM): $85M (2024) → $260M (2033) |
Penetration target: 3–5% of SAM $3M – $12M achievable in 3–5 years |
Enterprise B2B model with one-time deployment and integration fees, along with annual AMC and software licensing for analytics and platform usage. |
Ripik AI, Detect Technologies, Assert AI |
Through pilot deployments (PoCs), direct enterprise sales, industry networks, and partnerships with system integrators and industrial solution providers. |
Start with pilot deployments at plant level, demonstrate measurable impact (safety, efficiency, downtime reduction), and scale across multiple plants and use cases within the same enterprise. |
To become the intelligence layer powering industrial facilities, where AI continuously monitors, understands, and autonomously drives safety, process, and operational decisions in real time. |
Private Limited company |
Bootstrapped |
Yes |
We are looking to accelerate our growth through structured mentorship, industry connections, and access to pilot opportunities. Avinya’s ecosystem aligns well with our stage of scaling industrial AI deployments. |
a) Secure paid co-innovation pilots with industrial enterprises to validate and scale deployments b) Expand into logistics, refineries, and heavy industries with real-world use cases c) Build strategic partnerships for multi-plant rollouts d) Accelerate growth through structured GTM and scalable deployment models |
Yes, we are open to hybrid and virtual participation. |
Yes |
Core engine |
Our platform uses deep learning-based computer vision models (YOLO for detection, ResNet for feature extraction) with multi-object tracking and OCR pipelines. It runs on on-prem GPU infrastructure (NVIDIA CUDA) for real-time inference across multiple video streams. The system integrates with PLCs, relays, and IoT layers to trigger automated actions. Built using OpenCV, CUDA, and optimized C/C++ inference engines with a modular API-driven architecture. |
We leverage industrial datasets collected from real deployments across multiple plants, including annotated safety, process, and operational scenarios, continuously improving model accuracy and adaptability. |
industrial datasets and deployment learnings
Context-aware models tailored to real industrial workflows
Deep integration with PLC/SCADA systems enabling real-time intervention (not just detection) |
We measure precision, recall, and false alarm rates, along with real-world deployment accuracy (90%+ detection performance). Continuous monitoring and retraining pipelines ensure reliability across diverse industrial environments. |
Our system is deployed on-premise, ensuring that all video and operational data remain within the customer’s infrastructure. We follow client-specific data access controls, secure storage practices, and comply with enterprise security requirements. |
Minimal regulatory risk, as the system operates within client premises. However, we ensure compliance with data privacy norms and industry-specific safety and surveillance regulations. |
Primary challenges would be compute scalability and deployment standardization. We are addressing this through modular architecture, GPU scaling, and optimized inference pipelines for high-volume deployments. |
Yes, we have an in-house team specializing in computer vision, deep learning, and full-stack deployment, with experience in building and deploying real-time industrial AI systems across multiple environments. |
We use a combination of proprietary models, project-specific annotated datasets, and selected open datasets for base model training. |
Through continuous data collection from deployments, active retraining pipelines, feedback loops from real-world usage, and expansion of use-case-specific datasets. |
We are building for both India and global markets, with solutions designed to be adaptable across industrial environments worldwide. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/6rMxH4F30wRE-DocketRun-Tech-Private-Limited_deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/6rMxH4F30wRE-DocketRun-Tech-Private-Limited_deck.pdf |
https://workdrive.zoho.in/folder/hzqanc2337a398f8343dba16046edee90f4b2 |
Yes, we are a mission-driven startup focused on improving industrial safety, efficiency, and compliance. Our solutions aim to prevent accidents, reduce downtime, and create safer working environments by enabling real-time monitoring and intervention. We are working towards a future of zero incidents and zero downtime in industrial operations. |
NA |
Deshpande Startups |
checked |
| Apr 10, 2026 @ 5:21 PM |
Yeshwanth SH, Sunder G, Ameen Ahmed |
yeshwanthsh128@gmail.com |
https://www.linkedin.com/in/yeshwanth-sh-42b3532b6/ |
https://exora.solutions/ |
8095404788 |
Founder, Information science engineering background. Co-Founder - Electrical Science Engineering Co Founder - Information Science Engineering |
We met in college and we have been working together from the past 4 months |
All |
My biggest strength is my ability to stay consistent under uncertainty. Building a company over the past 7 months forced me to handle ambiguity, setbacks, and pressure, and still keep executing. That resilience is what helped me take it from an idea to actually hiring people. |
Exora Solutions Pvt Ltd |
https://exora.solutions |
bangalore |
Your Life, on autopilot |
Accessing and executing complex workflows across tools is still manual, fragmented, and inefficient. Users have intent, but there’s no system that can seamlessly understand it and take action across environments. |
IRA is a desktop application, that converts user intent into fully executed actions across software and workflows and apps / browsers. |
IRA is built to both decide and execute, closing the loop that most AI tools leave open. As it runs real workflows, it compounds in capability, creating a strong long-term moat. |
MVP |
Signups |
Students, msme's, enterprises and developers |
~$150B+ |
~$25B–$40B |
~$25B–$40B |
freemium |
NA |
NA |
NA |
To build a personal AI operating system that replaces traditional apps by turning user intent directly into execution. |
Completed |
NA |
Yes |
I’m building IRA as a long-term system-level product, not just an application, and I’m looking for an environment that supports deep thinking, rapid iteration, and meaningful feedback. Avinya stands out because it seems to prioritize builders who are focused on solving real problems with long-term impact. I want to be surrounded by people who are equally driven, technically strong, and thinking at the level of systems and platforms. The structure, mentorship, and peer group would help me refine both the product and my approach to building |
During the programme, my primary goal is to evolve IRA from an early-stage prototype into a solid, usable product with clear user value. This includes refining the core user experience (especially the desktop interaction model), validating key use cases, and building a stable architecture for real-time interaction and execution. I also want to gain clarity on product positioning, distribution, and how to prioritize features that actually drive adoption. Beyond the product itself, I want to improve as a builder |
Yes |
Yes |
Automation layer |
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Nzxr7G98D4hB-Exora_Pitch_Deck.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Nzxr7G98D4hB-Exora_Pitch_Deck.pptx |
http://- |
IRA is not just a productivity tool—it is an attempt to fundamentally change how people interact with computers. Today, software is fragmented, complex, and requires users to constantly adapt to tools. Our mission is to shift this paradigm by making technology adapt to human intent instead.
With IRA, we are building a desktop-native AI layer that allows users to simply express what they want, and have the system figure out how to execute it. This has the potential to significantly reduce cognitive load, save time, and make advanced capabilities accessible to a much broader range of users—not just technical experts. Over time, this can democratize automation, enabling individuals to achieve more without needing deep technical knowledge or navigating multiple tools.
From an impact perspective, IRA aims to improve how knowledge workers, students, and creators operate on a daily basis by removing friction between thinking and doing. In the long term, we see this evolving into a personal AI operating system that becomes the primary interface between humans and digital systems, making computing more intuitive, efficient, and accessible. |
No |
- |
checked |
| Apr 10, 2026 @ 4:13 PM |
Vaishnavi Ramesh Babaleshwar, S Sayeed Amaan |
vaishnavirb03@gmail.com |
https://www.linkedin.com/in/vaishnavi-r-babaleshwar-2660432a9/ |
https://dhanvam.com/ |
91087 06881 |
Vaishnavi R. Babaleshwar | Co-Founder & COO, S Sayeed Amaan | Co-Founder & CEO both B.E Graduates |
My Co-Founder and I met during our undergraduate studies in Artificial Intelligence and Data Science and have been working together for the past 1.5 years. This shared history means we already have a strong baseline of trust and a proven working dynamic. |
2 |
Our biggest strength is our execution balance. Because we both have AI degrees, we completely understand the product, but we divide the work strictly: Sayeed builds the tech, and I run the business. |
Sophyxis Technologies Private Limited |
https://sophyxis.com/ |
Belgaum |
Backed by Sophyxis Technologies, Dhanvam is an AI-powered financial co-pilot democratizing elite wealth management for the Indian market. |
The financial wealth management industry currently faces a dual crisis: 1. The "Advisor Bottleneck" (B2B): Financial Advisors spend nearly 70% of their time on manual data aggregation and compliance reporting, leaving little time for client strategy. This limits the number of clients they can serve and forces them to ignore the mass market. 2. The "Advice Gap" (B2C): Retail investors lack access to personalized, data-driven financial advice. Existing apps provide raw data but lack actionable insights, leading to poor financial health and investment decisions for the average user |
Sophyxis Technologies solves these challenges through DHANVAM, a unified AI-driven
ecosystem:
1.For Advisors: We provide an "AI Co-pilot" powered by our specialized FinLLM. This tool
automates research, risk profiling, and report generation, allowing advisors to serve 10x
more clients efficiently.
2.For Retail Users: We offer a Personal Finance Management (PFM) mobile application
that aggregates financial data and provides hyper-personalized, AI-driven advice to help
users manage expenses and investments effectively |
Our uniqueness lies in our "Deep Tech Moat": Proprietary FinLLM Architecture: Instead of relying on generic, off-the-shelf GPT wrappers, our goal is to build and train a bespoke Finance Large Language Model (FinLLM). We will train it specifically on localized financial data and regulatory datasets to ensure high accuracy and strict compliance in wealth auditing and advising. Bridged B2B and B2C Ecosystem: We are designing a dual-sided ecosystem. The professional workflows processed through our future B2B advisor platform (Dhanvam Assist) will be structured to continuously refine the hyper-personalized advice delivered on our B2C retail app, creating a powerful, defensible data feedback loop. |
MVP |
Pilots |
Dhanvam is designed to serve two distinct segments, reflecting our dual-sided approach: B2B (Financial Professionals & Institutions):Financial Advisors & CAs: Cloud-native advisory firms seeking AI automation. This addresses the massive gap where 1.5 Lakh advisors are needed in 5 years. Enterprise Institutions: Entities that will eventually require API licensing or custom, secure AI deployments. B2C (Retail Consumers & Investors): Everyday Consumers: The 76% of individuals who lack financial literacy and the 17.7% facing debt challenges. They need our app to automate budgeting.Tech-Ready Retail Investors: Users seeking hyper-personalized strategies that adapt autonomously to their spending habits. |
$21.9B+ |
$6.25B |
$74.4M |
Dhanvam operates on a scalable, high-margin model targeting both enterprise and consumer markets: Engine A: B2B Enterprise (Dhanvam Assist) SaaS Subscriptions: Recurring fees from financial advisory firms utilizing our AI Co-Pilot. Enterprise APIs: High-ticket licensing of our proprietary FinLLM to large financial institutions. Engine B: B2C Consumer (PFM App) Freemium Tier: Free basic tracking for rapid user acquisition; ₹199/month premium for autonomous AI financial coaching. Affiliate Commissions: Lead-generation revenue from recommending highly targeted financial products based on AI-generated user risk profiles. |
Wealthy.in and AssetPlus fit best here. They are heavily funded wealth-tech startups/scale-ups operating in your B2B advisory space. YNAB also functions like a highly scaled global tech startup. MobiKwik and Axio fall into this category. They are established, large-scale fintech companies with massive user bases that you are looking to disrupt. |
We acquire early users through our VTU VRIF incubation network for near-zero CAC, scale by leveraging our RBI Innovation Hub backing for institutional trust, and sustain long-term growth through an AI-driven content flywheel. |
Trust and customer acquisition cost (CAC) are our main hurdles. We leverage our existing backing to solve both: Phase 1: Early Adopters: We will launch our MVP directly to young professionals and students through our VTU VRIF co-incubation network, ensuring a very low initial CAC. Phase 2: Credibility-Led Growth: We will heavily market our backing by the RBI Innovation Hub (RBIH) and our Finnovate Top 10 status. This institutional trust de-risks the platform for retail users and helps secure B2B partnerships. Phase 3: AI-Driven Flywheel: As the AI gets smarter, we will generate anonymized financial insights (e.g., "How people in your age group save") to drive organic content marketing and lower our long-term acquisition costs. |
Our long-term vision is to democratize wealth creation in India. We aim to establish Dhanvam as the country's most trusted, sovereign AI financial advisor, transitioning millions of young professionals from basic savers into strategic, data-driven investors |
Active |
NA |
No |
We are applying to Avinya because our mission at Dhanvam—building a sovereign, localized FinLLM for India—perfectly aligns with ACSEL’s core focus on responsible, impactful AI. While our current co-incubation with the RBI Innovation Hub and VTU VRIF gives us strong regulatory grounding, we are looking for deep technical and strategic scaling support. Avinya’s dedicated FinTech track, combined with the unparalleled mentorship and market access provided by the IIT Alumni Centre Bengaluru (IITACB) network, makes this the exact ecosystem we need. We want to leverage this network to help us navigate the complex technical challenges of scaling deep-tech and transition our platform from early validation to mass-market deployment. |
Over the course of the 16-week programme, our primary goal is to elevate our platform from a validated MVP to an enterprise-grade, fully compliant solution. Specifically, we aim to achieve three things: Technical Stress-Testing: We want to leverage Avinya's IIT alumni mentors to stress-test our AI architecture, optimizing our FinLLM for large-scale data handling and strict domestic compliance. Pilot Expansion: We aim to utilize the program's industry network to secure B2B pilot opportunities for our 'Dhanvam Assist' platform with larger financial institutions. Fundraising Readiness: We plan to use the structured strategy sessions and the initial seed funding to aggressively refine our go-to-market execution and prepare our unit economics for a larger institutional funding round. |
Yes |
Yes |
Core engine |
Dhanvam employs a sophisticated multi-agent "Swarm" architecture orchestrated via LangGraph, allowing distinct agents (Planner, Quant, Trader, Guardian) to handle specific domains dynamically.
Core Model: The cognitive engine utilizes an open-source foundation model (Llama-3.1 8B) fine-tuned using the FinGPT framework to create our specialized FinLLM.
RAG Pipeline: We use a hybrid search architecture with layout-aware parsers to extract structured tables from financial documents, embedded into a Pinecone vector database.
Infrastructure & APIs: The backend logic and LLM inference run on Google Cloud instances. We integrate Setu for real-time cash flow aggregation and SnapTrade for broker-agnostic trade execution.
Frontend: The client applications are built using Flutter and Firebase for cross-platform delivery and real-time syncing. |
Our core advantage is the closed-loop data pipeline generated by our dual-product strategy (the B2C app and the B2B Dhanvam Assist). Furthermore, our system is designed to be highly "Stateful". By monitoring user behavior—such as interaction tone and login frequency during market downturns—we build a dynamic "Psychometric Profile". This continuous behavioral data allows our AI to deliver personalized "Advisor Gamma" by proactively initiating coaching dialogues to de-escalate panic, an advantage static robo-advisors lack. |
We avoid generic API wrappers. By utilizing the lightweight FinGPT framework, we can execute Low-Rank Adaptation (LoRA) fine-tuning daily or weekly, ensuring our model adapts to fresh market news far faster than proprietary monolithic models. Our defensibility is cemented by a hybrid architecture that fuses semantic LLM reasoning with deterministic mathematics; while the LLM handles strategy, a specialized "Quant Agent" executes portfolio optimization using robust frameworks like the Black-Litterman model to incorporate market views mathematically. |
Standard NLP metrics are insufficient for fiduciary-grade advice. We track domain-specific metrics like Intent Classification Accuracy and the recall rates of our cross-encoder RAG reranking. Crucially, we mandate Financial Chain-of-Thought (CoT) prompting within our agent framework, allowing us to explicitly audit the AI's step-by-step reasoning against regulatory standards. Reliability is ultimately measured by our Human-In-The-Loop system, tracking the acceptance and modification rates of AI suggestions by the professional advisors utilizing Dhanvam Assist. |
Security and compliance are non-negotiable, a standard validated by our incubation in the Reserve Bank Innovation Hub (RBIH). Compliance-as-Code: We utilize NVIDIA NeMo Guardrails to enforce deterministic input/output rails, blocking malicious prompts and preventing the AI from hallucinating guaranteed returns. Data Governance: We deploy PII Redaction Middleware, ensuring the LLM inference engine never processes raw, sensitive identifiers like Social Security or Aadhaar numbers. Access Control: We enforce the principle of least privilege, separating read-only research agents from the "Trader Agent," which requires explicit confirmation steps before accessing execution APIs. |
The primary risk involves navigating SEBI and RBI guidelines regarding automated advisory and the Unauthorized Practice of Law (UPL) when generating estate or tax strategies. We mitigate this by positioning Dhanvam Assist as an augmenting co-pilot for licensed professionals. For the autonomous B2C features, we strictly maintain an Audit Trail that logs every piece of generated advice with its corresponding "Chain of Thought" to ensure verifiable compliance with "Best Interest" regulations. |
At a 10x scale, our primary bottleneck will be the compute cost and inference latency of hosting a 8B parameter FinLLM on Google Cloud. Additionally, processing a massive surge of concurrent multi-agent workflows—especially the "Trader Agent" reacting to real-time feeds at ~20ms latency—will overwhelm our current Firebase MVP infrastructure. We will need to migrate our core transaction ledgers and state management to a highly optimized PostgreSQL database to handle the high-frequency read/writes |
Yes, product development is driven in-house by our founding team, Sayeed Amaan and Vaishnavi R B. Our technical leadership is rooted in a B.E. in Artificial Intelligence and Data Science from VTU, with extensive full-stack expertise spanning Python, NodeJS, React Native, TensorFlow, and Flutter. Having personally trained over 1000+ students in AI/ML and guided 46 internship projects, we possess deep internal R&D capabilities and direct access to a robust talent pipeline for scaling our engineering team. |
We utilize a blend of specialized open-source and proprietary datasets. For instruction tuning, we leverage datasets like Finance-Instruct-500k to teach the model multi-turn conversational advisory behaviors, and the Phinance Dataset to train the "Planner Agent" on complex financial trade-offs. For our RAG knowledge base, we actively ingest unstructured public data, including SEC 10-K Filings and earnings call transcripts, vectorized via specialized parsers. |
We implement Reinforcement Learning from Human Feedback (RLHF) to align the model with fiduciary standards. By capturing the daily interactions—the approvals, rejections, and nuanced edits—made by financial professionals using Dhanvam Assist, we generate high-fidelity, proprietary training loops. This ensures the underlying FinLLM continuously refines its tone, accuracy, and strategic logic based on expert validation. |
We are building for India first, targeting the massive need for accessible financial planning. The Flutter frontend is cross-platform, and the multi-agent framework orchestrating the LLM can easily be redirected to navigate global tax codes or integrate with international aggregators, positioning us for seamless global scaling once our Indian operations are solidified. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/EeAyQ7WNNPkU-DHANVAM-Pitch-deck-BIuild-with-AI.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/EeAyQ7WNNPkU-DHANVAM-Pitch-deck-BIuild-with-AI.pdf |
https://drive.google.com/file/d/16Sbf0kzRd_I0yFfsHLs-lQzQB8GVeH0u/view?usp=drivesdk |
Yes, Dhanvam is fundamentally impact-focused, driven by the mission to democratize high-quality, fiduciary financial guidance for the billions of people currently priced out of the market. The traditional wealth management industry relies on a high-touch, human-centric model that inherently limits access. By building an Autonomous Fiduciary Agent, our goal is to solve the "Gamma" problem—the quantification and delivery of value through financial planning, behavioral coaching, tax optimization, and estate management. Aligned with the vision of our early backers at the Reserve Bank Innovation Hub (RBIH), our overarching objective is to tackle localized financial inclusion challenges in India, ensuring that comprehensive, personalized wealth strategy is no longer a luxury reserved for the wealthy, but an automated, accessible utility for everyone. |
Yes. As a woman from a rural background and a first-generation entrepreneur, I come from an underrepresented segment in the tech ecosystem. These experiences shape my mission to create inclusive, accessible solutions for individuals who are often overlooked by traditional systems. |
NA |
checked |
| Apr 9, 2026 @ 10:10 PM |
Daniel D'Souza and Hassan Sesay |
danielmarcdsouza@gmail.com |
https://www.linkedin.com/in/daniel-marc-d-souza?utm_source=share_via&utm_content=profile&utm_medium=member_ios |
http://Na |
9108390479 |
Daniel Marc Dsouza : cofounder, full stack developer, Bachelor of Science in Computer Science, Pennsylvania State University. Hassan Sesay :cofounder, full stack developer, Bachelor of Science in Computer Science, Pennsylvania State University |
Met during class as we attended the same college, Pennsylvania State University, and were majoring in the same subject ( computer science). Known each other for 3 years now and have worked on multiple side projects during that span. Worked on our biggest project last year which won us our University department final project competition along with a startup competition we took part in. |
1 |
Though we both work as full stack developers we have different experiences and perspectives along with creative inputs that facilitates each other while making up for knowledge we would be individually weak in. |
Seva |
NA |
Mangalore |
Seva — the neighbourhood, digitised. |
India has 15 million small shops — salons, tailors, repair shops, kiranas — generating over ₹67 lakh crore in annual revenue, yet fewer than 15,000 have any digital presence. These businesses have no way to be discovered online, no system to manage bookings or workers, and no digital record of their own revenue. Meanwhile, customers have no reliable way to find, vet, or book skilled local workers beyond word-of-mouth. The gap is compounded by India's 22+ languages, which make communication between customers and shop owners across regions nearly impossible on existing platforms. |
Seva is a social media-style mobile app where every card is a small business — not an individual. Shop owners create a digital storefront in minutes, add their workers with skill profiles and track records, and start accepting bookings and UPI payments through the app. Customers browse a personalised, location-first feed of nearby shops, view staff profiles and ratings, and book directly. AI-powered live translation via India's Bhashini API bridges language barriers in real time, so a Tamil-speaking customer can seamlessly communicate with a Kannada-speaking barbershop. Shop owners also get a plain-language daily business summary in their own language — many for the first time seeing their own revenue data digitally. |
No competitor combines all four pillars Seva is built on. Urban Company (₹1,260 Cr revenue, listed 2025) targets premium urban professionals — not small shops, and has faced significant worker backlash over exploitative fees. JustDial is a directory with no booking or payments layer. WhatsApp Business enables communication but zero discovery or analytics. Seva's defensibility comes from four compounding moats: first, the shop-first data model creates network effects at the business level — every shop that joins brings its entire worker roster and customer base. Second, vernacular AI translation via government-backed Bhashini infrastructure is a technical capability competitors cannot quickly replicate. Third, Aadhaar-linked KYC verification builds trust that anonymous platforms cannot match. Fourth, by integrating with ONDC — India's government digital commerce network — every Seva shop is automatically discoverable across the national ecosystem, creating distribution no privately-funded competitor can buy. |
MVP |
Testimonials |
Small shop owners (salons, tailors, repair shops) with no digital presence, and everyday customers wanting to discover and book trusted local service workers nearby. |
₹38.4 lakh crore |
₹5.2 lakh crore |
₹180 crore |
Freemium subscriptions (₹0 / ₹499 / ₹999 per month), 2–5% transaction fee per booking, and featured listing fees (₹199–999/week). |
Urban Company, JustDial, OLX, WhatsApp Business |
Field sales in target neighbourhoods, WhatsApp onboarding bot, ₹500 shop referral incentive, ONDC network listing, and hyperlocal community seeding. |
Launch Bengaluru (1,000 shops) → 5 Tier-1 cities → Tier-2/3 India where competitors don't operate. |
To become the digital operating system for India's informal service economy, the layer of infrastructure that sits between every neighborhood shop and every Indian with a smartphone.Every transaction is digital, every worker has a verified track record, and every shop owner, regardless of the language they speak has the same access to customers and business intelligence that was previously available only to large organised players. Beyond India, the model is directly portable to any emerging market with a large informal service economy and linguistic diversity, i.e., Southeast Asia, West Africa, Latin America. |
Currently being worked on |
NA |
Yes |
For the excellent structured mentorship, investor access, and a network that understands the Indian market. Avinya's program is the right environment to pressure-test our go-to-market, refine our pilot strategy, and connect with the right people to launch. |
Solidify the direction to move in and gain clearer understanding to our path ahead. Assistance with initial launching, mentorship and funding. |
Yes |
Yes |
Supporting feature |
Two-stage pipeline. Stage 1: NLTK + regex for fast structured detection of sensitive data (phone numbers, Aadhaar, bank details). Stage 2: Gemma 2.0 9B LLM via FastAPI (Python) for nuanced moderation, i.e., fraud, harassment, fake reviews. NLP pre-processing reduces token input by 80–94% before LLM inference, cutting cost and latency. Separately, Bhashini API handles real-time bidirectional chat translation across 22 Indian languages. Feed ranking runs a weighted relevancy scoring algorithm on the ExpressJS backend using location, category preference, engagement history, and recency signals. |
At launch, none, we will use public models. The advantage builds over time: every booking, review, worker rating, and shop interaction becomes training signal unique to India's informal service economy. No competitor has granular data on small shop worker performance, hyperlocal service demand patterns, or cross-language customer-to-shop communication at this scale. That dataset doesn't exist yet but we build it. |
Three things. First, Bhashini integration, a government-backed API built specifically for Indian languages, that large Western platforms cannot easily replicate culturally or politically. Second, our moderation pipeline is trained on the specific fraud patterns of India's cash-dominant informal economy, not generic Western content. Third, as our proprietary booking and interaction dataset grows, our relevancy and fraud models improve in ways generic models cannot match without that data. |
Content moderation: precision and recall on flagged content, false positive rate
translation:BLEU score benchmarked against Bhashini's published baselines, plus qualitative human review across each supported language
Feed algorithm: click-through rate, booking conversion rate, and session depth per user
latency:translation responses must stay under 2 seconds for messages under 200 characters. |
All data encrypted in transit (TLS 1.3) and at rest (AES-256). Aadhaar verification status stored (raw Aadhaar data never retained). Payment data handled exclusively by Razorpay and Stripe (PCI-DSS compliant). User phone numbers and emails never exposed to other users. Compliant with India's Digital Personal Data Protection Act 2023. Supabase provides row-level security ensuring users can only access their own data. |
Three material risks. First, India's gig worker classification laws, Rajasthan (2023) and Karnataka (2025) bills may require welfare fund contributions per transaction, adding 1–2% cost. Second, DPDP Act 2023 enforcement which states that data localisation requirements could increase infrastructure costs if strict localisation mandates emerge. Third, RBI payment regulations, i.e., any change to UPI fee structures or escrow rules for marketplace payments could affect our transaction model. All three are manageable and affect every competitor equally. |
Gemma 2.0 inference cost and latency, self-hosted LLM inference does not scale linearly and will require GPU cluster investment or migration to a managed inference API. Bhashini API rate limit which we would need to negotiate enterprise tier access before hitting volume ceilings. Feed ranking will need to move from a simple scoring function to a proper ML ranking model as catalogue size grows. |
Not yet but depending on growth and scaling the team would be created to facilitate that. |
Bhashini API : government of India, free public access under the National Language Translation Mission.
NLTK corpora : open source (Apache 2.0).
Gemma 2.0 : Google DeepMind, open weights license for commercial use.
IndicTrans2 : AI4Bharat, MIT license.
No proprietary datasets yet; all current models use publicly licensed resources. |
Short term: human-in-the-loop review of moderation edge cases to build a labelled dataset of India-specific false positives. Medium term: fine-tune Gemma on our accumulated booking, review, and chat data to specialise it for informal service economy content. Long term: replace the static feed scoring function with a learned ranking model trained on booking conversion signals. Translation quality reviewed quarterly by native speakers across each supported language. |
India first, intentionally. The product is built around India-specific infrastructure that cannot be lifted and dropped elsewhere without meaningful rearchitecting. Once the model is proven in India, the same playbook applies directly to any emerging market with a large informal service economy and linguistic diversity: Indonesia, Nigeria, Brazil, Egypt. The IP and architecture travel; the market expansion follows proof of concept in India. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/zd31xA3caxgh-Seva_Pitch_Deck.pptx, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/zYfog8e4eDCC-Seva_Pitch_Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/zd31xA3caxgh-Seva_Pitch_Deck.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/zYfog8e4eDCC-Seva_Pitch_Deck.pdf |
http://NA |
Yes. Seva exists to bring India's invisible workforce online. The barber, the tailor, the electrician, skilled people running real businesses for generations, yet excluded from the digital economy because every platform was built in English, for fast internet, by people who never walked into their shops. Seva changes that: vernacular-first, low-data, cash-friendly, and simple. Every shop that joins gets a digital identity, every worker gets a verified track record, and every digital transaction creates a financial footprint that can unlock credit and welfare benefits for the first time. |
Yes, Hassan is African-american |
Anand Fernandes |
checked |
| Apr 9, 2026 @ 2:04 PM |
MANISH SHETTY |
manish6shetty@gmail.com |
https://www.linkedin.com/in/manish-shetty-519b852b0 |
https://myyuno.com/ |
8792344909 |
Manish Shetty- Co Founder: Technology Lead, Rajesh Sambaragi - Co Founder: Lead Developer, Smaran B M - Marketing & Content Lead |
All of us were connected from a long time from our school and college days, we understood the problems we faced as individuals and sat to a decision to build YUNO. We are currently working full time on our startup. We are working together from August 2025. |
All |
Biggest Strength as a team is that we prioritise the delivery and customer feedback. We know each other's flaws and we use them as triggers to become more productive. Important thing is that we constantly push each other to our limits and build what needs to be built. |
YUNO |
https://myyuno.com/ |
Bengaluru |
We provide digital companionship mainly for people suffering from loneliness. |
According to WHO reports 1 in every 8 individuals suffer from chronic loneliness. The world is more connected than ever but loneliness is at an all time high. People went from "therapy is for crazy people" to "therapy is cool", without really understanding if they need therapy. |
Our solution is to build a device where you can have human-like conversations, a device that is emotionally intelligent:
---> that understands you as an individual,
---> stays with you,
---> remembers every context
---> and grows with you over time
---> keeping all your information private on the device.
In simpler words, its like having a digital friend that lives for you, keeping all your information private, at your convenience.
Think of it like Wall-E & Eve.
**NOTE**- YUNO isn't an assistant, YUNO is a COMPANION. |
Uniqueness about our solution is that we are voice-first interface model that communicates in not just English, but also in Regional Languages as well. And our accuracy is really good compared to existing model. The most unique things are a vision for us at the moment, and with right access to resources, we can definitely make it a reality. |
MVP |
Users |
We are targeting college students and early working professionals in the corporate. The demography lies between 18-40. Psychography - Individuals who use tech as an alternative for emotional connection, who feel lonely in their life, who just want someone to stay with them without judging. |
$56.3B+ |
$2.4B |
$30M |
Hardware Device Sales, Saas Subscription, In App Customizations, B2B Enterprise Solutions. |
Pi AI, NeoSapien, Gemini, Meta, ChatGPT, Replika, Friend AI, TwinMind, Calmi AI, Superhuman. |
Our main acquisition comes from digital presence. Content, Collaboration with On ground events, micro influencer campaigns, and the final resort would be Google& Meta Ads. |
Engage with student lead communities, Creator led storytelling for content campaigns, Mental Health Ecosystem Partnerships, Being present at Corporate Events for potential collaboration/B2B Solution. |
Our vision is to redefine the relationship between humans and technology.
We imagine a world where:
---> AI is emotionally aware
---> Privacy is a standard, not a feature
---> Technology empowers connection, not distraction
---> Every person has a companion they can trust
YUNO is a step toward a future where technology protects our emotional world, not exploits it. |
MSME |
NA |
Yes |
4 Month Founders Program. Infrastructure for DeepTech Startups. Alumni & Post Cohort Support. Funding & Compliance support |
Build a production ready product with the programme's network and market access. Achieve Product Market Fit. Figure out potential market opportunities for product expansion. Potential Incubation/Accelerator Support, Funding & Infrastructure Opportunities, Post Cohort Support. |
Yes, we are interested to be physically present for the programme. |
Yes |
Core engine |
We have built our own proprietary model on top of existing open source models.
We are using all the available open source tools to build our AI backbone, constantly fixing the loopholes to make it as secure as possible. |
We are sure that people are ready to use the product we are proactively building with our community helping us build the product better. Top VC firms have come up with predictions that Personalized AI space will bring disruption in this year of 2026, makes our product valid and launch as it is the right time to do so. We also know that the big 6 and other resource backed companies are building what we're building, makes our product competitively viable. |
Our core defensibility comes from long-term user memory and personalization.
YUNO is not a stateless assistant, it continuously learns from user interactions, building a deeply personalized layer that compounds over time.
As more users interact with YUNO, the quality of personalization and memory graphs improves, creating a strong data moat.
Additionally, our agent-based architecture + real-world action layer makes it harder to replicate compared to simple chatbot interfaces.
Over time, the combination of user-specific data + behavior learning + integrations becomes our biggest defensible advantage. |
Task success rate (did YUNO complete the intended action correctly)
Response quality & relevance (context understanding, personalization accuracy)
Latency & reliability (response time, failure rate)
We also track user-level metrics like retention, daily interactions, and correction rate (how often users rephrase or fix YUNO), which directly reflect real-world performance |
We design YUNO with a privacy-first architecture from the ground up.
User data is securely stored and encrypted
We follow strict access control and authentication systems
Sensitive actions always require explicit user permission
Since YUNO relies heavily on personal context, we focus on transparency and user control, users can view, manage, or delete their data at any time.
Currently, we use secure cloud infrastructure to store and process user data. However, our long-term vision is to move towards a fully in-house architectural layer, where data is stored, processed, and handled locally without relying on cloud systems.
The goal is to ensure that we don’t have access to user data at all, making privacy not just a feature, but a core technology layer of YUNO.
As we scale, we also plan to align with global compliance standards (like GDPR-level practices) |
Yes, as a system handling user data and AI-driven actions, we are mindful of data privacy and AI regulation risks.
However, since YUNO is designed as a user-controlled assistant (not autonomous decision-maker), the risk is relatively manageable.
We are proactively building with compliance, consent, and transparency in mind to stay aligned with evolving regulations. |
At 10x scale, the biggest challenges would be:
Infrastructure load (handling millions of real-time interactions)
Latency in voice and agent execution
Memory retrieval efficiency at scale
These are expected scaling challenges, not fundamental limitations.
We are designing the system with modular architecture and scalable infra so we can optimize and distribute load as we grow. |
We are currently a small but highly focused technical team, with the core technology of YUNO being built directly by the two co-founders.
Our expertise primarily lies in:
Architectural layer design and system design
Agent-based system development
LLM integrations and orchestration
Building real-world action layers (calls, messaging, scheduling)
Moderate AI/ML operations and applied AI systems
We are very hands-on with the entire stack, from system architecture to product experience which allows us to move fast and build tightly integrated systems. A system that can continuously evolve the technology as YUNO grows. |
Currently, we primarily rely on pre-trained foundation models via APIs, so we are not training base models from scratch.
On top of that, the core dataset we build is user interaction data (with consent) conversations, preferences, and behavioral patterns that help YUNO personalize over time. This data is owned and structured by us at the application layer, not at the model training level.
We are very conscious about privacy, so this data is used strictly to improve personalization and system performance, not for external sharing or resale |
Our approach is very product-driven.
We improve performance through:
User interaction feedback loops (how users respond, correct, or engage)
Task success tracking (whether YUNO actually completes actions correctly)
Memory and personalization tuning (improving how context is stored and retrieved)
Instead of only focusing on model-level improvements, we optimize at the system level prompts, memory layers, and agent workflows.
Over time, as we gather more real-world usage data, YUNO is becoming more accurate, more personalized, and more reliable |
We are building YUNO for a global market from day one.
The problem we’re solving; digital overload and lack of personalized AI, is universal.
At the same time, starting from India gives us a strong testing ground because of the diversity in users, languages, and use cases.
Our goal is to build a system that is globally scalable, culturally adaptable, and deeply personal, regardless of geography |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/ciyCyNwM8law-ACSEL-PITCH-2026_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/ciyCyNwM8law-ACSEL-PITCH-2026_compressed.pdf |
https://drive.google.com/file/d/1Hk5kY4bOp5D3BCekPWrwsx7UiquMdqAB/view?usp=drivesdk |
YUNO is a mission-driven startup because its primary goal is to build a new category of technology centered around emotionally intelligent AI companionship that helps people feel understood. While the product naturally creates positive outcomes like reducing loneliness and improving emotional awareness, these are results of the mission, not the core positioning. The focus is on building a scalable, category-defining product, where meaningful impact emerges as a byproduct of delivering a deeply human-centered technological experience. |
NA |
KDEM, Venkatesh Deshpande sir |
checked |
| Apr 9, 2026 @ 1:50 AM |
Gaurav Dixit |
gdixit@skillops.ai |
https://www.linkedin.com/in/gauravdixit2/ |
https://skillops.ai |
+91-7676719650 |
Gaurav Dixit - CEO (15 yrs, Adobe/Yahoo/Capgemini, IIMB/BITS); Surya Tripathi - CTO (15 yrs, enterprise product architecture); Arun Sharma - CBO (15 yrs, pre-sales & enterprise sales); Sapna Raikar - CPO (19 yrs, B2B/B2C product at HP/Yahoo/Adobe) |
The founding team connected through professional networks and shared enterprise experience. Gaurav (CEO) and Surya (CTO) collaborated on enterprise delivery challenges at large organizations. All four founders have 15–19 years of experience in adjacent domains—product, tech, sales, and HR—within global companies like Adobe, Yahoo, HP, and Capgemini. They came together with a shared conviction: enterprises are flying blind on skill visibility, and AI can fix that. |
2 |
We combine deep domain expertise with real-world enterprise experience. Our team has lived the problem firsthand—working inside organizations where skill visibility was missing. Gaurav brings operational and product leadership (IIMB/BITS), Surya brings technical depth in enterprise architecture, Sapna brings 19 years of B2B/B2C product strategy, and Arun brings 15 years of enterprise sales and pre-sales. We are a full-stack founding team—no critical gap in product, tech, or go-to-market. |
SkillOps.ai |
https://skillops.ai |
Gwalior, Madhya Pradesh |
AI-powered Skill Intelligence Platform for optimizing workforce and work planning in global enterprises |
Enterprises lack clear insight into employee skills and skill gaps, leading to poor resource allocation, bloated hiring budgets, underutilised talent, slower GTM, and increased employee and customer dissatisfaction. HR, L&D, and delivery teams operate in silos with skills data scattered across spreadsheets and LMS systems—making it impossible for leadership to plan intelligently. |
SkillOps.ai is a Generative AI-powered, Skill-based SaaS platform that unifies skill intelligence, workforce planning, and work execution. It enables: (1) AI-Powered Skill Intelligence—real-time skill tracking beyond job titles; (2) Workforce & Resource Planning—combining skill data with workload forecasts; (3) Work & Task Intelligence with Internal Mobility—aligning right skills to right work; (4) Closing Skill Gaps with ROI—connecting L&D to measurable business outcomes; (5) AI Agents that break silos between HR, L&D, and delivery. |
First-of-its-kind SkillOps platform: Skill-First by Design (skills as primary unit of workforce intelligence), End-to-End (learning + scheduling + headcount in one AI layer), Outcome-Driven (measures cost saved, projects delivered, gaps closed—not just training hours), and AI Agents for enterprise cross-functional workflows. Built from Gwalior—proving world-class enterprise AI does not need a metro address. |
MVP |
Pilots, Signups, Testimonials |
Global mid-to-large enterprises (500+ employees) in IT Services, Financial Services, Manufacturing, and GCC (Global Capability Centers). Decision-makers: CHROs, L&D Heads, Delivery Directors, COOs, and CXOs. |
$15 Billion (skill-based & AI workforce planning — 30% of overall HR tech market) |
$9.4 Billion — 63% of organizations are planning to move to skill-based approaches |
$5.6 Billion — mid to large enterprises with 500+ employees accounting for 60% of market demand |
SaaS Subscription (Tiered): Basic Plan (skills inventory, basic planning), Pro Plan (AI-driven capacity planning, project forecasting, talent management), Enterprise Plan (custom pricing, full features, integrations, advanced analytics). Additional: Premium Integrations (SAP/Salesforce), Training & Consultancy, AI-based Data Insights as premium add-on. |
Workday, SAP SuccessFactors (legacy HR tech treating skills as metadata), Gloat and Fuel50 (internal mobility point solutions), Cornerstone OnDemand (LMS). Unlike all of these, SkillOps unifies skill intelligence + workforce planning + work execution in one AI layer. |
Direct B2B enterprise sales via LinkedIn outreach, custom demos & POCs; inbound via content marketing, SEO, webinars; partnerships with HRIS/ATS providers and HR consulting firms; community-led growth through SkillOps HR leader network and referral programs. |
Four-pillar GTM: (A) Direct B2B enterprise sales — LinkedIn outreach, custom demos, consultative selling; (B) Inbound marketing — content, SEO, webinars, LinkedIn thought leadership; (C) Partnerships — HRIS/ATS/HR consulting alliances; (D) Community-led growth — exclusive SkillOps community, referral programs, customer success stories. |
In 5 years, SkillOps.ai will be the centralized enterprise skill intelligence layer — the single source of truth for all workforce operations globally. Every enterprise HR, L&D, delivery, and leadership decision will be powered by SkillOps skill data. We will become the backbone of the skill-first economy for Indian enterprises and export that capability to global markets (Middle East, USA, UK), ultimately reshaping how 50M+ enterprise employees are managed, deployed, and grown. |
Private Limited Company — registered in India |
Bootstrapped — no external funding raised yet. Seeking to raise $300K (₹3 Crore) at $3M valuation with 10% equity dilution. |
Yes |
Avinya's structured mentorship, investor connections, and ecosystem access are critical for SkillOps.ai at this stage. As a deep-tech enterprise AI startup from Gwalior (Tier-2 city), we need the right strategic guidance to validate our enterprise GTM, accelerate sales cycles, and connect with investors who understand B2B SaaS. ACSEL's focus on AI startups and Karnataka's enterprise ecosystem is a perfect fit for our expansion plans. |
Raise our $300K seed round, close our first 3 paying enterprise customers, file key AI/planning patents, refine product-market fit for IT services and GCC verticals, and build a replicable sales playbook for enterprise B2B. We also want to leverage the ACSEL alumni network for introductions to relevant enterprise buyers and institutional investors. |
Yes — fully open to hybrid and virtual participation. Our team is distributed across Gwalior and can travel to Bengaluru for in-person sessions, workshops, Demo Day, and key investor meetings. |
Yes |
Core engine |
SkillOps.ai is built on a multi-agent Generative AI architecture: (1) Skill Intelligence Engine — uses NLP and LLMs to extract, classify, and map skills from resumes, profiles, project data, and HCM systems; (2) Workforce Planning Engine — ML-based capacity forecasting combining skill data with project pipelines and market benchmarks; (3) Work-Skill Matching Engine — AI-driven task-to-person matching using skill vectors and availability graphs; (4) Gap Analysis & L&D Recommendation Engine — connects skill gap identification to curated learning pathways; (5) Agentic AI Layer — purpose-built AI agents for HR, L&D, and delivery teams enabling cross-functional visibility. |
Enterprise skill graphs built from internal HR, project, and learning data — highly proprietary once ingested. Benchmarking data from 500+ enterprise skill taxonomies across IT, BFSI, manufacturing, and GCC sectors. Real-time skill demand signals from job market and project pipeline data. This multi-source skill ontology becomes a competitive moat that deepens with each enterprise customer onboarded. |
Defensible through: (1) Proprietary enterprise skill graphs that deepen with each customer; (2) Network effects — the more enterprises use SkillOps, the richer the benchmark dataset; (3) Deep enterprise integrations with HRIS, ATS, LMS, and project tools creating high switching costs; (4) First-mover advantage in the SkillOps category — no single competitor combines skill intelligence + workforce planning + work execution in one AI layer; (5) 60+ years of domain expertise in the founding team building barriers to replication. |
Model performance metrics: Skill extraction accuracy (precision/recall on skill tagging), Workforce planning forecast accuracy (MAPE on capacity predictions), Skill-to-work matching relevance scores, Skill gap closure rate (pre/post L&D intervention), User engagement and adoption metrics (DAU/MAU), Customer ROI metrics (time-to-staff reduction, bench utilisation improvement, hiring cost savings). Regular A/B testing and human-in-the-loop feedback loops with HR practitioners. |
Data Privacy & Security: Enterprise-grade security with SOC 2 compliance roadmap, data encryption at rest and in transit, role-based access control, tenant isolation in multi-tenant architecture. Compliance with DPDP Act (India), GDPR (for EU customers), and enterprise data residency requirements. All employee skill data is anonymised for benchmarking — no PII shared across tenants. Regular security audits and penetration testing planned. |
Regulatory risks: India's DPDP Act requirements around employee data consent and processing — mitigated by building consent management into the platform. GDPR compliance for European customers — addressed through data residency options. AI fairness and bias risks in skill matching and L&D recommendations — mitigated through bias audits and explainable AI features. No sector-specific regulatory barriers identified for current target markets (IT, BFSI, manufacturing, GCC). |
At 10x scale, the primary bottleneck will be the skill ontology and taxonomy management — keeping skill classifications current across thousands of enterprise contexts. Mitigation: investing in automated skill taxonomy evolution using LLMs and industry data feeds. Secondary risks: LLM inference costs at scale (mitigated via model distillation and caching), and data pipeline latency for real-time skill analytics (mitigated via streaming architecture with Kafka/Spark). |
Yes — in-house AI/ML capability with CTO Surya Tripathi leading technical architecture (15 years enterprise product and solutioning). The team has hands-on experience with LLM fine-tuning, NLP pipelines, ML-based forecasting, and enterprise system integrations. Currently hiring 2 senior ML engineers to expand the AI team. Advisors with deep ML expertise from IIT and IIM backgrounds are being onboarded. |
Current datasets: LinkedIn profile data (via authorized API integrations), internal HR/HCM data from pilot enterprises (licensed via enterprise agreements), publicly available job posting datasets for skill demand signals, O*NET and ESCO skill taxonomy databases (openly licensed). All enterprise customer data is contractually owned by the respective enterprise — SkillOps holds a limited processing license. |
Continuous model improvement via: (1) Human-in-the-loop feedback from HR practitioners validating skill extractions and matching recommendations; (2) A/B testing of skill matching algorithms with real enterprise workflows; (3) Regular retraining on new enterprise data with customer consent; (4) Active learning pipelines flagging low-confidence predictions for expert review; (5) Quarterly model audits for bias, drift, and performance degradation against enterprise benchmark datasets. |
Both — starting with India (GCCs and large IT/BFSI enterprises) as the primary market given the founders' networks and the massive enterprise workforce. Simultaneously pursuing Middle East (high-growth GCC market), followed by USA and UK within 18-24 months. The platform is built global-first: multi-currency, multi-language ready, and compliant with GDPR and DPDP from the ground up. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/H8nSdWlIngVj-SkillopsAI_Pitch_Deck_New_V2.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/H8nSdWlIngVj-SkillopsAI_Pitch_Deck_New_V2.pdf |
https://www.loom.com/share/23464ed1294747778aabcf6ae998a898?sid=94d73e11-a2a1-4930-8982-868f37ded1af |
Yes — SkillOps.ai is mission-driven on two levels. First, we are solving a real economic problem: enterprises globally waste billions annually because of invisible skill gaps and talent misallocation. We are building the infrastructure to fix that — making every person's skills visible, valued, and deployed to their full potential. Second, we are proving that world-class enterprise AI innovation can emerge from a Tier-2 city in India (Gwalior, MP) — challenging the notion that great startups need a metro address, and creating a template for deep-tech entrepreneurship from non-traditional geographies. |
Founders from Tier-2 city (Gwalior, Madhya Pradesh) — one of the few enterprise-grade AI product companies built from outside the traditional Indian startup hubs of Bengaluru, Mumbai, and Delhi. We represent a deliberate choice to build deep-tech from smaller cities. |
Startup Grants India (startupgrantsindia.com) |
checked |
| Apr 8, 2026 @ 11:29 PM |
Niharika Sri Parasa, Venkata Sridhar Parasa |
pnsri@advaita1.com |
https://linkedin.com/in/niharika-sri-parasa |
http://NA |
8247654106 |
Niharika, 8+ years experience in industry and research, Sridhar (Principal, Govt ITI, Bobbili,25+ years experience) |
We’ve known each other lifelong as father and daughter, but our professional collaboration began recently with the shared goal of building an edtech platform. While our backgrounds differ—one rooted in AI research and system design, the other in decades of educational leadership—this combination naturally led us to start working together. |
All |
Our biggest strength is the combination of deep technical capability and real-world educational insight.
We don’t just build AI systems—we build them grounded in how students actually learn. This allows us to move quickly from problem identification to practical, scalable solutions. |
iGuide |
NA |
Bengaluru |
Personalized learning, powered by intelligence. |
Adaptive personalised learning |
iGuide is an adaptive, personalized teaching system based on computational models of learning and pedagogy.
It integrates principles of mastery learning and scaffolded instruction to guide content sequencing.
The system models user skill and concept mastery in real time, enabling controlled progression and targeted support. |
iGuide is unique because it is not just AI-driven—it is instruction-driven. We embed proven teaching models into the system’s decision logic, ensuring that personalization follows sound learning principles rather than just data patterns. This creates a system that is both pedagogically grounded and technically adaptive, making it more effective and defensible than purely data-driven approaches. |
MVP |
Users, Revenue, Pilots |
Working professionals, EdTech platforms, Students |
$10B – $15B |
$2B – $4B |
$50M – $150M |
Subscription, Saas/Licensing, Freemium, Institutional Partnership |
Duolingo, ByJU's, Khan Academy |
We acquire users through content-led growth (daily challenges), convert via a freemium model, and scale through partnerships and referrals. |
Here’s a **crisp, one-line go-to-market strategy**: > **We launch with a focused student segment using content-led growth and freemium onboarding, then scale through partnerships and B2B integrations.** |
Our long-term vision is to build a cognitive learning platform that adapts to how people think, enabling personalized, scalable, and pedagogy-driven education globally. |
Private limited company |
NA |
Yes |
We are applying to Avinya because it aligns with our goal of building education systems that are both innovative and grounded in real learning outcomes. Avinya’s focus on impactful, long-term learning solutions and its emphasis on education at scale make it an ideal environment to refine and validate our approach. We believe the mentorship and ecosystem can help us translate our technical system into real-world educational impact. |
During the programme, we aim to validate our solution in real learning environments and refine it based on user feedback. We want to demonstrate measurable improvement in learning outcomes and engagement. Additionally, we aim to build a strong MVP, establish early partnerships, and define a scalable go-to-market strategy. |
yes |
Yes |
Core engine |
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/cgYgVTVGvOMr-AdvaitAI-PD.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/cgYgVTVGvOMr-AdvaitAI-PD.pptx |
http://iguide.advaita1.com |
Yes, iGuide is a mission-driven, impact-focused startup aimed at improving how people learn.
We focus on addressing the gap between content delivery and actual understanding by building systems that adapt to how individuals think.
By grounding our approach in pedagogical models and real-world educational insights, we aim to improve learning outcomes, especially in scalable and accessible ways.
Our goal is to make personalized, high-quality learning available to a broader population, beyond traditional classroom constraints. |
NA |
Chaitali Diwan |
checked |
| Apr 8, 2026 @ 6:23 PM |
Manish Shetty |
manish6shetty@gmail.com |
https://www.linkedin.com/in/manish-shetty-519b852b0 |
https://myyuno.com/ |
8792344909 |
Manish - Tech Lead, Rajesh - Lead Developer, Smaran - Marketing & Content Lead |
We’re basically classmates, and after meeting we got to know we’re facing similar issues individually, that’s where the idea was born. We have been working together from August of 2025. |
All |
We keep pushing ourselves in most ways we don’t think of everyday.
We know that balance is a myth.
Our only goal is to achieve what we want to build, and that’s the only priority we have. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 8, 2026 @ 6:04 PM |
Vaibhav |
ventureearth15@gmail.com |
http://www.linkedin.com/in/vaibhav |
https://www.ventureearth.in/ |
7676037130 |
Founder and CEO , Commerce |
We connected through shared interest in building practical, real-world solutions and have been working together closely on this problem for the past 6 months. Our collaboration started with exploring the problem on ground, and has evolved through experimentation, pilots, and execution. |
All |
Our biggest strength is our execution-first mindset. Instead of staying at the idea level, we quickly moved into real-world testing, built prototypes, ran pilots . We combine technical curiosity with ground-level understanding, which helps us build solutions that actually work in real conditions. |
Venture Earth Innovations Private Limited |
Na |
Bidar , North Karnataka |
DeDust builds source-level dust control solutions to reduce urban air pollution in a scalable and resource-efficient way. |
Air pollution is one of the biggest public health crises in India—17 of the world’s 20 most polluted cities are here, and millions are exposed to dangerous PM2.5 levels far beyond WHO limits. This leads to rising cases of respiratory diseases, reduced life expectancy, and massive economic loss. A major but ignored contributor is road and construction dust, which accounts for 30–60% of particulate pollution in cities. Despite this, cities still rely on daily water spraying, which lasts less than an hour and wastes huge amounts of water, fuel, and manpower without solving the root problem. This results in a system where pollution keeps coming back every day, despite continuous spending—making current solutions inefficient, unsustainable, and fundamentally broken. |
We provide a sprayable, water-based formulation that binds dust at the surface, preventing re-suspension for 3–5 days. It works with existing infrastructure, reduces water and fuel usage, and delivers more consistent dust control across urban and industrial environments. |
Our approach shifts from reactive suppression to preventive, source-level control, using a low-dose, non-corrosive formulation optimized for real-world conditions. Combined with field validation and application know-how, this creates strong operational and technical defensibility. |
Product-Market Fit |
Signups |
Municipal corporations, construction companies, industrial facilities, and logistics yards that spend regularly on dust control and are under pressure to reduce pollution and operational costs. |
₹15,000–20,000 crore annually in India across municipal dust control, construction, and industrial environments. |
₹3,000–5,000 crore focused on high-pollution cities and dust-intensive zones like construction and industrial sites. |
₹15–50 crore in the next 3–5 years by capturing a small share (0.5–1%) of the serviceable market. |
Recurring B2G and B2B model—charging per deployment cycle or area covered, based on usage frequency (every 3–5 days), rather than one-time product sales. |
Traditional methods like water sprinkling, chloride-based suppressants (magnesium/calcium), polymer solutions, and mechanical sweeping systems. |
Direct outreach, on-ground pilots, and converting trials into recurring contracts, supported by partnerships with contractors and industrial operators. |
Pilot → validate → convert into recurring contracts → expand city-by-city through contractors and municipal partnerships. |
To build a clean-air infrastructure company that reduces pollution at the source using scalable, resource-efficient solutions integrated with data and environmental systems. |
Done |
NA |
Yes |
We are applying to Avinya because we are at a stage where the right ecosystem can significantly accelerate our progress. Access to industry connections, domain experts, and strong references will help us validate and position the solution more effectively. Equally important is the opportunity for structured pilots and real-world deployment, which is critical for a solution like ours. We are also looking for support in R&D and formulation refinement, where technical guidance and lab access can help improve performance and consistency across conditions. |
During the programme, we aim to: Strengthen and optimize our core formulation through R&D Execute structured pilots in real environments Convert pilots into repeatable deployment models Build strong industry and government partnerships Refine our go-to-market strategy and scalability approach |
Yes, we are open to both hybrid and virtual participation, while also being available for any in-person engagements required for pilots, collaboration, or programme activities. |
Yes |
Supporting feature |
Currently, our core product is material-based (bio-derived formulation), but we are building a data + AI layer on top.
Planned architecture:
Data Layer: AQI sensors (PM10, PM2.5), satellite data, environmental data
Processing Layer: Python-based backend, data pipelines, APIs
Model Layer:
Time-series forecasting (PM levels)
Predictive models for dust re-suspension
Optimization models for spray frequency and dosage
Application Layer: Dashboard, alerts, and decision-support system
Future integrations:
Computer vision (dust density estimation)
AI-based spray optimization systems |
We are building a dataset combining:
Dust behavior vs application frequency
Environmental conditions vs performance
Field deployment data across sites
This creates a real-world operational dataset that is difficult to replicate without deployments. |
Combination of physical + data layer (hard to replicate)
Proprietary field data from pilots
Integration of material science + AI insights
Operational deployment feedback loop |
Reduction in PM10 / PM2.5 levels
Duration of dust suppression
Prediction accuracy (time-series error metrics like MAE/RMSE)
Field validation vs model predictions |
No personal data collection at current stage
Environmental data is non-sensitive
Secure APIs and controlled access
Future compliance aligned with Indian data protection norms |
Minimal on AI side.
Primary risks are on environmental compliance and formulation safety, which are being addressed through controlled testing and validation. |
Data consistency across locations
Sensor calibration differences
Model generalization across environments
We plan to address this through standardized data pipelines and model retraining. |
Currently early-stage, with founder-led development and external support, with plans to build a dedicated AI/ML team as we scale. |
Pilot deployment data (owned)
Environmental and AQI datasets (public sources)
Satellite datasets (open APIs / providers) |
Pilot deployment data (owned)
Environmental and AQI datasets (public sources)
Satellite datasets (open APIs / providers) |
We are currently focused on India, where the problem is most severe and immediate.
However, the solution is inherently scalable to global markets, especially regions like the Middle East where dust and water scarcity are critical challenges.
In the near term, we are building around a bio-derived dust control system, but over time, we plan to evolve into a full-stack clean-air platform, including:
AI-enabled smog control systems (smart sprayers / misting units)
SaaS platforms for real-time PM2.5/PM10 monitoring and prediction
Decision-support systems for cities and industries
Automated deployment optimization using AI
For example:
A city dashboard predicting high dust zones before they spike
Smart systems that trigger targeted spraying instead of manual operations
Industrial clients monitoring air quality compliance in real time |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/gFJ9Sx55jUIu-VentureDustLock-Pitch-DeckV4.pdf_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/gFJ9Sx55jUIu-VentureDustLock-Pitch-DeckV4.pdf_compressed.pdf |
http://NA |
Yes, we are building a mission-driven, impact-focused startup.
Air pollution is not just an environmental issue—it is a public health crisis. India has some of the most polluted cities in the world, with particulate levels far exceeding safe limits, directly contributing to respiratory diseases, reduced life expectancy, and lower quality of life. A significant portion of this pollution comes from road and construction dust, which remains largely under-addressed despite continuous spending.
Our mission is to reduce air pollution at the source in a practical and scalable way. Instead of relying on repetitive, short-term methods like daily water spraying, we are building solutions that deliver sustained impact while reducing resource consumption such as water and fuel.
What drives us is not just solving a technical problem, but creating real-world environmental and social impact. Cleaner air directly improves health outcomes for millions of people, especially workers and communities exposed to dust on a daily basis. At the same time, our approach supports more sustainable urban operations by making pollution control more efficient.
As we evolve, our vision is to build a clean-air infrastructure platform that combines material innovation with data and monitoring systems, enabling cities and industries to move towards measurable, long-term environmental improvement rather than temporary fixes.
For us, impact and scale go hand in hand—if it doesn’t improve real conditions on the ground, it doesn’t matter. |
North Karnataka |
NA |
checked |
| Apr 8, 2026 @ 4:54 PM |
Anurag Kunde |
sparkyaitech@gmail.com |
http://www.linkedin.com/in/anuragk15 |
https://www.sparkyai.in/ |
08123059031 |
Founder and CEO , B.Tech ( Artificial intelligence and machine learning ) |
We met through our shared interest in technology and problem-solving, especially around agriculture. We’ve been working together for over 2 years, building and testing solutions on-ground. |
All |
Strong execution with deep tech + ground understanding of agriculture. |
SparkyAI Private Limited |
https://www.sparkyai.in/ |
Bidar , North Karnataka |
Ai powered satellite imagery analytics . |
Agriculture is a $3T industry still running on guesswork. In places like Karnataka, climate patterns have broken down—rainfall is no longer predictable, temperatures are rising by ~1–1.5°C, and extreme weather events are becoming normal. Farmers can’t rely on experience anymore, but they also don’t have access to real-time data. The result: 30% crop losses 20% higher input costs 85% of farmers operating without diagnostics At the same time, massive amounts of data already exist—satellites, drones, soil—but it’s fragmented, inaccessible, and unusable at scale. The real problem isn’t lack of data. It’s the absence of a usable data layer for agriculture. |
We’re building the data infrastructure for agriculture.
SparkyAI uses satellite imagery + AI + geospatial models to turn raw environmental data into real-time decision intelligence. We process multispectral data (NDVI, NDWI, SAVI), combine it with weather and soil layers, and generate actionable insights for every farm—without any hardware.
Farmers get:
Crop health alerts
Irrigation and fertilizer optimization
Early disease detection
Enterprises and governments get:
API access to farm-level intelligence
Use cases in insurance, credit, land monitoring, and planning
We’re not building another advisory app.
We’re building the backend layer that powers them all. |
Most agritech startups build products. We’re building infrastructure. Our edge comes from: Deep geospatial AI stack using multispectral + hyperspectral data Spatio-temporal models trained on climate variability (rainfall shifts, heat stress patterns) Ability to detect crop stress at a spectral level before it’s visible We don’t rely on IoT or hardware, which makes us 10x cheaper and infinitely scalable. But the bigger vision is beyond agriculture. The same infrastructure can power: Panchayat-level digital land systems Rural mapping and planning Climate intelligence and carbon tracking Even defence-grade terrain analytics In 5 years, we’re not just solving farming. We’re building the geospatial intelligence layer for the real world. |
Users |
Pilots |
We serve a multi-sided market: Smallholder farmers (2–5 acres) who need affordable, real-time farm intelligence Agritech startups & enterprises (insurance, agri-inputs, fintech) that require farm-level data via APIs FPOs and cooperatives managing large farmer networks Governments & panchayats for land mapping, crop monitoring, and climate planning |
$375B+ (Global agriculture + geospatial + precision farming ecosystem) |
$45B (AI-driven agritech, satellite analytics, and farm intelligence platforms) |
$50M+ in 5 years (India + early global expansion through API and B2B adoption) |
We operate a multi-layered data + SaaS model: Subscription (B2C): ₹150–₹400 per acre/month for farmer advisory API & SaaS (B2B): Data access for enterprises (insurance, fintech, agritech) Data-as-a-Service: Selling structured geospatial datasets to corporates & governments Marketplace: Commission from agri-inputs and services |
Global platforms providing satellite data from outside India Agritech companies offering advisory or closed data ecosystems GIS and remote sensing firms focused on project-based services We differentiate by building a real-time, API-first data infrastructure layer, not just a product. |
Offline-first approach through FPOs, field demos, and community engagement Farmer referral networks (high trust, low CAC) Partnerships with agritech companies and input providers Government pilots and institutional collaborations |
Start with high-density farming clusters in Karnataka & Maharashtra Scale through B2B2C model (FPOs, agri-companies, govt programs) Expand into enterprise APIs for insurers, fintech, and supply chains Enter global markets (UAE, Africa, SEA) via pilot-based expansion |
To build the geospatial data infrastructure layer for agriculture and beyond.
In the next 5–10 years, SparkyAI aims to power:
1M+ farmers globally
100+ enterprises through APIs
Use cases beyond agriculture including panchayat-level mapping, climate intelligence, land monitoring, and defence analytics
We’re not building an agritech company.
We’re building the data backbone for the physical world. |
Done |
NA |
Yes |
We’re building something fundamental—a data infrastructure layer for agriculture—and we’re at a stage where the right guidance can significantly accelerate our direction. Avinya offers a strong ecosystem of mentors, operators, and deep-tech founders who understand both scale and complexity. We’re looking for that strategic push to move from a strong product to a category-defining company, especially in terms of positioning, partnerships, and global expansion. |
During the program, we want to: Refine our product into a scalable, API-first data platform Strengthen our B2B strategy and enterprise partnerships Build a clear path toward ₹1–2 Cr ARR and global pilots Improve our go-to-market and storytelling for investors |
Yes |
Yes |
Core engine |
Our system is built as a multi-layer geospatial AI pipeline that ingests satellite, environmental, and field-level data, processes it at scale, and converts it into decision intelligence.
At the data layer, we ingest multispectral imagery from sources like Sentinel-1/2, Landsat, and other providers, along with weather reanalysis datasets and soil layers. These are stored in cloud-based object storage (GeoTIFF, NetCDF) and processed using GDAL, Rasterio, and Geopandas pipelines.
At the modeling layer, we use a combination of classical ML and deep learning approaches. This includes:
Spatio-temporal models for crop growth and yield prediction
CNN-based architectures for image segmentation and LULC classification
Spectral index-based models (NDVI, NDWI, SAVI) combined with regression models for crop health scoring
Experimental pipelines using transformer-based models for time-series geospatial forecasting
We also integrate LLM-based layers (fine-tuned on agricultural and geospatial context) to convert raw outputs into human-readable advisory and decision workflows.
The system is exposed through an API-first architecture, enabling enterprises to consume predictions, classifications, and insights in real time. |
Our key advantage lies in ground-linked datasets—we combine satellite imagery with on-field validation data from farmers, FPOs, and pilot regions. Over time, this creates labeled, region-specific datasets across crops, soil types, and climatic conditions.
We are also building longitudinal datasets across crop cycles, which significantly improves model accuracy and generalization. |
Our defensibility comes from the data + infrastructure combination. While satellite data is public, the cleaning, harmonization, labeling, and temporal stacking of datasets is non-trivial and improves with scale.
Additionally, our models are trained on region-specific climate variability (rainfall irregularity, temperature shifts), making them more accurate in real-world conditions compared to generic global models. Over time, this builds a compounding data moat. |
We evaluate models using a combination of:
Classification metrics: Accuracy, F1-score (for crop classification, LULC)
Regression metrics: RMSE, MAE (for yield prediction, moisture estimation)
Spatial validation: comparing predictions with ground truth samples
Temporal validation: consistency across crop cycles and seasons
We also validate outputs through field trials and farmer feedback loops, ensuring real-world reliability. |
We follow a privacy-by-design approach. Farm-level data is anonymized and stored securely with role-based access control (RBAC). Sensitive datasets are encrypted at rest and in transit.
We align with data localization principles in India and ensure that ownership of farmer data remains with the user or partner organization. |
Potential risks include geospatial data regulations, satellite data licensing, and data-sharing policies. We mitigate this by using open or licensed datasets, complying with national geospatial guidelines, and structuring our platform to adapt to evolving data governance frameworks. |
At 10x scale, the primary challenges would be:
Data processing bottlenecks (large-scale raster computation)
Increased cloud storage and compute costs
Need for more robust data pipelines and orchestration systems
We are addressing this by moving toward distributed processing (parallel pipelines, tiling strategies) and optimizing model efficiency. |
Yes, we have an in-house team with expertise in machine learning, geospatial analytics, remote sensing, and deep learning, working on model development, data pipelines, and system optimization. |
We use a mix of:
Open datasets (Sentinel, Landsat, weather datasets)
Partner datasets (FPOs, pilot projects, field data)
Internally generated datasets through field validation and model outputs
Over time, we are building a proprietary dataset layer through continuous data collection and labeling. |
We improve models through:
Continuous data ingestion and labeling from new regions
Feedback loops from farmers and partners
Retraining models across multiple crop cycles and seasons
Integrating higher-resolution datasets (e.g., drone, hyperspectral)
This creates a self-improving system driven by real-world usage. |
We are building for India-first, global scale.
India provides diverse datasets across crops and climates, making it an ideal training ground. The same infrastructure can then be extended to Global South markets (Africa, Southeast Asia, Middle East) where similar agricultural challenges exist.
Our long-term vision is to become a global geospatial intelligence layer powering agriculture, climate, and land systems worldwide. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/RfJ7wAC9WTOO-SparkyAi_STPI_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/RfJ7wAC9WTOO-SparkyAi_STPI_compressed.pdf |
https://drive.google.com/file/d/18AozMJcFHJcjywSZ-asavPB8Nt8Z3m10/view?usp=sharing |
Yes, SparkyAI is deeply mission-driven. We are building this from a place of lived experience—coming from rural backgrounds where we’ve seen firsthand how farmers struggle with unpredictable weather, rising costs, and lack of access to timely information.
Our goal is to make agriculture data-driven, climate-resilient, and more profitable, especially for smallholder farmers who form the backbone of the global food system. Today, farmers face 30% crop losses, increasing input costs, and growing climate uncertainty, yet most of them still don’t have access to basic diagnostic tools.
Through SparkyAI, we are changing that by making advanced satellite intelligence and AI accessible and affordable, without requiring expensive hardware. The impact is already visible—farmers are saving water, reducing chemical usage, and improving yields.
In the long term, we aim to impact 1 million+ farmers globally, while also contributing to climate sustainability, soil health improvement, and efficient resource utilization. This is not just about technology—it’s about building a system where agriculture becomes more predictable, sustainable, and economically viable for the people who depend on it. |
Yes, I come from a rural background where access to advanced technology, mentorship, and startup ecosystems is limited, making this journey both challenging and deeply meaningful. |
NA |
checked |
| Apr 8, 2026 @ 4:34 PM |
Anurag |
sparkyaitech@gmail.com |
http://www.linkedin.com/in/anuragk15 |
https://www.sparkyai.in/ |
8123059031 |
Founder and CEO , B.Tech ( Artificial intelligence and machine learning ) |
|
one |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 8, 2026 @ 7:33 AM |
H |
yh |
http://hh |
http://hh |
Hh |
Hh |
Hh |
two |
U |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 7, 2026 @ 9:12 PM |
Sulmeet Kaur |
sulmeetkaur@aezo.ai |
https://www.linkedin.com/in/sulmeetkaur/ |
https://aezo.ai/ |
9833172606 |
Sulmeet Kaur, MBA, ISB Hyderabad, 15 yrs of work experience in technology. Led launch of projects /businesses for Tata Group, Amazon India, Govt of AP |
ISB Graduates / classmates, then worked in Govt of Andhra Pradesh as Strategy Consultants |
All |
Extensive experience in eCommerce, MSMEs, Technology |
Aezo Tech Pvt Ltd |
https://aezo.ai/ |
Hyderabad |
Mobile first, Multilingual AI Platform for Market Access for MSMEs |
India’s 70M+ MSMEs are digitally payment-ready, yet fewer than 4% actively sell online, <1% Export due to lack of professional catalogs, marketing content, and technical expertise required for eCommerce and digital discovery. This “missing middle” keeps millions of businesses invisible to national and global markets. While Global MSMEs have AI adoption rate at 8-12%, India's MSMEs languish at under 2%. Without urgent intervention, this competitiveness gap becomes permanent, locking an entire generation of entrepreneurs out of global markets for the next decade. As global businesses rapidly adopt AI, India faces a critical AI adoption window — India’s UPI Moment for Commerce. Without accessible AI tools, MSMEs risk falling behind in competitiveness, exports, and growth. Aezo AI addresses this gap by enabling MSMEs to become digitally discoverable and market-ready through simplified AI adoption. |
Aezo AI is a mobile-first, multilingual AI platform built exclusively for Bharat’s entrepreneurs, enabling MSMEs to access national and global markets by leveraging AI to unlock digital discovery, commerce readiness, and scalable growth. Aezo AI transforms a local MSMEs into digitally discoverable and globally competitive enterprise.
This AI Adoption Window represents India’s UPI Moment for Commerce — just as UPI democratized digital payments for over 400 million Indians, Aezo AI is democratizing market access for 70 million MSMEs. By eliminating technical complexity, cost barriers, and access to professional content infrastructure, Aezo enables entrepreneurs to adopt AI effortlessly.
The economics are transformative: content creation becomes 90% cheaper, 20x faster, while delivering up to 3.5x higher conversion rates. Aezo AI targets onboarding 30 million MSMEs within 18 months, with the potential to unlock over $120 billion in GVA through AI-powered national and global market access.
With a critical 12–24 month adoption window, Aezo ensures India’s entrepreneurs don’t just survive the AI race — they win it. |
Integrated AI Commerce OS: Combines content, proposals, export readiness, design, and prototyping in one platform—driving strong product stickiness. Promptless, Mobile-First: No skills needed; create high-quality outputs instantly from a smartphone. Region-Native AI: Built for India & SEA with cultural, language, and demographic accuracy. Data + Trend Flywheel: Learns from performance and identifies viral formats. 10x–20x Faster, 90%+ Cheaper: Replaces agencies and traditional workflows. |
Users |
Users |
MSMEs (Micro, Small & Medium Enterprises) looking to go digital, create content, and access national and global markets without technical expertise. Marketing & Performance Agencies managing multiple brands, needing high-velocity creative production for reels, ads, and social campaigns. Digital-first D2C brands (Fashion, Apparel, Footwear, Jewellery, Bags, Sports & Fitness, Home Décor) that rely on daily content (1–3 posts/reels) to drive engagement and sales. Social sellers & creator-led brands dependent on trend-driven reels, UGC ads, and consistent content to grow engagement and conversions. |
70M |
50M |
35M |
Aezo AI operates on a usage-based AI Credit Model, enabling SHGs and MSMEs to access enterprise-grade digital capabilities affordably. Users consume AI credits based on platform usage such as product catalog creation, AI image generation, marketing creatives, video creation and market intelligence services. Revenue is generated through subscription plans and prepaid AI credit packages aligned to actual usage, allowing users to scale adoption based on business needs. This pay-per-use approach ensures low entry barriers and predictable costs, enabling small producers to digitize products and access markets without upfront investment, while supporting sustainable platform growth through increased usage and adoption. |
Arcade AI, Photoroom |
Aezo AI adopts a B2B2B and partnership-led customer acquisition strategy to acquire exporters and manufacturers at scale. Customer acquisition channels include: Collaboration with Manufacturing Clusters and SEZ Offices Partnerships & Paid Promotions with Export promotion councils and Trade Associations Collaboration with Government MSME, export enablement programs, regional DICs, MSME-DI Institutes, Industrial COEs Corporate partnerships such as Bands, NBFCs, Logistics Service Providers Digital Marketing channels for reaching out to MSMEs Institutional partnerships enable bulk onboarding of supplier communities, while Aezo AI’s mobile-first, multilingual platform supports self-serve adoption. This hybrid approach ensures cost-efficient acquisition and scalable onboarding of MSMEs seeking global market access and export growth. |
Aezo AI adopts a B2B2B and partnership-led customer acquisition strategy to acquire exporters and manufacturers at scale. Customer acquisition channels include: Collaboration with Manufacturing Clusters and SEZ Offices Partnerships & Paid Promotions with Export promotion councils and Trade Associations Collaboration with Government MSME, export enablement programs, regional DICs, MSME-DI Institutes, Industrial COEs Corporate partnerships such as Bands, NBFCs, Logistics Service Providers Digital Marketing channels for reaching out to MSMEs Institutional partnerships enable bulk onboarding of supplier communities, while Aezo AI’s mobile-first, multilingual platform supports self-serve adoption. This hybrid approach ensures cost-efficient acquisition and scalable onboarding of MSMEs seeking global market access and export growth. |
To become the default AI Commerce Infrastructure for Indian MSMEs, powering how they create, sell, and scale across India and global markets. Aezo AI aims to be the operating system through which every MSME can access digital discovery, generate content, connect with buyers, and grow their business.
A core part of this vision is building sovereign AI capabilities, including India-first image and video models, ensuring culturally accurate, compliant, and locally optimized content at scale.
Over time, Aezo will evolve into a full-stack growth engine—integrating demand intelligence, buyer-seller matching, export enablement, and supply chain insights.
Our vision is to enable millions of Indian MSMEs to become globally competitive brands through AI, unlocking large-scale economic growth and exports. |
Registered as Private Limited |
NA |
Yes |
We are applying to Avinya to accelerate Aezo AI’s mission of building AI-led market access infrastructure for Indian MSMEs at scale. Avinya’s focus on deep-tech, India-first innovation, and real-world impact aligns strongly with our vision of creating sovereign AI capabilities for commerce. We see Avinya as a strategic partner to help strengthen our technology stack—especially in areas like India-first image and video models, multilingual AI, and scalable infrastructure—while also enabling deeper ecosystem integrations. Beyond capital, we value Avinya’s network, mentorship, and long-term approach to building category-defining companies that can drive national impact. |
Scale GTM & Distribution: Build strong distribution channels through ecosystem partners, marketplaces, agencies, and cluster-led onboarding to drive large-scale adoption. Accelerate MSME Onboarding: Design and execute high-velocity onboarding strategies across key sectors and regions to rapidly grow the user base. Partnerships & Market Access: Establish integrations with commerce platforms and networks to enable seamless selling and discovery. Expand into New Markets: Pilot expansion into Southeast Asia and global markets with similar user needs and behavior. |
Yes, we are open to both hybrid and virtual participation as needed. We will be available for in-person engagements as needed in the program. |
Yes |
Core engine |
Problem with Generic Third-Party Models
Foundation image and video models such as Google Nano Banana/Veo, OpenAI DALL·E/Sora and Flux series are primarily trained on Western datasets. As a result, they often fail to accurately generate:
Indian Demographics: Accurate representation of Indian facial features, skin tones, body diversity and regional appearance variations across geographies
Indian Crafts & Products: Context-aware rendering of traditional products such as Ikat (Odisha vs Pochampally), Banarasi vs Kanjivaram saree motifs, and specialised crafts including Kondapally toys, Madhubani art and Srikalahasti Kalamkari, preserving regional authenticity
Indian Cultural Contexts: Visual Generation aligned to diverse cultural settings such as regional wedding traditions (Andhra Pradesh, Kerala, Rajasthan), festivals including Diwali, Pongal and Ugadi, and variations in celebrations like Sankranti across North and South India, along with culturally accurate Indian home interiors and décor.
Local Environments & Lifestyles: Realistic depiction of Indian bazaars, non-metro cities, neighbourhood retail settings, regional architecture, everyday life scenes and rural–urban landscapes, reflecting Bharat’s consumption environments rather than globalized studio settings
This leads to inaccurate catalogs, ads and social media content that reduce marketplace acceptance and consumer trust..
Aezo AI’s Proprietary Improvement
Aezo AI overcomes limitations of generic foundation models through a proprietary Cultural & Demographic Skills Framework, acting as an intelligence layer over image and video AI models. The Aezo AI Agent:
Injects Indian cultural context, regional aesthetics and demographic realism
Grounds generation in authentic product usage, festivals and local environments
Enables promptless creation of culturally accurate commercial content
Benchmarking & Performance Outcomes
~90% improvement in cultural and contextual accuracy
3× higher first-pass catalog acceptance without regeneration
Generation retries reduced from 5–7 iterations to 1–2
Validation: A/B testing with 50+ MSMEs, user evaluations, expert reviews
Outcome: Aezo AI converts generic AI models into culturally intelligent commerce engines, enabling scalable, market-ready digital content creation for Indian MSMEs and SHGs. |
|
|
|
Aezo AI follows secure-by-design data governance practices aligned with data privacy and security standards. The application is live on App Store and Play Store and actively used by enterprises. All data transmission occurs via encrypted HTTPS protocols, with secure cloud storage compliant with Google Cloud security standards. Access is governed through role-based authentication controls, and only user-provided inputs required for AI processing are utilized. Aezo AI does not sell or share personal data. Data retention and deletion policies are implemented on request. The company is DUNS registered and compliant with applicable data protection and privacy requirements. |
|
|
Yes |
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/quBCrK4gRlBb-Aezo-AI-Pitch-AIH-Final-14-1_compressed-1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/quBCrK4gRlBb-Aezo-AI-Pitch-AIH-Final-14-1_compressed-1.pdf |
https://aezo.ai/applinks |
Yes, Aezo AI is a mission-driven startup focused on democratizing market access for Indian MSMEs. While millions of businesses are digitally payment-ready, most remain invisible to markets due to lack of content, discoverability, and access to digital commerce tools.
Aezo addresses this by making AI accessible on mobile—enabling entrepreneurs to create content, connect with buyers, and become export-ready without technical expertise. By reducing cost barriers by up to 90% and accelerating go-to-market, we unlock income growth, job creation, and global competitiveness for small businesses.
Our impact lies in enabling millions of entrepreneurs—especially in Tier 2/3 and rural India—to participate meaningfully in digital and global commerce, driving inclusive economic growth at scale. |
Women Founder |
NA |
checked |
| Apr 7, 2026 @ 2:12 PM |
Mohamed Aejaz Ahamed |
aejazahamed57@gmail.com |
https://www.linkedin.com/in/aejaz-ahamed-aa449bb0/ |
http://NA |
9738547792 |
Founder, CEO and CTO, Product Development |
We are brothers.
I work with Kommunicate as a Solutions Engineer and my brother is a computer science final year student. |
two |
Our biggest strength is execution speed combined with product clarity.
We focus on identifying real user pain points and quickly turning them into working solutions. DevGuard was built as a functional MVP with a strong emphasis on real-time usability and simplicity, rather than over-engineering.
We also bring a strong product mindset to security, focusing on user behavior and experience. This allows us to build solutions that are not only technically sound but also practical and easy to adopt. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 6, 2026 @ 7:20 PM |
Shailesh Kachi |
kachi.shailesh@adevair.com |
https://www.linkedin.com/in/shailesh-kachi/ |
https://adevair.com |
9673816155 |
Founder and CEO |
Shailesh Kachi (Founder) and Vasant Kachi (Co-founder) are father and son. We have worked together since incorporating Adevair Technologies in 2019 — over 6 years. Vasant brings 35+ years of government infrastructure experience, while Shailesh brings deep-tech IoT/AI expertise. |
2 |
Our complementary expertise and shared commitment. Shailesh brings 9+ years of IoT/AI deep-tech experience, double Master's from London, and product development rigor. Vasant brings 35+ years of government infrastructure execution — including large-scale, delay-prone project management. Together, we bridge cutting-edge technology with real-world deployment. Both are fully invested, having bootstrapped with ₹1.3 crore of personal savings over six years. |
Adevair Technologies Private Limited |
https://adevair.com |
Bengaluru |
Securio by Adevair Technologies is an AI-powered wearable that measures breathing-zone air quality, tracks posture, and enables hands-free SOS. |
Urban commuters breathe toxic air daily, but existing solutions fail. Smartwatches ignore the environment. Government monitors are sparse. Navigation apps rely on satellite estimates that cannot detect pollution spikes at traffic signals, construction zones, or underpasses. Air quality cannot be estimated at street level — it must be measured in real time at breathing zone. No existing product does this. |
Securio is an AI-powered wearable that measures real-time air quality at breathing zone, tracks posture with haptic nudges, and enables hands-free SOS. Every device becomes a mobile data node, anonymously uploading geotagged pollution data to build the world's first street-level, crowd-sourced air quality map — enabling healthier routes and smarter cities. Patent pending. Launching September 2026. |
Securio's unique approach is protected by a published pending patent (Application No. 202421100144) for "A System and a Method for Real-Time Environment Monitoring and Health Tracking." Additionally, Adevair holds a granted patent (No. 536729) for threat assessment methodology. PCT filed for international expansion. |
MVP |
Signups |
Urban commuters in high-AQI Indian metros (Delhi, Bengaluru, Mumbai, Chennai) — professionals, two-wheeler riders, parents, delivery partners, and asthma patients. Secondary targets: enterprises including logistics companies, insurance providers, and corporate wellness programs. Also smart city developers for DaaS licensing. |
$150,000,000,000 |
$50,000,000,000 |
$2,000,000,000 |
Three-layer: Hardware (one-time sale, 55-66% margin) + AURA Core filters (recurring, 3-4 per user/year, 70-80% margin) + DaaS (enterprise API licensing, 90%+ margin) |
Dyson Zone, LG PuriCare (wearable purifiers with static timers). BreezoMeter (satellite-based AQI estimation, powers Apple Maps). No direct competitor combines environmental measurement + biological tracking + data network. |
D2C via website and app with performance marketing (Google, Meta, LinkedIn). Content-led growth (LinkedIn, Instagram, YouTube). PR and media outreach. B2B2C channels: corporate wellness, logistics, insurance. Referral and affiliate programs. 400+ organic waitlist already validated. |
Phase 1: D2C launch in Indian metros (Delhi, Bengaluru, Mumbai, Chennai) with paid marketing and PR, leveraging 400+ waitlist. Phase 2: B2B2C partnerships (corporate wellness, logistics, insurance). Phase 3: International expansion via PCT. Retention through filter replacement cycle and app engagement. DaaS revenue scales with network density. |
To become the global infrastructure layer for real-time environmental risk awareness. Every Securio device becomes a node in the world's largest street-level air quality measurement network — enabling healthier routes for navigation apps, risk assessment for insurers, and data-driven decisions for smart cities. We don't just sell wearables. We map the invisible. |
Incorporated in Jan 2019 |
0 |
Yes |
Avinya's AI focus, ₹10 lakh seed funding, IIT alumni network, and founder-first mentorship will accelerate Securio's September 2026 launch. The pilot testing and ESG focus align perfectly with our wearable AI and sustainable filter model. |
We want to achieve three specific outcomes: First, product readiness. Complete our remaining tooling (currently 30%), finalize BIS/CE/FCC certifications, and execute our 1,000-unit pilot production run — moving from working MVP to launch-ready. Second, investor readiness. Refine our financial model and investor narrative with IIT alumni mentors, and close our ₹2 crore bridge round through the programme's investor network and demo day. Third, go-to-market validation. Leverage pilot testing opportunities to generate real-world user data, secure 3-5 LOIs from enterprise partners (navigation, logistics, insurance), and validate our filter replacement rate assumption before September 2026 launch. The 16-week structured mentorship, ESG integration focus, and access to IIT alumni expertise will accelerate us from pre-launch to market entry. |
Virtual |
Yes |
Core engine |
Securio uses on-device TinyML for real-time inference with no cloud dependency.
Models: Custom lightweight neural networks for posture detection (accelerometer/gyroscope fusion) and SOS gesture recognition (pattern classification with false-positive rejection).
Stack: TensorFlow Lite Micro for model training and deployment. C++ for embedded inference. Models run on low-power microcontroller (ARM Cortex-M series).
Frameworks: Custom data collection pipeline for posture and gesture training data. On-device inference only — no API calls.
Hardware: Accelerometer, gyroscope, haptic motor, LED ring.
No cloud APIs (OpenAI, Gemini, etc.) are used. This ensures sub-second response, offline functionality, and user privacy. |
Securio's proprietary data advantage is measurement, not estimation.
Unique data source: Every Securio node collects real-time, geotagged, breathing-zone AQI data. No competitor has this data. Government monitors are static and sparse. Satellite models (BreezoMeter) estimate from above.
Network effect advantage: More devices = denser data = higher value. Competitors cannot replicate this data network without deploying their own hardware at scale. Our first-mover advantage compounds as network grows.
Training data for AI: Proprietary posture and gesture data collected during our Delhi and Bengaluru testing. This data improves our on-device models and cannot be easily replicated.
Anonymized, aggregated data asset: Post-launch, the crowd-sourced pollution map becomes a licensable asset (DaaS) for navigation apps, insurers, and smart cities — creating a data moat that widens with every user. |
Four layers of defensibility:
First, on-device specialization. Our TinyML models are optimized for specific wearable use cases (posture, SOS gestures). General-purpose AI models cannot achieve the same latency (sub-second) or battery efficiency on resource-constrained hardware.
Second, proprietary training data. We have collected real-world posture and gesture data during our testing in Delhi and Bengaluru. Competitors would need to replicate this data collection.
Third, patent protection. Our pending patent (202421100144) covers the AI-integrated wearable for environment and health tracking. Our granted patent (No. 536729) covers threat assessment methodology.
Fourth, hardware integration. AI runs on our specific sensor suite (accelerometer, gyroscope, PM sensor, gas sensors). Copying the AI alone is insufficient without the complete hardware stack. |
Posture detection metrics: Accuracy (target >90%), precision, recall, F1-score. False positive rate (minimizing unwanted haptic nudges). Latency (target 95% target for safety-critical feature). False positive rate (extremely low – false SOS triggers erode trust). False negative rate (minimized – missed triggers defeat purpose). Latency (<500ms from gesture to alert).
Validation method: Cross-validation on collected dataset. Real-world testing during Delhi/Bengaluru pilot. Continuous improvement via post-launch anonymized feedback (user corrections, dismissals).
Model drift monitoring: Periodic re-evaluation with new data. Retraining as needed.
Hardware constraints: Battery consumption (target <5% daily for AI inference). Memory footprint (fits within microcontroller limits). |
Privacy-first by design: All AI inference (posture, SOS) runs on-device. No personal data leaves the device. User location and AQI data are anonymized before upload.
Consent-based data sharing: Users explicitly opt-in to share anonymized pollution data for DaaS and network mapping. No opt-out, no data collection.
Compliance: Working toward BIS (India), CE (Europe), and FCC (US) certifications. GDPR and India's Digital Personal Data Protection Act (DPDPA) compliance planned for post-launch.
Security: Encrypted BLE communication between device and phone. Secure cloud infrastructure (AWS) with access controls. No sensitive user data stored.
Transparency: Clear privacy policy and in-app controls. Users can delete their data at any time. |
Yes, but manageable.
Air quality data claims: We must avoid unsubstantiated health claims. We measure and report AQI – we do not diagnose or treat medical conditions. Legal review of marketing language is planned.
Medical device classification: Securio is not a medical device. It is a wellness wearable. We will ensure all communications maintain this distinction to avoid regulatory reclassification.
Data privacy regulations: India's DPDPA, GDPR (Europe), and CCPA (US) apply post-launch. We are designing for compliance from day one – anonymization, consent, user controls, data deletion.
Certifications: BIS (India), CE (Europe), FCC (US) are in progress (currently 15% complete). Target completion August 2026 before September launch.
PCT and international patents: Filed. No regulatory risks identified for IP.
Battery and electronics compliance: Standard certifications (RoHS, WEEE) planned.
No major regulatory barriers identified that would prevent launch or scaling. |
Potential breaking points:
First, on-device inference is fine. Our TinyML models run locally on each device. Scaling from 1,000 to 10,000 units does not impact per-device performance. No server load for inference.
Second, data aggregation backend will need scaling. Currently designed for serverless AWS. At 10x scale, we may need to optimize database queries, add caching layers, and increase cloud capacity. This is solvable with standard cloud scaling practices.
Third, model retraining pipeline. As we collect more posture and gesture data from thousands of users, our training dataset grows. We need automated retraining pipelines (CI/CD for ML) to update models without manual intervention. Currently manual. We plan to implement automated pipelines before scaling beyond 10,000 units.
Fourth, edge cases in gesture recognition. More users = more diverse movement patterns. False positives/negatives may increase. Mitigation: continuous monitoring, A/B testing of model versions, user feedback loops.
Nothing fundamentally breaks. All challenges have known solutions. |
Yes, but small. |
Pre-launch training data (posture, SOS gestures) was collected internally by Adevair and is owned by Adevair. No users involved.
Post-launch: Users own their personal data. Adevair owns anonymized, aggregated AQI and environmental data collected by Securio nodes. Users opt in to share this data. Individual user data is never sold. Environmental insights are licensed to enterprise partners (DaaS). Full compliance with DPDPA and GDPR. |
Both, with phased approach. |
For both India and Global Market |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Vl94QYYI5SaW-Securio-SOS-DaaS-2.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Vl94QYYI5SaW-Securio-SOS-DaaS-2.pdf |
https://drive.google.com/file/d/1MiUeNU40pC5IdXOMZ4Zh9XTk4IbdMfJL/view?usp=sharing |
Yes, Securio is a mission-driven startup. Our mission is to make invisible environmental risks visible and actionable for everyone.
Health impact: Air pollution causes 7 million premature deaths annually. Urban commuters in high-AQI cities breathe toxic air daily without real-time awareness. Securio empowers individuals with breathing-zone AQI measurements, posture AI for better breathing efficiency, and hands-free SOS for personal safety.
Environmental impact: Our AURA Core filter is biodegradable. The magnetic tray is reusable for 12+ months. Our return-for-credit model (₹20 off for returning used filters) incentivizes recycling and reduces waste — 90% less waste than disposable N95 masks.
Data for good: Every Securio node contributes to the world's first crowd-sourced, street-level pollution map. This data can help city planners identify pollution hotspots, navigation apps offer healthier routes, and policymakers make evidence-based decisions.
SDG alignment: Goal 3 (Good Health and Well-being) and Goal 9 (Industry, Innovation and Infrastructure).
We are not just building a product. We are building infrastructure for healthier cities and informed citizens. |
NA |
NA |
checked |
| Apr 6, 2026 @ 5:48 PM |
susmita Rajib kumar Roy, pankajl alok dheer, rajib kumar roy |
susmita.rajib@rbcind.pro |
https://www.linkedin.com/in/susmita-r-ab734b330/ |
http://www.rbcind.pro |
6289595007 |
Susmita Rajib kumar roy is founder director: Manages vendor/legal/ecosystem relationships, Dr. rajib kumar roy is CEO : manages all business function strategy to execution, tech development engineering, pankaj alok dheer is CMO: manages marketing connects and pipeline manaement |
Susmita Rajib & Rajib is husband and wife, we picked up pankaj through a founder requirement post from linkedin during 2024 august and from then we are working together |
All |
1. we are developing a product/solution based on our long survey based knowledge identification.
2. we are able to develop a team , remain committed to Dr. Roy's & team's vision of implementing "Chie"
3. we remain committed on "what we say, we remain committed to deliver" |
chiestra racehorse business pvt.ltd |
www.rbcind.pro |
bangaluru |
Chie is a Verticalized Compound AI System that performs a forensic, 360-degree "MRI" of startups and MSMEs to identify structural gaps and deliver an investor-ready strategic roadmap within 3-5 working days. |
Chie solves the "Structural Blind Spot" by replacing subjective, high-cost manual audits with an automated, 360-degree forensic "MRI" that identifies hidden operational and strategic gaps in 3–5 working days. |
We are developing Chie, an Adversarial AI Expert Engine designed to automate Tier-1 consulting rigor (specifically designed for incubator/accelarators). Chie is a Deterministic Diagnostic System that. We help startup to identify their business gaps and align business model by executing RBC's expert engine "Chie". Chie works like medical MRI for human body. Chie does the same for entire Business model, by identifies 100% of gaps, reducing current cost of effort to identify the gaps to 85% and improving >500% efficiency on gap discovery, impact analysis, progressive plan on across all function and operation of startup's business model. Chie will be a lifecycle partner for startup's progression and sustainability determination. Once the gap is analysed, RBC is having a process of business consultation to bridge the gap. |
Intellectual Property Moat ( we are planning to file three to five patent for core business logic). Chie’s defensibility lies in its Verticalized Compound AI System, which uses a proprietary "Adversarial Skeptic" engine and a patented 360-degree Scorecard Logic to cross-verify founder claims against live market and statutory data. Unlike generic LLM wrappers, Chie provides a deterministic, high-fidelity "Truth Score" and a 15-month execution roadmap, creating a structural "Trust Moat" that is impossible to replicate with simple prompt engineering. |
MVP |
Pilots |
B2B: all incubators from university to private, government. B2C: startups and MSME for Tier -1 assessment |
55250 cr. INR |
2800 cr. INR |
38 cr. INR |
B2B (80%) & B2C (20%) |
there is no such similar solution, we are bringing the top MNC comapny like BCG/KPMG/capgemini's consulting gap analysis model to indian startups/msme |
startups in B2B through incubator, incubators through ecosystem connects by attending events, provide talks by complementing incubators operational efficiency |
Position Chie as a "Mentorship Force Multiplier" that automates high-fidelity due diligence, allowing incubators to provide 360-degree forensic roadmaps to their entire cohort with 85% less manual effort. Implement an Institutional Integration model where Chie is mandated as the standard "Maturity Assessment" tool for cohort selection, graduation milestones, or government grant eligibility. Provide leadership with a "Portfolio Health Dashboard" that uses Chie’s deterministic "Truth Scores" to de-risk capital allocation and significantly increase the "Investor-Readiness" of their startups |
create LLM for indian startups business consultation in 3 years timeline |
Priate limited |
Boot strapped |
Yes |
We need grant and focused ecosystem connects with investors and B2B/B2C customers |
Our target is to reach TRL level by next 4-6 weeks by executing at least 20-30 startups in operative enironments |
virtual participation |
Yes |
Core engine |
CHIE is a Next.js 14 (App Router) application backed by a PostgreSQL database (via
Prisma ORM) and MinIO for S3-compatible object storage.
Core Components
Frontend: Next.js 14, React, TailwindCSS, Framer Motion
Backend: Next.js API Routes (Serverless functions)
Database: PostgreSQL (managed via Prisma)
Storage: MinIO (S3 compatible) for document uploads
AI Engine: Multi-agent LLM pipeline (Normalizer, Validator, Scorer, Risk,
Synthesizer) using OpenAI/Anthropic/Gemini via OpenRouter
External APIs: Cashfree (Payments), InstaFinancials/Zauba/Tofler (CIN/PAN
Verification), ZeptoMail (Transactional Emails)
AI Engine Layers
The AI Engine is strictly separated into 4 layers ( src/lib/ai-engine/layers/ ):
1. Instructions: Static versioned system prompts per agent.
2. Memory: Ephemeral per-run context assembly (intake + KYB + signals).
3. LearningHistory: SME feedback patterns, read-only at inference.
4. UsageAccounting: Token/cost/latency tracking per LLM call ( ApiUsageLog ). |
our core logic in patentable, team is working on skeptic and stress test assumption model and finalizing at least 3-5 technical/process patents |
Chie’s defensibility lies in its Verticalized Compound AI System, which uses a proprietary "Adversarial Skeptic" engine and a patented 360-degree Scorecard Logic to cross-verify founder claims against live market and statutory data. Unlike generic LLM wrappers, Chie provides a deterministic, high-fidelity "Truth Score" and a 15-month execution roadmap, creating a structural "Trust Moat" that is impossible to replicate with simple prompt engineering. |
very nice question,
1. for operational performance we will carry out page reasponse time and stress/reliability test (concurrent access to discoery).
2. for data and analsysis report we will have inbulit audit and review process along with HITL
top measurement areas for data analysis and reports are
Adversarial Detection Rate
Grounding Faithfulness
Expert-Approval Velocity |
we are aligned to DPDP act and we will be going for ISO 27001 |
Yes , we need to hae iso 27001 to start with, we are already iso 9001:2015 |
I believe we should have stronger data center as backbone with stronger load balancing process. the process of having multiple servers like dev-test-demo-staging-prod-DR need better infra |
YES we have, we have a team of three strong technical team having average experience of 18 years in top MNCs. |
every data set/system we will use are either licensed /opensources. for LLMs we will be using higher ersion for better precisions |
1. review HITL-feedback from end users-feedback to product change management
2. industry connect for technology change and process change management |
we are currently focused on india market with 5-6 llocal languages intially with next 6-8 weeks. after 1 years in production we can go to any market, customisationlocalisation will take only 6-8 weeks time |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/3kcsTXRZOCPJ-Chie-GA-2026.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/3kcsTXRZOCPJ-Chie-GA-2026.pdf |
http://MVP%20link%20will%20be%20available%20within%20next%202%20weeks |
It's an impact focused startups
As, Chie ( It's an AI based expert engine of RBC) will provide the dynamic questions to startup founders based on sector, industry, type of product, product/solution definitions, target markets either hyper local, domestic india, domestic global , competitors, challenges and once those questions are answered, chie will validate the adequacy of the answers and post that chie will produce startup gap analysis, impact on business operation/performance due to gap, competitors landscape, sectorial outlook, develop strategy based on gap and many other analytic report from multiple LLM using agentic AI and MCP server. |
NA |
KDEM and susmita rajib (director) |
checked |
| Apr 5, 2026 @ 7:26 PM |
Ash Bagla |
abagla@ucdavis.edu |
https://www.linkedin.com/in/ash-bagla/ |
https://wellvantage.ai/ |
5309658347 |
CEO |
iim-l |
2 |
Respect |
WellVantage |
https://wellvantage.ai/ |
Noida |
"WellVantage’s long-term goal is to build a personalized nutrition and health platform. Currently, it operates as an integrated B2B–B2C fitness system where gyms manage memberships and attendance, while users access workouts, wellness content, and AI-powered nutrition tracking to improve engagement and retention. " |
"WellVantage’s long-term goal is to build a personalized nutrition and health platform. Currently, it operates as an integrated B2B–B2C fitness system where gyms manage memberships and attendance, while users access workouts, wellness content, and AI-powered nutrition tracking to improve engagement and retention. " |
Gyms face high churn due to poor engagement and lack of differentiation, while users rely on fragmented apps and generic plans that fail to deliver real health outcomes. Rising metabolic diseases require integrated, personalized solutions. WellVantage is an AI-powered platform combining gym management software with a consumer wellness app to improve retention, engagement, and personalized health outcomes. |
We combine gym SaaS with an AI-driven wellness app, creating a closed-loop system of real-world data (nutrition + activity). This enables scalable, personalized recommendations while providing gyms with a low-CAC distribution and retention advantage. |
Users |
Signups |
Gyms and B2C Cutomers |
See Pitch |
See Pitch |
See Pitch |
B2B SaaS (gym subscriptions), B2C premium subscriptions, and B2B2C via gym distribution. Future revenue from insurance, employers, and partnerships. |
Mindbody, HealthifyMe, MyFitnessPal, and generic gym CRM software. |
Primary: gyms (B2B → B2C distribution). Secondary: referrals and organic growth via gym members. |
Phase 1: onboard mid-sized gyms. Phase 2: expand via member referrals and boutique studios. Phase 3: partnerships with insurers, employers, and retailers. |
Build a precision health platform delivering personalized nutrition and preventive care using real-world behavior data. |
Incorporated |
NA |
Yes |
We are building a preventive health platform and need clinical validation, healthcare partnerships, and guidance to transition from fitness-tech to a scalable, personalized healthcare solution. |
Validate our personalized nutrition model, build healthcare/insurance partnerships, strengthen scientific credibility, and define a clear path to scale as a precision health platform. |
Yes |
Yes |
Supporting feature |
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/vmDdElEYUe31-India-WellVantage-Pitch.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/vmDdElEYUe31-India-WellVantage-Pitch.pdf |
https://drive.google.com/drive/u/0/folders/12oWnwW22lFbAOMGSlcK1KnO0EPS7tztl |
Yes. We are building a preventive health platform addressing rising lifestyle diseases. Our focus is shifting from reactive care to daily, personalized interventions using nutrition and activity data. By making health measurable and actionable, we aim to improve healthspan, reduce long-term disease burden, and make preventive care scalable and accessible. |
Yes LGBTQ |
Website |
checked |
| Apr 4, 2026 @ 12:26 AM |
Sameep Tripathi |
carrusmobilitsolutions@gmail.com |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 3, 2026 @ 5:06 PM |
Ajay Nair |
ajay@ai-swar.com |
https://www.linkedin.com/in/ajay-nair-6a8369/ |
https://swar-ai.com |
+91-8762803277 |
Founder CEO |
We have been colleagues earlier and know each other for about 10 years |
2 |
Together we have over 55 yrs of deep tech and systems experience |
Swar AI |
https://swar-ai.com |
Bangalore |
Worlds only sub Rupee Voice AI solution |
Telecalling by AI systems |
Ours is the worlds only sub Rupee Voice AI. Coneventional Voice AI players do not offer benefits in terms of economics especially to price concious customers in India and elsewhere. |
We beat our competition on economics by 10X |
Users |
Pilots |
BFSI, Fintech |
400billion$ |
200billion$ |
250Million$ |
SAAS |
Yellow.ai,Gnani.ai |
Advertising and network |
Partnerships with existing players |
We are positioned to capture a substantial part of the telecalling business in India. |
Private Limited |
0 |
Yes |
I believe being an IIT alumnus, Avinya may be more relevant |
Investment, mentorship, network |
Yes |
Yes |
Core engine |
ASr/LLM/TTS |
We know how to optimize GPU usage. Thats the unfair advantage we have over competition |
We are at a miinimum 10X more economical than our competitors |
Latency, IFEVAL score |
Data is entirely on our VMs and never exposed to any external interfaces or network |
That is actually an advantage for us as we dont use any Cloud APIs |
Nothing |
55 yrs of end to end AI and systems expertise |
We use pretrained models so dont need datasets |
Keep reading about newer and better opensource models available |
Primarily India to begin with. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/5czt0GbvK2pT-Swar-AI-2.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/5czt0GbvK2pT-Swar-AI-2.pptx |
https://swar-ai.com/demo |
We aim to create an impact by making telecalling services accessibile to every business. |
No |
MD Ramaswami |
checked |
| Apr 3, 2026 @ 3:46 AM |
Vaibhav Sharma, Agniva Ray |
vaibhav@kubar.tech |
https://www.linkedin.com/in/vaibhavsharma7403/ |
https://kubar.tech |
+91 7872260684 |
Vaibhav Sharma — Founder & Head of Product. Agniva Ray — Co-Founder & COO. |
Met during undergrad, worked together 18+ months through 3 pivots in 2025. |
2 |
Complementary depth — product+AI (Vaibhav) and operations+BD (Agniva). Clix Capital live lender, AgriGrader exclusive partnership, RBI FinVision Award — all without external capital. |
Kubar Labs |
https://kubar.tech |
Bengaluru |
NavDhan — AI-powered embedded credit infrastructure converting MSME transaction data into creditworthiness. |
63M MSMEs face Rs 25-30L crore credit gap. Traditional lenders reject 80%+ for lack of formal credit history. |
NavDhan embeds credit infrastructure into B2B platforms MSMEs already use. Sector-specific AI on 150+ data points enables 35-40% approval rates at Rs 74/loan vs Rs 3,000-5,000 industry standard. |
Sector-specific AI, exclusive AgriGrader data, embedded distribution with near-zero CAC. |
Revenue |
Revenue, Pilots |
B2B marketplaces, NBFCs/banks, thin-file MSME borrowers in Tier 2/3 cities. |
Rs 25-30 lakh crore |
Rs 6-8 lakh crore |
Rs 500-800 crore by Year 3 |
Platform API fee + lender commission (5-10 bps) + data intelligence subscription. Asset-light. |
CreditEnable, CredAble, Aye Finance, FinBox, NBFC DSA networks. |
Platform-embedded — AgriGrader (exclusive) and Clix Capital (live pilot, 25 borrowers). |
Phase 1: 1,000 loans via Clix+AgriGrader. Phase 2: 2 more lenders + 2 verticals. Phase 3: ONDC open API. |
Intel Inside of MSME credit — essential intelligence layer for every lender/marketplace. 10M profiles, Rs 1L crore facilitated, then SEA expansion. |
Registered Private Limited — Kubar Labs Private Limited, India |
No external equity. Rs 2M+ in grants: NVIDIA Inception, Google, Microsoft, WTFund. |
Yes |
ACSEL's AI-native Karnataka focus aligns with NavDhan. Seeking lender network access, regulated AI mentorship, and Rs 10L for Q2 2026 integrations. |
1. Onboard one Karnataka NBFC via ACSEL. 2. Integrate one Karnataka agri-marketplace. 3. Validate AI on 500 Karnataka MSME profiles. |
Yes, open to hybrid and virtual. |
Yes |
Core engine |
XGBoost/LightGBM/RF ensemble on 150+ features. AA+GST+UPI data. SHAP explainability. React widget SDK + WhatsApp Agentic DSA. eNACH/UPI lifecycle automation. |
Exclusive AgriGrader data + MSME credit graph. 18-24 month replication advantage. |
Data network effects, exclusive distribution, regulatory-first architecture. 12-18 month compliance build advantage. |
Approval rate uplift (35-40% vs 8-15%), 90-day NPA EWS, weekly PSI, 100% SHAP attribution. |
AA consent arch, DPDP Act 2023, RBI DL Guidelines, AES-256/TLS 1.3, OAuth 2.0, no PII in training. |
Pure LSP (no DLG), AA fallback, NBFC Year 3, modular for ONDC policy changes. |
Async queuing for lender API limits, MLflow MLOps, Agentic DSA escalation, event-driven AA ingestion. |
Yes — Vaibhav (AI/ML arch), Ayan Guha (conversational AI), contract ML engineer. Python, XGBoost, FastAPI, SHAP, MLflow. |
AA (RBI-certified), GST API (NIC), AgriGrader (exclusive), CIBIL/Experian, Clix Capital pilot data. DPDP compliant. |
Monthly Clix updates, GeM/IndiaMart 50+ features Q3 2026, champion-challenger shadow testing, lender feedback. |
India-first, then Indonesia/Vietnam. DPI-agnostic architecture. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/0HxpbIO6hAeC-NavDhan_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/0HxpbIO6hAeC-NavDhan_compressed.pdf |
https://drive.google.com/file/d/1kH2a9DwAyKyPJX-aol0KIT12wMfCPit1/view?usp=sharing |
90% borrowers are first-time credit users. FinVision Award from NIBM/RBI recognizes inclusion impact. Every loan is an MSME traditional finance wrote off. |
Vaibhav — first-gen entrepreneur from Noida, no legacy network. Agniva — first wave from his community in deep fintech. |
None |
checked |
| Apr 3, 2026 @ 1:16 AM |
Dinesh Rampalli |
dineshr@roxonn.com |
https://www.linkedin.com/in/dinesh-rampalli-9b94b0169/ |
https://www.roxonn.com |
8886158111 |
Founder & CTO |
Dinesh and Ravi met through the T-Hub XDC Finternet Accelerator program in 2024. Both were exploring blockchain infrastructure opportunities and found complementary skills — Dinesh on technical/product, Ravi on business operations. Jagadish (Dinesh's brother) joined to lead community and growth.
Working together full-time since January 2025 (~15 months). Before going full-time, collaborated on Roxonn part-time from June 2024 while building the initial product and raising our first round. |
2 |
Execution speed with zero resources.
We shipped 3 live products on mainnet in under 12 months — from a tier-2 city in India with no startup ecosystem, no VC connections, no prior startup experience. Most teams with our background would still be planning. We launched.
Clear role separation helps: Dinesh owns product and tech, Ravi handles operations and compliance, Jagadish drives community and partnerships. No overlap, no ego battles. We move fast because everyone knows their lane.
Second strength: We're builders, not talkers. 500+ users, 25K community, 10K+ transactions — all organic, zero marketing spend. We let the product speak. |
Roxonn Future Tech Private Limited |
https://www.roxonn.com |
Hyderabad |
GitHub-integrated platform that pays developers instantly when their work is approved. |
Developers lose 20-40% to platform fees on Upwork/Fiverr and wait 2-4 weeks for payment. Their reputation is locked in silos — switch platforms, start from zero. For a developer in India earning ₹50K/month, losing 30% is rent money. |
Roxonn integrates with GitHub to pay developers instantly when work is approved. Fees are 2-5% vs 20-40% on traditional platforms. We offer three products: a developer marketplace with instant payments, an AI coding assistant (Roxonn Code), and a compute network where users earn from idle hardware (Roxonn Node). Developers build portable, verified work history that follows them everywhere. |
1. GitHub-native — Payment triggers when work is approved, zero new workflow for developers 2. 10x cheaper — 2-5% fees vs 20-40% on Upwork/Fiverr 3. Instant payment — Seconds, not weeks 4. Portable reputation — Verified work history follows developers across platforms 5. Three-sided network — Developers + businesses + compute providers create compounding lock-in 6. First-mover — Only platform with GitHub-to-payment automation live in production |
Revenue |
Revenue |
Primary: Freelance developers (age 22-35) in India currently using Upwork/Fiverr or between jobs, earning ₹30K-2.5L/month. They want to keep more earnings and get paid faster. Secondary: Early-stage startups and SMEs who need affordable software development without agency markups. Price-sensitive, open to new platforms if quality is guaranteed. |
$541 Billion — Global freelance and IT outsourcing market |
$50 Billion — Freelance developer payments in emerging markets (India, Philippines, Africa, LATAM) |
$50 Million — Realistic capture in 5 years with 100K active users at $500/year average transaction value |
1. Platform fees (2-5%) — Cut on every payment processed through the platform 2. Subscriptions (₹500-2000/month) — Premium features for developers and businesses 3. Compute fees — Margin on jobs processed through Roxonn Node 4. Enterprise licensing — White-label infrastructure for companies Target: ₹1 Crore ARR by Year 3 |
1. Upwork/Fiverr/Toptal — 20-40% fees, 2-4 week payments, locked reputation 2. Braintrust — Decentralized talent network but no GitHub integration or instant payments 3. Deel/Remote — Global payroll, not project-based developer payments Our edge: GitHub-native, 10x cheaper, instant payment, portable reputation. No one else has this combination live in production. |
1. Organic community — 25K members built through Discord, Telegram, LinkedIn, Twitter 2. GitHub integration — Frictionless onboarding for developers already using GitHub 3. Word-of-mouth — Developers share when they get paid instantly for the first time 4. Accelerator networks — T-Hub, RTIH partnerships bring warm introductions 5. Content marketing — Technical blogs, webinars, developer tutorials |
India-first, then global. Phase 1 (Now): Acquire developers organically through community and GitHub integration Phase 2 (6 months): Focus on demand-side — onboard Indian startups and SMEs posting projects Phase 3 (12 months): Expand to Philippines, Nigeria, LATAM Phase 4 (24 months): Enterprise features for larger companies Wedge: Developers keep 30% more money. Simple math drives adoption. |
Transform India from cheap labor provider to product-building nation.
Roxonn becomes the default payment infrastructure for software work globally. Every developer — whether in Bangalore or Nellore — earns fairly for their skills. Every business accesses quality talent without middlemen extracting value.
5-year goal: 500K active users, ₹50 Crore ARR, expanded to 20+ countries.
10-year goal: The way Stripe is for payments, Roxonn is for developer work — invisible infrastructure that powers millions of transactions. |
Private Limited Company — Roxonn Future Tech Private Limited (CIN: U62013TS2024PTC187100), incorporated June 2024, DPIIT recognized (DIPP230873) |
Bootstrapped with ₹50L+ in cloud credits from AWS, Google, Microsoft, NVIDIA startup programs. No equity funding raised to date. |
Yes |
Three reasons: 1. GTM support — We have 200+ users and 3 live products but need help acquiring businesses (demand-side). Avinya's network can open doors to Indian startups and SMEs who need developer talent. 2. Fundraising guidance — We're raising our first equity round. Mentorship on structuring the raise, investor positioning, and warm introductions would accelerate our timeline. 3. Scaling playbook — We built fast from a tier-2 city with no ecosystem. Now we need structured guidance on scaling from 200 to 50,000 users without breaking what works. We execute. We need partners who can help us scale smarter. |
1. Onboard 50+ businesses posting projects on the platform (solve demand-side bottleneck) 2. Secure 3-5 investor meetings with aligned VCs for our ₹1.5-2 Crore raise 3. Refine GTM strategy for Indian market — pricing, packaging, sales playbook 4. Grow from 200 to 2,000+ active users 5. Achieve first ₹1-2 Lakhs in monthly revenue Clear, measurable outcomes. We'll hold ourselves accountable. |
yes |
Yes |
Supporting feature |
Roxonn Code uses a lightweight API-based architecture:
1. LLM Integration — Third-party model APIs (Claude/GPT) for code suggestions and natural language processing
2. Context Layer — Extracts relevant code context from user's current file and project
3. Prompt Engineering — Custom prompts optimized for developer workflows (code review, debugging, generation)
4. Response Processing — Filters and formats AI outputs for clean developer experience
5. Feedback Loop — User interactions improve prompt quality over time
No custom model training currently. Focus is on smart prompt engineering and context management to maximize value from existing LLM APIs. |
1. Developer work history — Verified records of completed tasks, code quality, and project outcomes
2. Task-code mapping — Dataset linking project requirements to actual code submissions
3. Platform interaction data — How developers use Roxonn Code, which suggestions they accept/reject
4. Pricing signals — Real transaction data on what businesses pay for different types of work
This data compounds over time and cannot be replicated by competitors starting fresh. |
1. Workflow integration — Roxonn Code is embedded in GitHub workflow where developers already work. Competitors need to replicate this integration, not just the AI.
2. Context advantage — Our AI has access to task requirements, payment history, and developer reputation. Generic coding assistants don't have this business context.
3. Feedback loop — Every accepted/rejected suggestion improves our prompts. More users = better AI = more users.
4. Combined value — AI alone isn't defensible. AI + instant payments + portable reputation together is defensible. Competitors would need to build all three. |
1. Suggestion acceptance rate — % of AI suggestions developers accept
2. Task completion time — Does AI reduce time to complete tasks?
3. Code quality scores — Error rates in AI-assisted vs non-assisted submissions
4. User satisfaction — NPS and direct feedback on AI features
5. Latency — Response time for AI suggestions (target: <2 seconds)
6. Cost per query — API costs vs value delivered
Currently tracking manually. Building automated dashboards as we scale. |
1. Data minimization — Only collect data necessary for platform functionality
2. User consent — Clear terms on how code and data are used
3. No training on user code — We don't train custom models on user submissions
4. API security — All LLM API calls use secure connections, no code stored by third parties
5. Access controls — Role-based access to sensitive data
6. Encryption — Data encrypted at rest and in transit
7. Compliance roadmap — Planning for GDPR compliance as we expand to EU
Early stage but building security-first culture. |
1. AI regulations — Emerging AI governance laws (EU AI Act) may require transparency in AI-assisted work. Low risk for our use case (coding assistance, not high-stakes decisions).
2. Cross-border payments — Different countries have varying rules on digital payments. Mitigating by starting India-first with clear KYC.
3. Employment classification — Platform workers vs employees debate. We're a marketplace, not an employer, but monitoring regulatory developments.
Overall risk: Low to moderate. No high-risk AI applications (healthcare, finance decisions). Payment infrastructure is the bigger regulatory consideration. |
1. API costs — Current LLM API costs are manageable at 200 users. At 2,000+ users, costs could become significant. Mitigation: Negotiate volume pricing, explore open-source models (Gemma, Llama).
2. Latency — More concurrent users may slow response times. Mitigation: Caching common queries, load balancing, CDN for static assets.
3. Context management — Larger projects = more context to process. Mitigation: Smart context windowing, summarization techniques.
4. Prompt quality — Current prompts optimized for early users. May need refinement for diverse use cases at scale.
5. Support overhead — More AI users = more edge cases = more support tickets. Mitigation: Better documentation, in-app guidance. |
Yes — 1 dedicated AI engineer
Nikitha Samudrala (Gen AI Engineer)
- Specializes in LLM integration and prompt engineering
- Building and optimizing Roxonn Code
- Experience with OpenAI, Anthropic, and Google AI APIs
Founder (Dinesh Rampalli) also has applied AI experience from building the initial AI features.
Planning to expand AI team with funding — hire 1-2 ML engineers focused on model fine-tuning and evaluation. |
1. Platform interaction data — Owned. User actions, task completions, code submissions on our platform.
2. Public code datasets — Licensed/open. Standard coding examples for prompt development and testing.
3. LLM APIs — Licensed. Using commercial APIs (OpenAI/Anthropic/Google) under their standard terms.
4. No scraped or unauthorized data. All datasets either owned (platform data) or properly licensed. |
1. User feedback loop — Track which suggestions are accepted/rejected, use patterns to refine prompts
2. A/B testing — Test different prompt strategies, measure impact on acceptance rates
3. Model upgrades — Evaluate new LLM releases (GPT-5, Gemini 2.0, Claude 4) and migrate when beneficial
4. Domain fine-tuning — As we gather more data, explore fine-tuning open-source models for developer-specific tasks
5. Human review — Periodic manual review of AI outputs to catch quality issues
6. Benchmarking — Regular evaluation against coding benchmarks and competitor tools |
Both — India-first, then global.
Phase 1 (Now): India — 1.5M engineering grads/year, underserved by current platforms, highest pain from fees
Phase 2 (12 months): Southeast Asia — Philippines, Indonesia
Phase 3 (18 months): Africa — Nigeria, Kenya
Phase 4 (24 months): Global — US, Europe, LATAM
India is our wedge. Global is the vision. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/iVDexd5enGLb-Deck_Roxonn.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/iVDexd5enGLb-Deck_Roxonn.pdf |
https://www.youtube.com/watch?v=S7fuBCxjiYY |
Yes — and at Roxonn, the mission and the business model are the same thing.
A developer in Hyderabad doing the same quality work as someone in San Francisco earns a fraction of the pay, waits weeks for payment, and loses 20–40% to a platform that adds almost no real value. AI compute infrastructure is controlled by a handful of cloud giants, pricing out smaller developers and emerging market contributors entirely. These are systemic failures affecting hundreds of millions of people in the global tech workforce.
Roxonn fixes this at the infrastructure level. Smart contracts automatically verify GitHub contributions and release $ROXN payments instantly — no intermediary, no geography tax. Our DePIN compute layer lets anyone contribute spare computing power to AI infrastructure and earn real income from it. DAO governance ensures the people building and powering the platform have a voice in how it evolves.
I'm from Hyderabad. I've seen firsthand how much talent gets undervalued and excluded by systems never designed to include them. Roxonn is my answer to that — not as charity, but as infrastructure. Because the most durable impact is economically self-sustaining, where doing the right thing and building a great business point in exactly the same direction. |
NA |
NA |
checked |
| Apr 2, 2026 @ 6:22 PM |
Ramalingeswara Rao K V |
kvramalingeswararao@gmail.com |
https://www.linkedin.com/in/RamalingeswaraRaoKV/ |
http://www.RamalingeswaraRaoKV.blogspot.com |
9449501439 |
Founder |
Single Founder |
1 |
Bootstrapped full-time single Founder |
NIRAM |
www.LanKeys.blogspot.com |
Bengaluru |
NIRAM LanKeys (www.LanKeys.blogspot.com) is a patented Multi-purpose Natural Language Processing Tool. |
Cyber Security issues; Communication in Mother-tongue; Innovation in NLP, AI, TTS, Transliteration, etc.; etc. |
✳️ Information Technology (IT):
⭐️ NIRAM LanKeys (www.LanKeys.blogspot.com) is a patented Multi-purpose Natural Language Processing Tool.
I have participated in the Launchpad programme at NSRCEL in IIMB.
⭐️ Network (Pitch & Honor): https://LanKeys.blogspot.com/p/network.html
⭐️ Videos:
Demo: https://youtu.be/7TAhbNFFd48
Product Life Cycle: https://youtu.be/qRImjCgOdYo
⭐️ Number of Patents: 2 (Published-1 and Granted-1).
Patent #1 [Published]: METHOD AND SYSTEM FOR TRANSLITERATING TEXT FROM ONE LANGUAGE TO MULTIPLE LANGUAGE SCRIPTS.
Intellectual Property BHARAT/India 🇮🇳 Patent Application No. 931/CHE/2012.
World Intellectual Property Organization 🌐 Patent Application No. PCT/IN2013/000155.
Patent #2 [Granted]: SYSTEM AND METHOD FOR AUTOMATIC NAMING CONVENTION OF AUDIO FILES BASED ON USER PREFERENCE.
Intellectual Property BHARAT/India 🇮🇳 Patent No. 473442. Patent Application No. 2686/CHE/2014. |
Patents-2 |
MVP |
Testimonials |
B2G, B2B, B2C: Global (Digital) People |
-NA- |
-NA- |
-NA- |
Patent license; products & services. |
-NA- |
-NA- |
Network |
IPO |
In process |
0 |
No |
Network & Grant |
Grant |
In-person (Bengaluru) preferable |
Yes |
Automation layer |
Patent [Granted]: SYSTEM AND METHOD FOR AUTOMATIC NAMING CONVENTION OF AUDIO FILES BASED ON USER PREFERENCE.
Intellectual Property BHARAT/India 🇮🇳 Patent No. 473442. Patent Application No. 2686/CHE/2014. |
|
Patents |
|
|
|
|
Founder is an IT Expert. |
|
|
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/dyNnWVOY6vL4-NIRAM_LanKeys_PitchDeckBM_22Mar2026.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/dyNnWVOY6vL4-NIRAM_LanKeys_PitchDeckBM_22Mar2026.pdf |
https://youtu.be/7TAhbNFFd48 |
⭐️ Network (Pitch & Honor): https://LanKeys.blogspot.com/p/network.html
⭐️ Product Life Cycle: https://youtu.be/qRImjCgOdYo |
Physically Challenged Person |
Suman |
checked |
| Apr 2, 2026 @ 5:41 PM |
CH MADAN MOHAN SANTOSH |
madan.mohan.santosh@gmail.com |
|
https://www.busivold.com |
+919490265963 |
Founder/Tech |
N/A |
one |
New thinking |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 2, 2026 @ 5:13 PM |
Khyati Jain & Vanshika Modi |
jain.khyati@cycleofsamsara.com |
https://www.linkedin.com/in/khyati-jain1/ |
http://www.cycleofsamsara.com |
9874704678 |
Khyati Jain – CEO & Co-Founder Leads business, growth, and operations. Background in D2C brand building, marketing, and community-led growth. Has driven early traction for COS including user acquisition, seller onboarding, and partnerships. Vanshika Modi – CPTO & Co-Founder Leads product, design, and technology. Background in UI/UX, full-stack development, and customer journey design. Built COS’s end-to-end platform including onboarding, authentication flows, and escrow-based systems. |
We met in London during our post-graduation at the London College of Fashion, where we were flatmates, and quickly realized we had complementary strengths and a shared interest in building something of our own.
Living together, we often discussed gaps we observed in consumer behavior, especially around fashion and underutilized wardrobes. These conversations eventually led to the idea behind the Cycle of Samsara.
We’ve been working together since the inception of the company, building everything from scratch — from early validation and seller onboarding to product development and go-to-market. This shared journey has helped us build strong alignment, trust, and a highly collaborative working style. |
All |
Our biggest strength is the combination of strong execution and product thinking.
We balance business, growth, and operations with product and technology, allowing us to move quickly from idea to implementation. We are highly hands-on and willing to do unscalable work early — from onboarding sellers manually to testing offline channels — to gain real insights.
We also share a clear vision and long-term conviction, which helps us stay aligned while solving complex problems like trust, logistics, and behavior change in resale |
Cycle of Samsara |
www.cycleofsamsara.com |
Kolkata |
Cycle of Samsara is a trust-first resale platform for premium and luxury fashion in India. |
Reselling high-value fashion in India is fragmented, unreliable, and low-trust. Sellers struggle with pricing, logistics, and delayed payouts, while buyers hesitate due to concerns around authenticity and product quality. Most existing platforms focus on volume rather than verification, leading to poor experiences. As a result, premium and occasion wear remains underutilized, despite strong demand for affordable luxury and sustainable consumption. |
We are building a structured, trust-led resale platform for premium and luxury fashion.
Our platform integrates seller verification, listing-level approvals, authentication layers, and escrow-secured payments to ensure safe transactions. We also simplify selling through pricing guidance, logistics support, and curated listings.
By focusing on high-quality inventory and seamless user experience, we make resale reliable and accessible for both buyers and sellers. |
Our defensibility lies in building trust as infrastructure, not as a feature. We combine multiple layers — verification, authentication, escrow, and curation — to create a high-trust marketplace optimized for premium transactions. This improves conversion, retention, and supply quality. Additionally, our focus on high-AOV categories like luxury and occasion wear creates a niche where trust is critical and harder to replicate. Over time, our data on pricing, authentication, and user behavior, combined with AI-led tools, will further strengthen our moat. |
Users |
Users, Pilots, Signups |
Our target customers are urban consumers (Tier 1 and Tier 2 cities, ages 18–45) who own or aspire to own premium and luxury fashion. We focus on: • Sellers – Individuals with underutilized high-value items such as luxury, designer, and occasion wear (especially wedding fashion) • Buyers – Value-conscious consumers seeking authenticated premium fashion at accessible prices This segment values quality, trust, and convenience, and is increasingly open to circular consumption. |
TAM: ~$9–10B India’s total apparel resale and recommerce market (projected) |
SAM: ~$1.5–2B Premium, luxury, and occasion wear resale segment in urban India |
SOM: ~$15–30M GMV Near-term obtainable market (~1% of SAM), focused on high-AOV transactions |
We operate on a commission-based marketplace model. • 25% commission per transaction • Additional revenue streams: – Authentication services – Buyer protection fees – Cleaning and refurbishment services Our model is aligned with high-value transactions, improving margins through higher AOV. |
• Horizontal marketplaces: OLX, FreeUp • Peer-to-peer resale platforms • Emerging curated resale platforms in India Most competitors focus on volume and open listings, leading to low trust and inconsistent quality. |
• Instagram and community-led growth • Content and creator collaborations • Offline pop-ups for validation and trust-building • Organic referrals and word-of-mouth We focus on building a strong, trust-driven community rather than purely performance-led acquisition. |
Our GTM is phased: 1. Pre-launch: Build supply and community (seller onboarding, awareness) 2. Launch: Validate product and operations through controlled growth 3. Growth: Scale traffic and transactions via digital channels and partnerships 4. Scale: Establish COS as a premium resale destination We prioritize curated inventory, strong trust signals, and high-quality user experience to drive conversion and retention. |
To build India’s most trusted circular commerce ecosystem for fashion.
We aim to make resale a default behavior, not an alternative, by creating infrastructure for authentication, pricing, and seamless transactions.
Starting with premium fashion, we plan to expand into broader categories and enable brands and consumers to participate in a fully circular economy — where products are reused, resold, and retained in value over time. |
Yes |
Na |
Yes |
We are applying to Avinya to accelerate our transition from early validation to a scalable, structured business. At this stage, we have proven demand and built a working product, but we are now focused on strengthening our product, operations, and go-to-market strategy. Avinya’s mentorship, ecosystem, and hands-on support can help us refine our business model, improve unit economics, and scale efficiently. We are particularly interested in learning from experienced operators and building the right foundation to grow a trust-first marketplace in a complex category like resale. |
During the programme, we aim to: • Improve key product metrics such as sell-through rate, onboarding speed, and repeat transactions • Strengthen our unit economics, including CAC, margins, and retention • Build and validate scalable acquisition channels and partnerships • Refine our product roadmap, especially around authentication and pricing intelligence • Prepare for the next stage of growth and fundraising with a clear strategy and metrics Our goal is to move from early traction to a repeatable and scalable model |
Yes, we are open to both hybrid and virtual participation and can fully commit to the programme requirements. |
Yes |
Supporting feature |
We are building AI as a support layer for trust, pricing, and verification.
Current stack: Python-based backend with REST APIs, integrating third-party vision models for image validation and anomaly detection. We are experimenting with pretrained computer vision models (e.g., CNN/transformer-based models) for image authentication and classification.
Planned architecture includes:
• Image-based authentication pipeline
• Price prediction engine using historical resale data
• Seller risk scoring models
We are using cloud infrastructure (AWS/GCP) with scalable APIs and modular services. |
We are building proprietary datasets from:
• Verified listings (images, condition, brand, pricing)
• Transaction data (AOV, sell-through, buyer behavior)
• Seller quality and performance data
Over time, this structured dataset for premium resale in India becomes a strong moat, especially for pricing and authentication. |
Our defensibility comes from:
• Proprietary, high-quality labeled data from verified transactions
• Integration of AI into core workflows (not just as an add-on)
• Focus on premium, high-AOV inventory where accuracy matters more
• Feedback loops from real transactions improving model quality
This creates a data + workflow advantage that improves with scale. |
We use:
• Accuracy and precision (for classification/authentication)
• False positive / false negative rates
• Pricing prediction error (MAPE)
• Impact on business metrics (conversion rate, sell-through)
We also validate outputs through human review in early stages. |
• Secure cloud infrastructure with access controls
• Encryption for sensitive data
• Limited data access internally
• Compliance with applicable data protection standards
We prioritize user data protection and minimize unnecessary data storage. |
Current risks are limited but may include:
• Data protection and privacy regulations
• Compliance around digital transactions and escrow systems
We are building with compliance in mind to adapt as regulations evolve. |
• Manual verification and authentication processes
• Data labeling bottlenecks
• Infrastructure load on image processing systems
We are addressing this by automating workflows and building scalable pipelines. |
We have an in-house full-stack AI engineer working on backend systems and early AI integrations.
We plan to expand into a dedicated AI/ML team as we scale. |
• Internal data from listings, users, and transactions (owned)
• Public/pretrained datasets for model initialization
• Third-party APIs for image validation (licensed) |
• Continuous data collection from transactions
• Feedback loops from users and manual verification
• Regular retraining of models
• Iterative testing tied to business outcomes |
We are starting with India, focusing on premium resale in urban markets.
Our long-term vision is to expand globally, especially in markets with similar trust gaps in resale. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Sbuyr60eHeoX-COS-Pitch-Deck-_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Sbuyr60eHeoX-COS-Pitch-Deck-_compressed.pdf |
https://youtu.be/fcjHzPRToRQ |
Yes, Cycle of Samsara is a mission-driven startup focused on enabling circular fashion in India.
The fashion industry is one of the largest contributors to waste, especially in categories like occasion and luxury wear that are used only a few times. Our mission is to extend the lifecycle of these products by making resale trusted, accessible, and scalable.
By enabling users to buy and sell pre-owned fashion with confidence, we reduce waste, lower the need for new production, and promote more responsible consumption. At the same time, we create economic value by helping individuals monetize underutilized assets.
Our approach combines sustainability with strong business fundamentals — making circular fashion not just an ethical choice, but a practical and desirable one. |
Yes. As women founders building in tech and marketplaces, we are part of an underrepresented group in the startup ecosystem. |
Na |
checked |
| Apr 2, 2026 @ 5:04 PM |
Shivkumar Gurram |
gs@sourcingxpress.com |
https://www.linkedin.com/in/shivkumargurram/ |
https://www.sourcingxpress.com/ |
8050020304 |
Shivkumar Gurram is founder and has 15+ years of experience in tech recruitment with companies like Amazon, Twilio, IBM etc... Abhishek Das is co-founder and has been freelancing Software Projects since 10 years |
It started with a simple conversation.
While Shiv (Founder) was building the prototype of SourcingXPress, juggling between working with interns and running a recruitment business, I happened to speak with Abhishek Das.
He had applied for a role.
But that call didn’t feel like a typical screening.
Somewhere in the conversation, we both said the same thing:
“Recruitment is broken.”
From his side; as a job seeker.
From mine; as a recruiter.
Different experiences.
Same frustration.
That moment changed everything.
What started as a discussion turned into alignment.
And that alignment turned into action.
We decided to build something together.
Something that actually fixes hiring, not just for recruiters, but for candidates too.
That’s how SourcingXPress was born.
And now, it’s been more than a year of building, learning, and pushing boundaries together.
Where I (Shiv) speaks to the recruiters and candidates day-in-day-out and come up with product feature based on real user feedback and Abhishek (CTO + Co-founder) who leads the Architecture and teach team. |
2 |
Team it self is our strength. Founder has a core domain expertise and co-founder has core Software Engineering expertise. |
Get SourcingXPress Pvt LTD |
https://www.sourcingxpress.com |
Bangalore |
We are Amazon for Hiring |
SourcingXPress solves the most painful problem in modern hiring: too many irrelevant applications, too little accuracy, and expensive paywalls that block real hiring outcomes. |
Traditional job portals bombard employers with high traffic but low-quality matches. Recruiters waste hours screening unqualified profiles, while job seekers struggle for visibility. On top of that, platforms lock every essential feature, search, candidate contact, shortlists, behind heavy paywalls, forcing employers to spend more without guaranteeing better hires.
SourcingXPress removes both barriers:
The accuracy gap, A zero-shot matching algorithm ranks candidates purely on skill and relevance, giving recruiters instant, high-precision shortlists without manual effort.
The paywall problem , Employers can post unlimited jobs for free, browse applicants for free and pay only for what actually matters: verified, high-quality unlocks. There are no subscriptions, no hidden charges, and no forced upgrades.
This creates a radically fair hiring ecosystem. Recruiters save time, reduce costs, and get better shortlists. Job seekers gain equal visibility because the system rewards talent, not paid boosts.
In short, SourcingXPress solves the industry's two biggest issues, irrelevant applications and overpriced paywalls, by delivering fast, accurate, affordable hiring for everyone. |
SourcingXPress is unique because it flips the hiring model from paying for access to paying for confidence, where recruiters can see an explainable and evidence backed match score including skill-wise experience and context of usage before unlocking a candidate, eliminating wasted time and cost. Unlike keyword-based platforms, it uses structured skill intelligence and graph-based modeling aligned with real recruiter decision making rather than just resumes. This creates strong defensibility through a combination of deep domain expertise, proprietary matching logic, and behavioral design that improves decision velocity while reducing noise. As hiring shifts from access to intelligence, this approach positions SourcingXPress as a system that improves hiring outcomes, making it more efficient, sticky, and hard to replicate. |
Revenue |
Users, Revenue, Pilots, Signups, Testimonials |
Anyone hiring for Tech Roes which includes but not limited to Recruitment Freelancers, Corporate Recruiters, Recruitment Agencies, Startups, Enterprise as we are not just operating in B2B we are also operating in Business to Recruiter (B2R) industries first model in hiring. |
NA |
NA |
NA |
Platform Subscription, API access and Managed Career Site |
Naukri, Instahyre, CutShort |
Already acquired 550+ employers/recruiters through word of mouth |
Currently it's only founders network as he runs a community of 10000+ Recruiters in India and provide corporate trainings. |
To build an Hiring Eco System where anyone who is laid-off or on a long career break and wanted to make a come back they can see SourcingXPress as a resilient platform. |
Private Limited - Registered 1 year ago |
550000 |
Yes |
Funding |
Funding |
Anything is fine as long as it's in Bangalore |
Yes |
Core engine |
I will let my CTO explain once we are shortlisted for pitching |
Already 550+ Employers and 1 Lakh+ Active tech job seekers on the platform and an average of 5-10 recruiter/companies and 3000-5000 new users joining every single week. |
I will let my CTO explain once we are shortlisted for pitching |
NA |
Currently in process of ISO20007, SOC 2 (Type 2), DPDPA |
Yes, and we are working on it. |
I will let my CTO explain once we are shortlisted for pitching |
Yes, I will let my CTO explain once we are shortlisted for pitching |
We are hosted on Azure and additionally - I will let my CTO explain once we are shortlisted for pitching |
I will let my CTO explain once we are shortlisted for pitching |
We have started with India but eventually we plan to scale global market and already have some ready to pay customers from USA. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/HAv6QXEEzeBv-SISF_sourcingxpress-pitch-deck-dec-2025.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/HAv6QXEEzeBv-SISF_sourcingxpress-pitch-deck-dec-2025.pdf |
https://www.youtube.com/watch?v=HuHC6_u-5n4%20%20%20%20%20%20also%20check%20%20%20%20%20https://www.youtube.com/watch?v=J7df4iP8qVo&t=13s |
Both. |
NA |
NA |
checked |
| Apr 2, 2026 @ 4:14 PM |
Anil Dhupadal |
anildhupadal@naksysglobal.com |
http://www.linkedin.com/in/anil-dhupadal |
http://www.naksysglobal.com |
7022530079 |
Technology professional with experience in application development, system architecture, and platform engineering. Focused on building scalable, integrated solutions and leading end-to-end product development. |
We have known each other personally and have built a strong working relationship over time. Combining our complementary skill sets in SAP consulting and technology development, we have collaborated on understanding enterprise challenges and identifying gaps in existing systems. While this is our first venture together, our aligned vision, trust, and shared experience in solving real-world business problems form a strong foundation for building and scaling this startup. |
All |
Our biggest strength is the combination of deep domain expertise and strong technical capability. With one founder specializing in SAP and enterprise business processes and the other in technology and product development, we bring a balanced approach to solving real-world problems. This allows us to design solutions that are not only technically robust but also aligned with actual business needs. Our shared vision, hands-on experience, and ability to execute quickly enable us to build a practical, scalable, and enterprise-ready platform. |
NAKSYS GLOBAL TECHNOLOGIES INDIA PVT LTD |
www.naksysglobal.com |
Bengaluru |
A unified, enterprise-grade booking platform that centralizes scheduling, resource management, and SAP-integrated operations across services and locations. |
Businesses struggle with fragmented booking systems, manual scheduling, and lack of integration across services and locations, leading to inefficiencies, errors, and poor visibility. Most existing solutions do not connect with core enterprise systems like SAP, creating data silos and operational delays. This results in suboptimal resource utilization and a poor customer experience, highlighting the need for a unified, integrated booking platform. |
Our solution is a Unified Booking Platform that centralizes booking, scheduling, and resource management across multiple services, locations, and channels within a single system. It provides real-time availability, automated workflows, and seamless coordination between users and resources.
The platform is designed with an API-first and integration-driven approach, enabling smooth connectivity with enterprise systems such as SAP for synchronized data across operations like inventory, billing, and procurement.
With features like dynamic scheduling, automated notifications, role-based access, and real-time analytics, the solution improves efficiency, reduces manual effort, and enhances customer experience. Built as a scalable, cloud-ready platform, it supports complex enterprise use cases while remaining flexible for different industries. |
Deep Enterprise & SAP Integration Native integration with SAP and core enterprise systems ensures bookings are directly linked to business processes like inventory, procurement, and billing—creating a strong technical moat. True Unified Architecture Built as a single platform (not a combination of tools), enabling real-time synchronization, data consistency, and seamless operations across all booking channels. Designed for Complex Enterprise Use Cases Supports multi-location, multi-service, and resource-based scheduling with configurable business rules, making it difficult for simpler tools to replicate. API-First & Scalable Design Flexible architecture that allows easy integration with third-party platforms and enterprise ecosystems, ensuring long-term adaptability. Domain Expertise Advantage Founders’ strong background in SAP and enterprise processes ensures the product is grounded in real-world operational needs, not just generic booking functionality. This combination of integration depth, architecture, and domain knowledge creates a defensible and scalable solution. |
Idea |
Pilots |
Our target customers are mid to large enterprises and service-based organizations that require efficient, centralized management of bookings, scheduling, and resources. Key customer segments include: Healthcare Providers – Hospitals, clinics, diagnostic centers Travel & Hospitality – Hotels, resorts, tour operators Logistics & Transportation – Fleet operators, cargo, and delivery services Professional Services – Consulting firms, training institutes, and service providers Enterprise Organizations – Companies managing internal resource scheduling (rooms, assets, workforce) The platform is best suited for organizations with multi-location operations, high booking volumes, and complex scheduling requirements. |
$250B |
$45B |
$250M |
Our revenue model is primarily based on a SaaS (Software-as-a-Service) subscription model, combined with enterprise services: Subscription Fees: Monthly or annual pricing based on number of users, locations, or bookings Enterprise Licensing: Custom pricing for large organizations requiring advanced features and integrations (e.g., SAP) Implementation & Integration Services: One-time fees for onboarding, customization, and system integration Support & Maintenance: Ongoing support packages and premium service plans Add-ons & API Usage: Charges for advanced analytics, additional modules, and third-party integrations This hybrid model ensures recurring revenue while also capturing value from enterprise deployments and customization needs. |
Our competitors include both global booking platforms and niche scheduling solutions: Generic Booking Platforms Tools like Calendly and Acuity Scheduling that focus on simple appointment scheduling but lack enterprise-level capabilities and deep integrations Industry-Specific Solutions Platforms such as Bookeo (tourism/services) and Mindbody that cater to specific sectors but are not designed for cross-industry unified use Enterprise Scheduling Tools Solutions within ecosystems like Salesforce or Microsoft Dynamics 365 that offer scheduling features but are not purpose-built as unified booking platforms Custom In-House Systems Many enterprises rely on internally built or fragmented systems, which lack scalability, standardization, and real-time integration |
We adopt a focused B2B go-to-market approach to acquire and scale customers: Direct Sales & Enterprise Outreach Targeted outreach to enterprises in healthcare, logistics, and hospitality through networking, referrals, and industry connections Pilot Programs & POCs Offering proof-of-concepts to demonstrate value, build trust, and convert early adopters into paying customers Partnerships & System Integrators Collaborating with IT consulting firms and SAP partners to access existing enterprise client bases Digital Marketing & Thought Leadership Leveraging LinkedIn, content marketing, and case studies to build credibility and attract inbound leads Industry Events & Networking Participating in startup programs, tech events, and business forums to increase visibility Customer Referrals Encouraging early customers to refer new clients, creating a network-driven growth model This multi-channel approach helps us build credibility, validate the product, and scale customer acquisition efficiently. DEVELOPER MODE |
We adopt a focused B2B go-to-market approach to acquire and scale customers: Direct Sales & Enterprise Outreach Targeted outreach to enterprises in healthcare, logistics, and hospitality through networking, referrals, and industry connections Pilot Programs & POCs Offering proof-of-concepts to demonstrate value, build trust, and convert early adopters into paying customers Partnerships & System Integrators Collaborating with IT consulting firms and SAP partners to access existing enterprise client bases Digital Marketing & Thought Leadership Leveraging LinkedIn, content marketing, and case studies to build credibility and attract inbound leads Industry Events & Networking Participating in startup programs, tech events, and business forums to increase visibility Customer Referrals Encouraging early customers to refer new clients, creating a network-driven growth model This multi-channel approach helps us build credibility, validate the product, and scale customer acquisition efficiently. |
Our long-term vision is to become a global, enterprise-grade Unified Booking Platform that serves as the central system for managing bookings, resources, and scheduling across industries.
We aim to evolve into a fully integrated operational layer that connects booking workflows with core enterprise systems like SAP, enabling seamless, data-driven decision-making. Over time, we plan to incorporate advanced capabilities such as AI-driven demand forecasting, intelligent scheduling, and automation.
Our goal is to empower organizations worldwide to replace fragmented tools with a single, scalable platform that improves efficiency, enhances customer experience, and drives digital transformation across complex business environments. |
ACTIVE |
NA |
Yes |
We are applying to Avinya to accelerate our journey from product development to market traction. As an early-stage startup, we are looking for structured mentorship, ecosystem access, and strategic guidance to refine our product, validate our business model, and scale effectively. Avinya’s focus on innovation and its strong network of industry experts, partners, and investors align well with our needs. We aim to leverage the program to secure pilot opportunities, build partnerships, and strengthen our go-to-market strategy. Additionally, access to funding support and expert feedback will help us accelerate product development and achieve faster market adoption. |
During the programme, we aim to achieve the following key milestones: Product Validation & Enhancement Refine our Unified Booking Platform based on mentor feedback and real customer use cases Pilot Customers & Early Traction Secure pilot projects/POCs and convert early adopters into paying customers Go-to-Market Strategy Strengthen our positioning, pricing, and sales approach for targeted industries Partnership Development Build collaborations with enterprises, system integrators, and ecosystem partners Funding Readiness Prepare for fundraising through improved pitch, metrics, and investor connections Operational Scaling Establish a strong foundation for team expansion, processes, and product scalability Overall, our goal is to transition from early-stage development to validated product-market fit and initial revenue generation. DEVELOPER MODE |
Yes, we are open to both hybrid and virtual participation. We are flexible and can actively engage in remote sessions while also attending in-person interactions when required, ensuring full commitment to the programme and its activities. |
Yes |
Automation layer |
Our AI architecture is designed as a modular, scalable, and integration-first system that enhances booking intelligence, automation, and decision-making within the Unified Booking Platform.
Data Layer
Aggregates structured and real-time data from bookings, user interactions, and integrated enterprise systems (e.g., SAP), ensuring a unified data model
Processing & Integration Layer
Uses APIs and event-driven pipelines to process booking events, synchronize data, and enable seamless interoperability across systems
AI/ML Layer
Incorporates machine learning models for:
Demand Forecasting (predicting booking trends)
Intelligent Scheduling (optimal resource allocation)
Anomaly Detection (identifying conflicts or unusual patterns)
Recommendation Engine (suggesting time slots/services)
Application Layer
Exposes AI-driven features through dashboards, APIs, and user interfaces for both customers and administrators
Feedback & Learning Loop
Continuously improves models using historical data, user behavior, and system outcomes
The architecture is cloud-native, API-first, and scalable, enabling real-time insights, automation, and seamless integration with enterprise ecosystems. |
While we are in the early stage, our proprietary advantage will be built through enterprise-integrated booking data and operational insights:
Deep Enterprise Data Integration
By integrating with systems like SAP, we capture not just booking data but also linked operational data (inventory, procurement, billing), creating a richer and more contextual dataset
Cross-Industry Booking Intelligence
Aggregating data across industries (healthcare, logistics, hospitality) enables us to identify patterns and optimize scheduling in ways single-industry tools cannot
Behavioral & Usage Data
Continuous collection of user behavior, booking trends, and resource utilization helps improve recommendations and forecasting models
Workflow & Process Data
Insights into how enterprises manage approvals, scheduling rules, and exceptions create a unique dataset for automation and optimization
Over time, this combination of transactional, operational, and behavioral data will form a strong, defensible data moat that improves our AI capabilities and platform value. |
Our AI component is defensible due to a combination of data depth, domain integration, and continuous learning capabilities:
Enterprise-Linked Data Advantage
Unlike generic booking tools, our AI is trained on data tightly integrated with enterprise systems (e.g., SAP), combining booking, operational, and transactional data—creating a richer and harder-to-replicate dataset
Domain-Specific Intelligence
Our models are designed around real-world enterprise workflows (procurement, inventory, resource allocation), making them more context-aware and relevant than generic scheduling algorithms
Closed Feedback Loop
Continuous learning from user behavior, booking patterns, and outcomes improves accuracy over time, strengthening performance with scale
Customization & Configurability
Ability to adapt AI models to specific business rules and industry needs creates high switching costs for customers
Integration-Driven Moat
Deep integration into enterprise ecosystems makes our AI an embedded part of operations, not just an add-on feature
This combination creates a data-driven and integration-led moat that becomes stronger as adoption grows. |
We evaluate our AI models using a combination of accuracy, efficiency, and business impact metrics:
Forecasting Accuracy
Metrics like MAE (Mean Absolute Error) and RMSE (Root Mean Square Error) to measure demand prediction accuracy
Scheduling Optimization Efficiency
Resource utilization rate, booking success rate, and reduction in scheduling conflicts
Recommendation Quality
Click-through rate (CTR), acceptance rate of suggested slots/services, and user engagement metrics
System Reliability
Uptime, API response time, latency, and error rates for real-time booking operations
Anomaly Detection Performance
Precision, recall, and false positive/negative rates for identifying conflicts or unusual patterns
Business Impact Metrics
Increase in booking conversion rates, reduction in no-shows, and improvement in operational efficiency
Continuous Monitoring & Feedback
Model drift detection, retraining frequency, and user feedback loops to maintain long-term reliability
This multi-layered evaluation ensures our AI remains accurate, scalable, and aligned with real business outcomes. |
We follow a security-by-design and privacy-first approach to ensure data protection and regulatory compliance:
Data Protection & Encryption
End-to-end encryption using HTTPS/TLS for data in transit and strong encryption standards (e.g., AES-256) for data at rest
Access Control & Authentication
Role-based access control (RBAC), multi-factor authentication (MFA), and strict identity management to prevent unauthorized access
Compliance Readiness
Designed to align with global and regional regulations such as GDPR and applicable Indian data protection laws, with clear data handling and consent mechanisms
Secure Architecture
Cloud-native infrastructure with network isolation, secure APIs, and regular vulnerability assessments
Monitoring & Incident Response
Continuous monitoring, audit logs, and defined incident response processes to detect and mitigate threats
Data Minimization & Anonymization
Collect only necessary data and apply anonymization/pseudonymization where possible for analytics and AI models
Regular Audits & Updates
Periodic security audits, penetration testing, and timely patching to maintain system integrity
This ensures our platform is secure, compliant, and trustworthy for enterprise use. |
While our platform is broadly applicable, there are certain regulatory considerations that could impact operations:
Data Protection & Privacy Laws
Compliance with regulations such as India’s Digital Personal Data Protection Act (DPDP) and global standards like GDPR, especially when handling user and enterprise data
Industry-Specific Regulations
Sectors like healthcare and transportation may have additional compliance requirements related to data handling, consent, and operational processes
Data Localization Requirements
Potential mandates to store and process data within specific geographic regions
Digital Payments Compliance
Adherence to RBI guidelines and payment security standards when integrating with payment systems
Cross-Border Data Transfers
Restrictions when expanding internationally and handling data across jurisdictions
We mitigate these risks through a compliance-first architecture, flexible data handling policies, and ongoing monitoring of regulatory changes. |
As we scale, certain components of our AI system may face stress if not proactively optimized:
Data Pipeline Bottlenecks
Increased volume and velocity of booking events can strain ingestion, processing, and real-time synchronization pipelines
Model Latency & Inference Speed
Real-time recommendations and scheduling decisions may experience higher latency under heavy concurrent loads
Model Drift & Data Quality Issues
Rapidly changing usage patterns across new customers and industries can reduce model accuracy without frequent retraining
Infrastructure & Compute Costs
Scaling AI workloads (training + inference) can significantly increase cloud and processing costs
Multi-Tenant Complexity
Supporting diverse enterprise workflows may require more customized models, increasing system complexity
Monitoring & Reliability Challenges
Detecting anomalies, failures, or performance degradation becomes harder at scale without robust observability. |
Currently, we do not have a dedicated in-house AI/ML team as we are in the early development stage. The core team is driving initial AI capabilities with a strong foundation in enterprise systems, data handling, and integration.
We plan to build a specialized AI/ML team as we scale, focusing on expertise in:
Machine learning and predictive analytics
Time-series forecasting and optimization algorithms
Data engineering and real-time processing
AI model deployment and MLOps
In the interim, we are designing the platform with an AI-ready architecture, ensuring seamless integration of advanced AI capabilities as the team expands. |
Currently, we are in the early development stage and are primarily using:
Synthetic & Test Data
Generated datasets to simulate booking scenarios, scheduling patterns, and resource allocation use cases
Sample / Open Datasets
Publicly available datasets (where applicable) for initial model experimentation and validation
Internal Structured Data Models
Designed schemas based on real-world enterprise workflows (e.g., booking, inventory, scheduling logic)
At this stage, we are not using any proprietary customer data. As we onboard customers, we will utilize client-owned data with proper consent and agreements, ensuring compliance with data protection regulations.
Our long-term strategy is to build proprietary datasets from platform usage, creating a strong data advantage while maintaining strict privacy and security standards. |
We follow a structured MLOps-driven approach to continuously improve model performance:
Continuous Data Collection
Capture real-time booking, usage, and outcome data to enrich training datasets
Feedback Loops
Incorporate user actions (accept/reject recommendations, rescheduling patterns) to refine models
Regular Retraining
Periodic and event-driven retraining to address model drift and adapt to changing patterns
A/B Testing & Experimentation
Test multiple model versions to evaluate improvements in accuracy, efficiency, and user engagement
Model Monitoring & Drift Detection
Track performance metrics (accuracy, latency, error rates) and detect deviations in real time
Feature Engineering Enhancements
Continuously improve input features based on new data and domain insights
Scalable MLOps Pipeline
Automated pipelines for data processing, training, validation, and deployment
This ensures our AI models remain accurate, adaptive, and aligned with real-world usage as we scale. |
We are building for both India and global markets, with a phased approach.
India (Primary Focus)
We are initially targeting the Indian market to validate our product, onboard early customers, and refine our solution based on real-world use cases.
Global Expansion (Secondary Phase)
Once validated, we plan to expand into international markets such as Southeast Asia and the Middle East, where there is strong demand for scalable, enterprise-grade booking solutions.
Our platform is designed to be globally scalable, with flexibility to adapt to different industries, regulations, and operational requirements. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Q8Zz1kIZjaKM-Pitch-Deck-2-Business-Model.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Q8Zz1kIZjaKM-Pitch-Deck-2-Business-Model.pdf |
http://NA |
Yes, we are building a mission-driven startup focused on improving operational efficiency and digital accessibility for businesses.
Our mission is to eliminate fragmented booking systems and manual processes by providing a unified, intelligent platform that simplifies scheduling, resource management, and service delivery. By doing so, we aim to help organizations reduce inefficiencies, improve resource utilization, and deliver better customer experiences.
We also focus on enabling digital transformation for growing businesses, especially in emerging markets, by providing scalable and accessible technology solutions. In the long term, our platform can support more transparent, efficient, and data-driven operations across industries, contributing to overall economic productivity and innovation. |
No, we do not identify as part of an underrepresented group. However, we strongly believe in building an inclusive organization and are committed to promoting diversity in our team as we grow |
Discovered via LinkedIn |
checked |
| Apr 2, 2026 @ 3:31 PM |
Manish Shetty |
manish6shetty@gmail.com |
https://www.linkedin.com/in/manish-shetty-519b852b0 |
https://myyuno.com/ |
8792344909 |
Manish - Tech Lead, Rajesh - Lead Developer, Smaran - Marketing & Content Lead |
We’re basically classmates, and after meeting we got to know we’re facing similar issues individually, that’s where the idea was born. We have been working together from August of 2025. |
All |
We keep pushing ourselves in most ways we don’t think of everyday.
We know that balance is a myth.
Our only goal is to achieve what we want to build, and that’s the only priority we have. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Apr 1, 2026 @ 10:28 PM |
Ajith Sharma and Geerthivasan Mahendran |
ajith0016@gmail.com |
http://www.linkedin.com/in/ajith-sharma-95368a90 |
https://vasool-app.web.app/ |
9972654336 |
Ajith Sharma-CEO- CS and CA Finalist with over a decade of experience in Compliance, taxation, and governance. Experience in enterprise software sales and proven track record in scaling financial services and SaaS products. Passionate about using AI to solve real business cash flow challenges. Geerthivasan, CTO- Built and scaled high-performance engineering teams to deliver complex AI solutions. Drove $10M+ in new annual revenue and cut operational costs by 40% with Nissan. Launched analytics products generating $12M in annual value while cutting cloud costs by 30%. |
We met on LinkedIn in June 2024 and quickly bonded over a shared interest in building something meaningful in the compliance and business services space. We initially started working on a techno-legal product around incorporation and regulatory workflows.
As we spent more time speaking with founders and operators, we kept encountering a recurring theme—while compliance was important, cash flow was what truly kept them up at night. That insight pushed us to rethink our approach.
We then evolved into building an AI-driven CFO platform, but even within that, one problem consistently stood out: receivables and delayed payments.
That’s what led us to double down and build Vasool AI.
Over the past 9–10 months, we’ve worked closely through multiple iterations and pivots for Vasool AI, which has helped us build strong conviction, clarity, and execution alignment as a team. |
2 |
I (Ajith) bring strong domain expertise in finance, compliance, and SME operations, along with hands-on experience in sales and customer relationships. This helps us stay deeply grounded in real customer problems and drive go-to-market effectively.
Geerthi brings an AI-first technical mindset, with the ability to architect and build scalable, intelligent systems from the ground up.
This combination allows us to bridge both sides—deep problem understanding and strong technical execution. We’re able to not just identify high-value problems like receivables, but also rapidly build and iterate on solutions like Vasool AI with speed and clarity |
Vasool AI |
https://vasool-app.web.app/ |
Bengaluru |
Vasool AI is an AI-powered receivables platform that helps businesses collect payments faster with intelligent follow-ups and automation |
Businesses struggle to collect their receivables on time, leading to cash flow issues despite having strong revenues. Today, collections are highly manual—tracked on Excel, followed up over WhatsApp, and dependent on inconsistent human effort. There is no prioritization of which customers to follow up with, when, or how. As a result, businesses face delayed payments, poor visibility on outstanding dues, and significant time spent chasing customers instead of focusing on growth. Vasool AI solves this by bringing intelligence, automation, and structure to the entire receivables and collections process |
Vasool AI is an AI-powered receivables intelligence platform that automates and optimizes the entire collections process.
It analyzes outstanding invoices, predicts which customers are likely to delay payments, and prioritizes follow-ups accordingly. The platform then executes personalized, multi-channel communication—across WhatsApp, email, and voice—reducing manual effort and improving recovery rates.
It also provides real-time visibility into receivables, helping businesses track cash flow and take proactive actions.
We start with soft collections and are building towards a full-stack collections platform, expanding into hard collections, recovery workflows, and eventually credit intelligence |
Our defensibility comes from combining deep domain expertise with an AI-first approach to a highly underserved problem. Unlike generic CRMs or reminder tools, Vasool AI is built specifically for receivables and collections. We don’t just automate follow-ups—we bring intelligence into the process by prioritizing debtors, personalizing communication, and continuously learning from payment behavior. Over time, this creates a proprietary data layer on payer behavior, response patterns, and recovery outcomes, which strengthens our prediction models and workflows. Additionally, our integration into core systems (like accounting tools, communication channels, and payments) makes us part of the daily workflow, increasing stickiness. As we expand into hard collections and credit intelligence, this data and workflow ownership become even more defensible. |
MVP |
Pilots |
SMEs and mid-sized businesses with high receivables volume—especially in sectors like trading, manufacturing, logistics, and services—where delayed payments are frequent and collections are still manual. |
$20B+ global market across receivables management, collections, and credit infrastructure. |
$3–5B across India and similar emerging markets with ~60M SMEs facing cash flow and collections challenges. |
$100M+ by targeting digitally active SMEs in India over the next 5–7 years. |
SaaS subscription based on number of invoices/customers managed, with additional revenue from payment integrations and future monetization through collections and credit products. |
Traditional ERPs (Tally, Zoho), reminder tools, and global players like HighRadius and Upflow. However, most are either too generic, enterprise-focused, or lack AI-driven automation for SMEs. |
Founder-led sales, partnerships with CAs/accountants, and inbound interest through LinkedIn and SME networks. Early traction is driven by direct outreach and demos. |
Start with India-focused SMEs via direct sales and CA/accountant partnerships, expand through integrations with accounting platforms, and scale via product-led growth and referrals. |
To become the default financial operations layer for SMEs—starting with receivables, expanding into full-stack collections, and eventually building a credit intelligence and underwriting platform. |
Incorporated on 10.01.2025 |
Not raised any funding |
Yes |
We’re applying to Avinya to accelerate our journey from early validation to a scalable, category-defining product. Given our focus on solving a deep-rooted SME problem in India, we believe Avinya’s ecosystem, mentorship, and network can help us sharpen our product, GTM, and fundraising strategy. We’re particularly looking to learn from operators and investors who understand building for Indian SMEs and scaling AI-first products |
Our primary goal is to achieve strong product-market fit by onboarding and retaining a core set of paying customers. We want to refine our GTM strategy, build a repeatable sales motion, and strengthen our product based on real user feedback. Additionally, we aim to position ourselves for our first institutional raise with a clear narrative, traction, and roadmap. |
Yes, we are fully open to hybrid or virtual participation and can actively engage in the programme in whichever format works best. |
Yes |
Core engine |
Models: We use an ensemble approach featuring Google Gemini 1.5 Pro for complex multi-hop reasoning (cashflow forecasting and risk modeling) and Gemini 1.5 Flash for high-velocity user interactions and real-time "Neural Analysis."
The Stack: Our frontend is built on React.js/Vite for a responsive "Desktop-Class" experience. The backend logic is powered by a high-concurrency FastAPI layer (Python) optimized for asynchronous ERP synchronization.
Persistence & Sync: We use Firebase/Firestore for real-time data streaming and Zod for strict schema validation, ensuring 99.9% data integrity across sensitive financial records.
APIs: Deep native integrations with Zoho Books, QuickBooks, and Xero via OAuth 2.0, creating a seamless "Data Gravity" effect. |
Our advantage lies in the Relationship Context Layer. Beyond raw invoice data, we possess:
Interaction Velocity Data: Real-time tracking of communication "touchpoints" (Dialer, WhatsApp, Email) and the resulting payment behaviors.
Sentiment & Negotiation Signatures: Proprietary datasets of successful vs. failed debt collection dialogues, allowing our AI to predict "Payment Intent" with high accuracy.
MSME Network Effects: As our platform grows, we build a cross-vendor reliability score for debtors, identifying "chronic late payers" before they impact a new user’s cashflow |
Vertical Fine-Tuning: Generic LLMs are "polite" but ineffective in collections. Our models are specialized for the Psychology of Debt Recovery, balancing persistence with professional relationship management.
Neural Debugging: We have built an automated "AI Verdict" system that identifies and fixes data inconsistencies between different ERPs and bank statements—a task that typically requires hours of manual accounting.
In-Workflow Agency: We aren't just a dashboard; we are an Active Agent. Vasool autonomously executes the entire collections lifecycle, from identifying a delay to simulating negotiations and reconciling the payment. |
We track performance via a specialized "Collections ROI" dashboard:
DSO (Days Sales Outstanding): The primary measure of how much we reduce a business's payment cycle.
Neural Latency: Maintaining a response time of <500ms for AI-driven insights to ensure a seamless "Copilot" experience.
Entity Extraction F1-Score: Measuring the accuracy of the AI in pulling structured data from unstructured interaction logs.
PTP Reliability Rate: The percentage of AI-extracted "Promises to Pay" that result in actual bank credit. |
Compliance: We adhere to the RBI’s Digital Lending Guidelines and are preparing for DPDP Act (India) and GDPR (Global) alignment.
Data Minimization: We only store what is necessary for the AI to function; sensitive PII is encrypted at rest and in transit (AES-256).
Security: Regular automated vulnerability scanning and a strict Role-Based Access Control (RBAC) framework ensuring data siloization between different organizations |
Primary risks involve the evolving regulatory framework for digital debt collection in India. We mitigate this by:
Hard-Coded Compliance Rails: Our AI is incapable of violating "Fair Practices Codes" (e.g., calling at inappropriate hours or using non-professional language) because the logic is governed by strict deterministic rulesets. |
Our architectural roadmap is designed to transform high-growth pressures into a competitive advantage. Rather than patching existing systems, we are evolving our infrastructure to ensure performance, cost-efficiency, and intelligence remain uncompromised at 10x and 100x scale.
1. High-Velocity Event Orchestration
To manage the surge in data-processing requirements, we are transitioning from traditional API polling to a fully WebHook-driven, Serverless architecture. By implementing a message-queue system (Google Pub/Sub), we decouple data ingestion from AI processing. This ensures zero-latency intake, allowing the system to absorb massive batch imports without impacting the user experience.
2. Petabyte-Scale Data Intelligence
As log volume increases, we are evolving our storage strategy from standard Firestore writes to a Data BigLake Architecture via BigQuery. This move serves two critical functions:
Operational Efficiency: Maintains high-performance "Neural Analysis" ingestion without the exponential costs of traditional databases.
Real-Time Optimization: Allows for "cold data" to be used for real-time AI training and trend analysis without putting any load on the production environment.
3. The Federated AI Mesh
To maintain superior unit economics while increasing throughput, we are moving away from a monolithic LLM approach. Our Federated AI Mesh uses a tiered intelligence strategy:
Gemini Flash: Deployed as high-speed, specialized agents for routine triage and rapid response.
Gemini Pro: Reserved for high-stakes negotiations and complex reasoning. This distributed mesh ensures that the most sophisticated "brain power" is used exactly where it adds the most value, keeping the system lean, fast, and infinitely scalable. |
Yes. Our core founding team has deep expertise in NLP (Natural Language Processing) and Financial Systems Engineering. We specialize in "Agentic RAG" (Retrieval-Augmented Generation) and low-latency LLM orchestration, specifically tailored for the complexities of the Indian business vernacular and fintech regulations. |
Owned: Anonymized interaction logs and payment outcomes from our user base.
Licensed/Public: We leverage macroeconomic indicators and banking sector data for broader risk benchmarking. |
RLHF (Reinforcement Learning from Human Feedback): We allow Admins to "Vote" on the AI’s negotiation verdict, creating a closed-loop system where the model learns from the most successful human collection strategies.
Shadow Deployment: We run updated model versions in "Shadow Mode" to compare their predictions against live results before promoting them to production. |
India-First, Global-Ready. We are solving a critical pain point for the 63 million MSMEs in India—the "Payment Delay" culture. However, the underlying technology (Automated Collections, AI CFO, Cashflow Forecasting) is built on a modular architecture that is ready for deployment in other credit-intensive global markets like Southeast Asia and the US. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/7r7h2T7454Ld-The-Autonomous-Receivables-Assistant-for-Indian-MSMEs-Ginserv-Demo-Day.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/7r7h2T7454Ld-The-Autonomous-Receivables-Assistant-for-Indian-MSMEs-Ginserv-Demo-Day.pdf |
https://vasool-app.web.app/ |
Yes—we’re building a mission-driven startup focused on improving the financial health of SMEs.
Across India and similar markets, millions of small businesses struggle not because of lack of demand, but because of delayed payments and poor cash flow visibility. This directly impacts their ability to pay salaries, reinvest, and grow.
Our mission with Vasool AI is to bring structure, intelligence, and predictability to receivables—helping SMEs get paid on time and operate with confidence.
By solving collections at scale, we aim to unlock working capital for millions of businesses, reduce financial stress for founders, and ultimately contribute to a more resilient and efficient SME ecosystem. |
NA |
Ginserv |
checked |
| Apr 1, 2026 @ 6:26 PM |
JIYAULHAQ |
jiaulhaq1786@gmail.com |
https://www.linkedin.com/in/jiyaul-haq-a12771290?utm_source=share&utm_campaign=share_via&utm_content=profile&utm_medium=android_app,,, |
https://%20jiyaxo.co.in |
09667772386 |
Founder & director |
Since 10 years , |
All |
We are the right team for this venture because we combine deep technical expertise with strong ground-level understanding of agriculture. The founder brings hands-on experience in mechanical design, battery systems, prototyping, and patent development, enabling us to build innovative and practical solutions. Our team includes members with 20–30 years of experience in rural operations, product development, and manufacturing, ensuring real-world feasibility and scalability.
Additionally, we have strong networks with farmers, FPOs, and incubation ecosystems (IIT Ropar, Sharda Launch Pad, AIC-BMU), which help in validation, partnerships, and deployment. This blend of technology, execution capability, and market access positions us uniquely to successfully build and scale this venture. |
Jiyaxo g techno pvt ltd |
https://jiyaxo.co.in |
New delhi |
We are building sustainabiale irrigation solution for farmers |
We are solving the problem of high-cost, polluting, and unreliable irrigation faced by small and marginal farmers who depend on diesel pumps and inconsistent electricity. Diesel pumps increase fuel expenses and emissions, while existing alternatives are expensive, fixed, or inefficient for deep borewells. This problem matters because it directly impacts farmer income, crop productivity, and environmental sustainability. High irrigation costs reduce profitability, and delayed watering affects yields. At the same time, diesel usage contributes significantly to carbon emissions and rural air pollution, making it critical to transition toward clean, affordable, and reliable irrigation solutions. |
Our solution addresses the problem by replacing diesel and grid-dependent irrigation with a Li ion powered portable deep suction water pump that delivers reliable performance without fuel or electricity constraints. It enables farmers to draw water from up to 200 ft depth, operate for 8 to 10 hours on a single charge, and irrigate fields at a much lower cost, ensuring timely irrigation and improved productivity.
What makes it novel is the combination of high-performance deep suction with battery-powered portability and a patented one second quick fit coupling, which allows instant tool free installation across borewells. This integrated system delivers diesel level performance with zero emissions reduced cost and unmatched ease of use, which is not available in existing solutions. |
Li ion powered portable deep suction water pump delivering up to 200 ft suction 8 to 10 hour runtime and 80 to 100 PSI pressure with patented one second quick fit coupling enabling instant tool free installation diesel free operation low cost irrigation and easy mobility for small farmers Defensibility (Competitive Moat) Protected by granted design patent and filed technical patents covering pump system and quick fit coupling strong technical integration of battery motor and deep suction performance first mover advantage in portable deep borewell segment cost efficiency and farmer centric usability creating high adoption barrier for competitors combined with continuous IP expansion and ecosystem partnerships |
Users |
Users |
Our target customers are small and marginal farmers, along with FPOs, Custom Hiring Centers, and agri-entrepreneurs who rely on borewell irrigation and currently use diesel or unreliable grid-powered pumps. We have validated this problem through direct field interactions and discussions with farmers, where key pain points identified include high diesel cost, frequent power cuts, difficulty in installation, and lack of portability. Farmers expressed a strong need for a low-cost, portable, and reliable irrigation solution. Initial prototype demonstrations and feedback indicate high interest and willingness to adopt a solution that reduces cost and simplifies irrigation operations. |
The Total Addressable Market (TAM) for irrigation pumps in India is estimated at ₹1.5–2 lakh crore, |
while the global market exceeds $25–30 billion. |
Our Serviceable Market (SAM) focuses on diesel and off-grid farmers, and our initial Serviceable Obtainable Market (SOM) targets high-demand states, enabling scalable growth. |
B2b,b2c, |
Our main competitors include Kirloskar Pumps, CRI Pumps, and Shakti Pumps, along with low-cost battery pump manufacturers. These players offer diesel, electric, or solar pumps but lack a combination of portability, deep suction, and ease of installation. We will win because our solution uniquely delivers deep suction up to 200 ft, long battery runtime, and a patented one-second quick-fit coupling that enables instant, tool-free installation. Unlike competitors, we provide diesel-level performance with zero emissions, lower operating cost, and true portability, making it highly suitable for small and marginal farmers. |
Customers will acquire our product through a multi-channel distribution approach designed for rural reach and scale. We will sell directly via local distributors and dealers, supported by on-field demonstrations to build trust among farmers. For institutional reach, we will partner with FPOs, Custom Hiring Centers, NGOs, and government schemes, enabling bulk adoption. Additionally, we will promote a shared usage / rental model through agri-entrepreneurs, making the product accessible to farmers who cannot afford upfront purchase. Digital outreach and referral-based sales will further support customer acquisition. |
We will adopt a hybrid B2C + B2B + B2G approach for rapid and scalable market entry. 1. Pilot & Validation (0–6 months): Deploy 20–30 units with farmers and FPOs through field trials and demonstrations to build trust and collect performance data. 2. Institutional Sales (B2B/B2G): Partner with FPOs, Custom Hiring Centers, NGOs, and government schemes for bulk procurement and faster adoption. 3. Distributor Network (B2C): Establish local agri-dealer and distributor channels in high diesel-dependent regions for direct farmer sales. 4. Service Model: Enable irrigation-as-a-service via agri-entrepreneurs to reduce upfront cost barriers for farmers. 5. Awareness & Adoption: Conduct on-field demos, farmer meetings, and referral programs to drive grassroots demand. 6. Scale Strategy: Expand regionally → pan India → international markets (South Asia, Africa). This strategy ensures fast adoption, low customer acquisition cost, and scalable |
Our vision is to become a global leader in clean-energy agricultural mechanization, transforming irrigation into a diesel-free, affordable, and accessible solution for every farmer. We aim to replace millions of diesel pumps with our portable Li-ion powered systems, reducing carbon emissions and improving farmer income at scale.
Over the next few years, we plan to build a strong presence across India and expand into Asia and Africa, where similar irrigation challenges exist. We will continuously innovate in battery technology, smart irrigation, and user-friendly design, creating a complete ecosystem of sustainable farm solutions.
Ultimately, our goal is to enable climate-resilient agriculture, improve rural livelihoods, and contribute significantly to global sustainability by making clean irrigation the new standard. |
Pvt Ltd, |
1000000 |
Yes |
We are struggling for funding to commercialisation our product, |
Funding support and mentoring, |
Yes we ready for both |
Yes |
Supporting feature |
We have implemented ai based monetoring in our system, |
Na |
Real time monetoring |
Na |
Through patent of algorithm and copy right of coding, |
Na |
Na |
Yes we have two ai expert in our team |
Owened |
Na |
For both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/8GVM5Z5DE25e-pitch-desk-2026-1_compressed-2.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/8GVM5Z5DE25e-pitch-desk-2026-1_compressed-2.pdf |
https://drive.google.com/file/d/1XFGtQMyd7Yw936Aq0wH-MwctvHeGPMHM/view?usp=drivesdk |
Yes, we are a mission-driven, impact-focused startup.
Our mission is to make irrigation clean, affordable, and accessible for small and marginal farmers by eliminating dependence on diesel and unreliable electricity. We are addressing a critical intersection of agriculture, climate change, and rural livelihoods, where high irrigation costs and pollution directly impact farmer income and sustainability.
Our Li-ion powered portable water pump reduces carbon emissions, fuel expenses, and operational complexity, enabling farmers to irrigate efficiently while improving productivity and income. Each unit has the potential to significantly cut CO₂ emissions and promote climate-resilient farming practices.
Beyond technology, we aim to create inclusive impact by supporting FPOs, agri-entrepreneurs, and rural communities through shared access models. Our long-term vision is to scale this solution across India and globally, contributing to sustainable agriculture, clean energy adoption, and rural economic development. |
We come from a rural background and grassroots ecosystem, which is often underrepresented in deep-tech and startup innovation spaces. Our journey is driven by first-hand exposure to real agricultural challenges, giving us unique insights into farmer needs and practical problem-solving. Despite limited access to advanced infrastructure and resources, we have built IP-backed innovations and working prototypes, demonstrating resilience, self-driven learning, and execution capability. Our perspective bridges the gap between technology and real-world rural problems, which is rarely addressed by conventional urban-centric startups. This background strengthens our mission to build inclusive, farmer-centric technology and create impact at scale. |
Mail |
checked |
| Apr 1, 2026 @ 4:22 PM |
Shyam Sundar , Vinay Kumar A R |
official.nexrtech@gmail.com |
https://www.linkedin.com/in/shyam-sundar-9881671b/ |
https://nexrtech.vercel.app/ |
9739277227 |
Shyam Sundar: CEO & Founder of a Health tech startup, leading product development, business strategy, and partnerships for immersive VR and AI-based wellbeing solutions. Background in HR and Student Success with 9+ years of experience in talent development, mentoring, and building people-centric programs. Vinay Kumar A R: COO at a Health tech startup, leading operations, scaling strategy, cross-functional execution, and market expansion. with background of Mechanical Engineer with 10+ years of experience in aviation design and renewable energy. Entrepreneur with 8+ years in the pharmaceutical industry with expertise in venture building, operations, regulatory processes, and commercialization. |
We met 8 years ago through a mutual friend and built a strong friendship over shared interests in technology and entrepreneurship. About 3 years ago, we decided to combine our complementary strengths and started working together as co-founders, focusing on building and scaling technology-driven solutions in the mental wellness space. |
All |
Our core strength lies in our complementary expertise across technology, business, people strategy, and clinical mental health. With Shyam bringing HR and people development expertise, Vinay contributing engineering and entrepreneurial execution, and Dr. Harish providing 8+ years of psychiatric and clinical insight, we combine domain knowledge with execution capability. This multidisciplinary foundation allows us to build solutions that are not only technologically scalable but also clinically relevant and human-centered. |
NEXR technology PVT LTD |
https://nexrtech.vercel.app/ |
HOSPET, Bellay district |
Nexr is an AI-powered mixed reality mental wellness platform that combines therapy, immersive experiences, and anonymous support to improve access, engagement, and outcomes. |
Mental healthcare in India suffers from three key barriers: 1, Limited access to professionals (severe availability gap) 2, Social stigma barrier for people from seeking help 3, Low engagement and adherence in traditional therapy (high dropout & relapse rates) |
NEXR is an AI powered mixed reality platform with three core components:
1, Multilingual AI Coach (Melo.AI): 24/7 conversational support for emotional guidance and therapy assistance
2, Metaverse Community (Melo-World): Anonymous, avatar-based environment for stigma-free interaction, peer support, and group therapy
3, VR Therapy: Clinician-led immersive therapy sessions with progress tracking, symptom monitoring, and AI-assisted documentation |
1, End-to-end integration: Combines AI, metaverse, and clinical VR therapy in one platform 2, Anonymity-first design: Avatar-based engagement reduces stigma significantly 3, Clinical + tech fusion: scientifically backed by psychiatrists with structured therapy protocols 4, Higher engagement & outcomes: Immersive therapy improves adherence and recovery rates |
MVP |
Pilots |
1, Primary: Corporates (HR/EAP) with high-stress, customer-facing employees (IT, BPO, Banking, Healthcare) 2, Secondary: Clinics & hospitals (therapists, psychiatrists) |
TAM: ₹3,640 Cr ($438M) – Total mental health market in India |
SAM: ₹728 Cr ($80M) – Corporate high-stress sectors |
SOM: ₹36.4 Cr ($4M) – Early adopters in Tier-1 cities |
1, B2B Enterprise SaaS: ₹1,300 per employee (full platform access) 2, App Subscription: ₹600 per user (metaverse - only access) 3, Clinics & Hospitals: ₹8,000/month per location or 15% revenue share 4, Corporate wellness Programs: ₹12K–₹15K per session |
1, Wundrsight 2, Avika 3, Amaha 4, Mindpeers 5, YourDost |
1, Corporate pilots and freemium trials 2, On-site VR wellness events (“stigma-free” experiences) 3, Partnerships with HR/EAP providers 4, Clinical partnerships (hospitals, mental health institutions) |
Phase 1: Validate (0–4 months) • Run freemium pilots • Partner with clinics (e.g., DIMHANS) • Pilot with 10 corporates + 10 coworking spaces Phase 2: Monetize (4–12 months) • Convert pilots into paid enterprise contracts • Publish case studies and clinical outcomes • Build HR/EAP partnerships |
To build the world’s most accessible and effective mental healthcare platform powered by AI and immersive technology |
Yes - we have incorporated |
NA |
Yes |
We believe Avinya will help us gain mentorship, clinical validation support, and market access to scale NEXR from pilot stage to a clinically credible, enterprise-ready platform. This incubation will help us accelerate product development, strengthen partnerships, and achieve scalable product-market fit. |
During this period, we want achieve clinical validation, convert pilots into paying customers, and build a scalable, enterprise-ready mental wellness platform. |
Yes |
Yes |
Core engine |
Core AI Layer (Models & Capabilities)
Built on multilingual LLM-based conversational models
• Key functions:
• Emotional guidance & psychoeducation
• Structured check-ins and symptom tracking
• Session assistance for therapy workflows
• AI-assisted therapist support (summaries, notes, recommendations)
• Designed to support Indian languages + English
Voice AI Stack
• Real-time communication: LiveKit (low-latency audio streaming)
• Speech-to-Text (STT): Converts user voice to text for processing
• Text-to-Speech (TTS): Generates natural voice responses
• Enables continuous, conversational voice interaction for therapy and support use cases
Intelligence Layer
Manages:
• Session context and memory
• Symptom check-ins and user state tracking
• Escalation triggers (risk detection, distress signals)
• Clinician handoff workflows
• Acts as the decision engine connecting AI models with clinical logic |
At this stage, our strongest advantage is not a massive closed dataset, but a structured product-design and therapy-protocol advantage built through deep study, repeated experimentation, and close alignment with mental health use cases in India. We are developing a proprietary interaction layer that combines multilingual emotional support, voice conversations, anonymous avatar-based participation, and immersive therapy journeys in one product. Over time, this creates differentiated data around engagement patterns, adherence, session progression, symptom tracking, and outcome-linked behavioral signals across text, voice, and XR interactions.
As we scale pilots, our proprietary advantage will increasingly come from anonymized and consent-based product usage patterns, therapy engagement pathways, and workflow data that help us improve personalization, retention, and clinical utility. |
1, System-Level Integration (Not Just LLM Access):
Our defensibility comes from integrating AI, voice, XR/VR, and clinical workflows into a single stack—far beyond a standalone chatbot or LLM wrapper.
2, Multimodal Architecture: We combine
• Multilingual conversational AI
• Real-time voice interaction
• Anonymous avatar/metaverse environments
• Clinician-assisted VR therapy
3, Therapy-Informed Design:
Our AI is built on clinical frameworks, therapy protocols, and behavioral insights (engagement, dropout, stigma), making it purpose-built for mental healthcare—not generic conversations.
The combination of clinical framing, immersive environments, voice infrastructure, and product workflow depth makes the system harder to replicate than a standard AI mental health assistant. |
We evaluate performance across two layers: model/system metrics and care-delivery outcomes
Model & System-Level Metrics
Response quality: relevance, coherence, and helpfulness
Multilingual accuracy: comprehension and response quality across languages
Latency: response time for text and voice interactions
Conversation stability: interruption handling and session continuity in real-time use
Error & hallucination rate: factual and contextual correctness in guidance and summaries
Escalation accuracy: correct triggering of risk flags and clinician handoffs
Product & Outcome-Level Metrics
User engagement: active usage, repeat sessions
Session completion rate: % of sessions successfully completed
Adherence: follow-through on recommended activities/interventions
Dropout reduction: decrease in user attrition over time
Consistency: frequency of symptom check-ins |
We treat privacy and trust as core product requirements. Our approach includes consent-based data collection, role-based access controls, encrypted communication, secure session handling, and controlled storage of user interaction data. Sensitive user information is separated from operational workflows wherever possible, and we are designing the platform to support data minimization, audit trails, and configurable retention policies.
For anonymous community use cases, the product is designed to reduce exposure of personal identity by default through avatar-based participation. For clinical workflows, therapist and institution-level access is permissioned. We are also careful about how AI outputs are used: the system is positioned as assistive and supportive, with escalation and clinician involvement for higher-risk scenarios rather than unsupervised diagnosis or treatment claims. |
Yes. Mental health and AI-assisted care naturally sit close to privacy, clinical responsibility, and medical-claim boundaries. The main regulatory risks are around handling sensitive health-related data, how therapeutic outcomes are represented, and ensuring the platform is not mispositioned as an unsupervised medical substitute where clinician oversight is needed.
We manage this by keeping a clear distinction between wellness support, therapy assistance, and clinician-led interventions. Our clinical modules are intended to work with mental health professionals, and we plan to continue strengthening consent, documentation, privacy controls, and safety escalation processes as part of product maturity. |
At 10x scale, key pressure points include real-time voice concurrency, latency, session stability, and infrastructure costs, which can impact response quality—especially across multilingual and immersive experiences
We mitigate this with a modular architecture (separate scaling of voice, AI orchestration, and analytics), along with strong monitoring, escalation logic, and human-in-the-loop quality control to maintain reliability |
Yes. Our in-house team combines expertise in AI/ML, product engineering, real-time voice systems, and immersive (XR/VR) development.
We have hands-on experience in LLM-based workflows, multilingual conversational AI, and real-time voice pipelines, integrated into production-grade applications. Additionally, team have spent substantial time on therapy theory, conversational safety, behavior design, adherence challenges, and how to make AI useful in actual mental wellness settings rather than standalone chatbot solutions. |
We use a combination of licensed foundation-model capabilities and internally developed, therapist-informed datasets and workflow layers built specifically for mental wellness use cases. These include structured conversation pathways, multilingual interaction patterns, therapy-support frameworks, escalation boundaries, and safety rules refined through close collaboration with therapists, repeated product trials, and real-world testing.
Our advantage is not just in raw data volume, but in the quality of domain shaping behind the system. We have invested heavily in understanding therapeutic theory, session flow, emotional support boundaries, and how AI should behave safely in sensitive mental wellness contexts. This has allowed us to build an internal applied-intelligence layer that is much more specialized than a general-purpose conversational AI system.
As pilots scale, we are also building a consent-driven and anonymized data asset around user engagement, adherence patterns, symptom progression signals, therapist feedback, and outcome-linked behavioral insights. This creates a strong long-term owned data advantage while keeping privacy, consent, and critical AI safety boundaries central to the platform. |
We plan to improve the system through an iterative loop of pilot feedback, clinician input, usage analytics, and targeted evaluation of failure cases. In practical terms, this means reviewing low-quality conversations, monitoring escalation events, testing multilingual performance, improving prompts and orchestration logic, and refining symptom-check and therapy-assist flows based on real user behavior.
We also plan to improve performance at three levels: model and prompt quality, workflow quality across voice, text, and XR sessions, and outcome quality in terms of adherence, engagement, and clinical usefulness. The goal is not just smarter responses, but safer, more context-aware, and more effective support over time.
Improvements are made across three levels:
• Model & prompt quality (accuracy, relevance)
• Workflow quality (voice, text, XR interactions)
• Outcome quality (engagement, adherence, clinical usefulness) |
We are building for both, with India as the immediate and primary market. India is the right starting point because the mental health access gap, stigma barrier, language diversity, and shortage of professionals make the problem especially urgent here. At the same time, our architecture is being designed to scale beyond India, especially for multilingual, enterprise wellness, and digitally delivered therapy-support use cases in other markets. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/NMsZVsaGG5nG-NEXR-ASCEL-PITCH-DECK.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/NMsZVsaGG5nG-NEXR-ASCEL-PITCH-DECK.pdf |
https://drive.google.com/file/d/1P6z1JysI5STlfGav1CzL6Hi2J_HEe_fp/view?usp=sharing |
NEXR is a both mission-driven, impact-focused AI startup addressing one of India’s most critical gaps—accessible and stigma-free mental healthcare.
Our mission is to make mental wellness support accessible, engaging, and effective at scale by leveraging AI and immersive technologies. Today, a large portion of the population does not seek help due to stigma, lack of access, and low engagement in traditional therapy. We are solving this by building an AI-powered platform that combines multilingual conversational support, real-time voice interaction, and immersive XR-based therapy, designed specifically for real-world adoption.
From an impact perspective, we aim to:
Improve access by providing 24/7, multilingual AI support
Reduce stigma through anonymous, avatar-based environments
Increase adherence and outcomes using immersive and engaging therapy formats
Augment clinicians to scale mental healthcare delivery efficiently
As we scale, our goal is to become a core digital infrastructure for mental healthcare, driving measurable improvements in productivity, well-being, and clinical outcomes across enterprises and healthcare systems. |
Yes, we come from a rural background and are first-generation entrepreneurs, which places us within an underrepresented group in the tech startup ecosystem. This journey has given us a deep understanding of access gaps, stigma, and resource limitations, especially in areas like mental healthcare. It strongly influences our approach to building NEXR—focused on accessibility, affordability, and real-world usability for diverse populations. Our background not only reflects resilience but also helps us design solutions that are inclusive and relevant to underserved communities at scale. |
Kunal from Cipherion & Ram Subramanian Sir from IIIT Dharwad |
checked |
| Apr 1, 2026 @ 11:29 AM |
Vaibhav Sharma, Agniva Ray |
vaibhav@kubar.tech |
https://www.linkedin.com/in/fenestbuc/ |
https://kubar.tech |
+917872260684 |
Vaibhav Sharma - CEO & Head of Product. 22-year-old founder with deep product and fintech infrastructure expertise. Built NavDhan's credit scoring model, lender integration architecture, and product strategy. Former Polkadot ecosystem contributor. Agniva Ray - COO. Leads fundraising, external partnerships, and operations. Co-leads lender and marketplace engagements as deals mature. |
Vaibhav and Agniva connected through the India startup ecosystem while independently building at the intersection of fintech and AI. We have been working together for over a year, initially as collaborators and now as co-founders of Kubar Labs. Our complementary strengths - Vaibhav in product and technical architecture, Agniva in operations and fundraising - make us well-suited to build a capital-intensive B2B credit infrastructure company. |
All |
We combine deep technical product thinking with strong business development instincts. Vaibhav's ability to go from whiteboard to working prototype fast - while staying grounded in lender and MSME realities - has let us build a credit scoring model and agentic interview stack that no competitor in our segment has. Agniva's operational rigor keeps us accountable on timelines and capital efficiency. We have signed a live pilot with Clix Capital (25 borrowers, growing), established exclusive partnerships with AgriGrader, and are in active conversations with SBI and PNB - all within our first year. |
Kubar Labs (Product: NavDhan) |
https://kubar.tech |
Noida, Uttar Pradesh |
NavDhan is India's sector-specialized embedded MSME credit platform - the intel-inside layer that helps lenders reach, assess, and disburse to underserved small businesses in 1-3 hours. |
India has 64 million MSMEs, but only 11% have access to formal credit. The $330B MSME credit gap persists not because lenders lack capital, but because they lack the tools to assess and reach thin-file borrowers at scale. Traditional credit scoring misses 85% of an MSME's real financial picture. Loan disbursals take weeks. CAC for lenders is prohibitively high. DSA networks are opaque, expensive, and unable to scale into Tier 2/3 markets. On the borrower side, MSMEs struggle with complex forms, language barriers, and no visibility into loan eligibility. |
NavDhan is the embedded credit intelligence layer for India's MSME ecosystem with four core components: (1) Agentic DSA - A voice-based AI agent replacing human DSAs. It calls borrowers in their native language and conducts a mandatory 10-15 minute credit interview capturing 100% of relevant borrower data vs the 30% captured by traditional methods. (2) Sector-specific credit scoring - Proprietary model using GST/invoice data, ULI (Unified Lending Interface), marketplace signals (ONDC, IndiaMart, GeM), account aggregator data, and Kubar sector intelligence. Targets 35-40% approval rates vs 15-20% industry average. (3) Lender ops console + SaaS analytics - Full LOS/LMS integration with lender-facing dashboard and real-time portfolio insights. Launching as SaaS at Rs 25K-1L/month from Month 12. (4) Telemetric module - Reduces marketplace and ERP integration time from one quarter to two weeks. Current priority: GeM Purchase Order Financing - pre-shipment loans for MSME exporters with verified government POs. |
Three moats: First, proprietary sector intelligence. NavDhan builds sector-specific credit models (textiles, agri, manufacturing, GeM exporters) that generic NBFC scorecards cannot replicate. Our AgriGrader partnership gives us exclusive credit scoring rights for a verified agricultural marketplace - a data advantage that compounds with every loan. Second, the Agentic DSA stack. Our voice-based multilingual credit interview is the only automated system in Indian MSME lending that replicates a human DSA interaction while capturing structured credit data. Creates a defensible data flywheel: more interviews = better models = higher approval rates = more lender trust. Third, embedded distribution. By integrating into platforms MSMEs already use (ONDC, GeM, IndiaMart, Tally, AgriGrader), NavDhan appears at the point of need. CAC below Rs 500. We also won the RBI FinVision Award. |
Revenue |
Revenue, Pilots |
Two-sided market: Lenders (primary paying customers) - Banks (SBI, PNB), NBFCs (Clix Capital, Shriram Finance, IDFC First), and co-lending platforms looking to expand MSME portfolios without rebuilding underwriting infrastructure. MSMEs (end borrowers) - Sector-specialized micro and small businesses in textiles (Surat, Tiruppur), manufacturing (Pune, Noida), agriculture (via AgriGrader), and GeM-registered MSME exporters with Rs 25L-1Cr annual turnover, thin CIBIL files, but verifiable GST and marketplace transaction data. |
$330B - India formal MSME credit gap. Global MSME credit gap is $5T. |
$50B - addressable segment of tech-enabled MSME lending through digital platforms in India over the next 5 years. |
Rs 606 Cr (~$72M) in loan portfolio by Month 24, targeting 15,000 loans across textiles, agri, manufacturing, and GeM exporters. |
Asset-light, no FLDG (First Loss Default Guarantee): (1) Transaction fee - % of each loan disbursed, primary revenue, scales with volume. (2) SaaS subscription - Rs 25K-1L/month per lender for ops console and analytics dashboard, launching Month 12. Unit economics: Rs 74 cost per loan, CAC below Rs 500, NPA target below 1.5%. |
Recur Club, Kinara Capital, NeoGrowth - traditional MSME lenders/NBFCs with balance sheets. NavDhan does not lend - we are the intelligence and distribution layer. CreditEnable, Yubi (CredAvenue) - generic credit marketplaces. NavDhan is sector-specialized and embedded. Perfios, Bureau.id - credit data providers only. NavDhan goes beyond data aggregation into agentic loan origination. No competitor combines embedded marketplace distribution + sector-specific AI scoring + voice-based agentic DSA in one stack. |
Lenders: Direct B2B sales through founder-led outreach and warm ecosystem introductions (met PNB at FinTech Edge NIBM, Clix Capital through network). MSMEs: Embedded acquisition through marketplace partners (AgriGrader, GeM, ONDC, IndiaMart). MSMEs encounter NavDhan inside tools they already use - zero cold outreach required. CAC below Rs 500. |
Phase 1 (M1-6, current): 1,000 loans in textiles and manufacturing across Surat, Tiruppur, Pune via Clix Capital and AgriGrader. Prove NPA below 1.5% and lender satisfaction. Phase 2 (M7-12): Onboard SBI or PNB. Launch GeM PO Financing. Launch lender SaaS console. 5,000 loans. Phase 3 (M13-24): Scale to 15,000 loans, Rs 606 Cr portfolio. Expand Agentic DSA to 5+ regional languages. |
NavDhan becomes the intel-inside layer for MSME credit in India - the intelligence and origination layer that every lender and marketplace plugs into. 10-year vision: Kubar Labs evolves into the Kubar Protocol, extending this credit infrastructure globally through tokenized trade finance and programmable liquidity pools, solving the $5T global MSME credit gap. |
Incorporated - Private Limited Company, India |
Pre-seed stage, bootstrapped to date. Active fundraise in progress: targeting $1.5M at $20M valuation. In diligence with Google AIFF program. Pipeline includes IPV IdeaSchool (top 100, Rs 7 Cr pre-money) and STPI FinGlobe GIFT City. VC conversations ongoing with 3one4, Blume, Better Capital, Accion Venture Lab. |
Yes |
We are at a critical inflection point - first lender live, product validated, fundraise underway. What we need now is structured mentorship from founders and operators who have built B2B fintech companies through the messy middle: how to close enterprise sales cycles, build a lender-grade compliance stack faster, and scale a two-sided marketplace without losing unit economics. Avinya network and programme structure can compress years of learning into months. Being in an AI-focused cohort will also sharpen our product thinking. |
1. Close our seed round with programme support and network introductions. 2. Sign 2-3 new lender partnerships (targeting SBI/PNB co-lending and Shriram Finance). 3. Launch GeM PO Financing with first 500 disbursals. 4. Refine Agentic DSA architecture with expert mentorship on AI model evaluation and compliance. 5. Build relationships with founders who have navigated regulated fintech at scale in India. |
Yes |
Yes |
Core engine |
NavDhan AI architecture has three layers: Layer 1 - Agentic DSA (Conversational AI): Built on LLMs (using Gemini for TTS, evaluation, and multilingual voice synthesis) with a custom conversation design framework. The agent follows a structured interview protocol covering 40+ credit-relevant signals across business operations, vendor relationships, cash flow patterns, and sector-specific factors. Operates in Hindi, English, and regional Indic languages. Layer 2 - Sector-Specific Credit Scoring: Proprietary model combining Bureau data (CIBIL, DPD, enquiries), GST and invoice analytics (buyer concentration, invoice consistency, GRC scores), banking transaction data, marketplace signals (ONDC, GeM, AgriGrader), and ULI data. NTC borrowers scored on an alternate pathway. Layer 3 - Lender Intelligence Console: Real-time portfolio analytics, predictive NPA signals, and sectoral benchmarking powered by BigQuery and Vertex AI. |
Yes - multiple owned data assets: (1) Exclusive credit scoring rights for AgriGrader marketplace transaction data (agricultural buyer-seller flows, auction results, quality grades). (2) GeM transaction and PO verification data for government MSME exporters. (3) Agentic DSA interview transcripts - structured credit interview data from 10-15 minute sessions that no competitor is collecting at this granularity. (4) Sector-specific training data from textiles (Surat/Tiruppur) and manufacturing (Pune) clusters. This data is owned, not licensed. It compounds with every loan. |
The Agentic DSA creates a data flywheel: more interviews generate richer training data, which improves credit models, which increases approval rates, which attracts more lenders, which drives more borrowers. The sector-specific scoring cannot be replicated without the same embedded marketplace partnerships - competitors would need to rebuild our distribution agreements from scratch. Our conversation design IP (interview protocol, language handling, and trust-building frameworks) is protected as trade know-how. |
Approval rate vs NPA correlation (primary metric): targeting 35-40% approval with sub-1.5% NPA. Interview completion rate and drop-off analysis per language and sector. Credit model backtesting on historical bureau and GST data. Gemini-based automated evaluation of Agentic DSA conversation quality (coherence, data capture completeness, borrower sentiment). Human review loop for edge cases and model disagreements. |
Account Aggregator (AA) framework compliance for financial data consent. RBI digital lending guidelines adherence. ULI (Unified Lending Interface) integration for regulated data flows. DPDPA (Digital Personal Data Protection Act) compliance in data handling. PII encryption at rest and in transit. Lender-specific data segregation architecture. No storage of raw financial documents beyond verification window. |
Primary: RBI regulation of LSP (Loan Service Provider) and DSA activities. We are structured as a technology partner, not a lender, which keeps us out of NBFC licensing requirements. RBI evolving digital lending guidelines require ongoing monitoring. We maintain active compliance discussions with IDFC and legal counsel. Secondary: DPDPA implementation uncertainty for AA-based data flows. We are building consent-first architecture to be compliant regardless of final regulatory interpretation. |
(1) Agentic DSA concurrency - current architecture handles around 100 simultaneous voice sessions. At 10x, we need distributed session management and model serving infrastructure, addressed via Modal for serverless GPU scaling. (2) Credit model drift - sector-specific models need retraining as macroeconomic conditions shift. At 10x volume, we need automated drift detection and retraining pipelines. (3) Lender onboarding - each integration currently requires custom LOS/LMS mapping. Our Telemetric module provides a standardized API abstraction layer at scale. |
Yes: Ayan Gangopadhyay - AI/NLP Lead, conversation design architect, Agentic DSA system design. Rayansh Srivastava - Founding Engineer, model serving and backend infrastructure. Parth Ahuja - Engineer, Agentic DSA implementation. Debpriyo Ghoshal - Engineer, credit model integration. |
CIBIL/Experian bureau data (licensed via lender partnerships), GST transaction data (borrower-consented via AA framework), AgriGrader marketplace data (owned - exclusive partnership), GeM PO data (owned via API integration), ULI financial data (RBI-regulated consent-based access), Agentic DSA interview transcripts (owned, borrower-consented). All data handling is consent-based or licensed. |
Monthly retraining cycles using disbursed loan and repayment outcome data. Automated evaluation of Agentic DSA conversations via Gemini. Sector-specific model versioning with A/B testing on approval criteria. Feedback loops from lender ops teams on portfolio performance. Active research monitoring on alternative credit scoring for thin-file borrowers. |
India first, global next. NavDhan is built specifically for India regulatory and market context (AA framework, ONDC, GST, ULI). Once the India stack is proven, Kubar Protocol will extend this infrastructure globally via tokenized trade finance and MLETR cross-border rails, targeting the $5T global MSME credit gap. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/8E92cO4y67fo-navdhan_deck_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/8E92cO4y67fo-navdhan_deck_compressed.pdf |
https://drive.google.com/file/d/1kH2a9DwAyKyPJX-aol0KIT12wMfCPit1/view?usp=sharing |
Yes. NavDhan is built around a single conviction: India's 64 million MSMEs are the backbone of the country's economy, yet 89% of them are excluded from formal credit. This is not just a market gap - it is a structural barrier to economic mobility. Every loan we enable is a small business that can fulfill an order, hire a worker, or survive a cash crunch without turning to informal moneylenders at 3-5% monthly interest. Our RBI FinVision Award recognized this mission. We are building for financial resilience in India's agricultural and manufacturing value chains - sectors that directly impact rural livelihoods. Kubar's 10-year vision is anchored in the belief that inclusive access to capital is the most powerful lever for economic mobility. |
Vaibhav Sharma is a 22-year-old dropout founder building in Tier 2 India (Noida) - outside the IIT/IIM pipeline that dominates Indian startup funding. Kubar's team is deliberately non-metropolitan, with most team members from outside Delhi/Bengaluru's typical startup corridors. We are also building for borrowers in Tier 2/3 cities and rural India - communities consistently underrepresented as end users in VC-funded startups. |
NA |
checked |
| Mar 31, 2026 @ 4:12 PM |
Nitin Jagadish |
nitin@augustinnovate.com |
|
|
09845375943 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 31, 2026 @ 11:58 AM |
Gaurav Sisodia |
gaurav.sisodia@kecserv.com |
https://www.linkedin.com/in/gaurav-s-3324033b/ |
https://www.kecserv.com/ |
9718000301 |
Gaurav Sisodia - Founder, CEO & Director, Manini Shah - Head of Brand & Institutional Positioning, Jagdish Yadav - Chief Technical Officer, Akhil Achari - Key Account Manager, Gourish Bakshi - Head Analytics, Cynthia - Program Head, Vikram Singh - Manager IT Infra, Sumit Kumar - AI Engineer, Vikas Gupta - Executive Systems & IT |
The core team came together through professional networks and prior industry collaborations in BFSI and technology. While some members have worked together earlier, the full team has been actively working together over the past year to build and scale FinotoAI. |
1 |
Our biggest strength is the combination of deep BFSI domain expertise and strong AI/technology capabilities, enabling us to understand real lending challenges and build scalable, practical solutions for financial institutions. |
KECSERV OPC Private Limited |
https://www.kecserv.com/ |
New Delhi |
FinotoAI is an AI-powered platform that automates the entire credit lifecycle for banks and NBFCs, enabling faster lending, lower costs, and better risk decisions |
Current lending processes are largely manual and fragmented, leading to 5–10 days of loan processing time, 24–72 hours for KYC, 60–70% manual underwriting, and ₹500–₹8,000 cost per file, with 15–25% rework due to errors and weak early NPA detection. This results in delays, high costs, and inconsistent credit quality, reducing efficiency and borrower satisfaction. |
We are developing an AI-powered unified credit lifecycle platform that automates lending processes from customer onboarding to loan recovery. Our solution uses AI, OCR, and predictive analytics to reduce processing time, operational costs, and credit risk.
We are solving this problem for banks, NBFCs, HFCs, fintech lenders, and financial institutions.
Finoto AI offers a unified, AI-powered credit lifecycle platform with features including AI-based lead qualification, OCR-enabled KYC and document verification, automated underwriting and risk scoring, technical and legal valuation support, predictive financial analysis, smart loan disbursement, AI-driven collections management, and multilingual customer support. The platform provides real-time dashboards, audit-ready compliance reporting, secure API integration with existing systems, and human-in-the-loop governance to ensure accuracy, transparency, and regulatory alignment. |
Our solution is unique because it provides a unified, end-to-end AI platform covering the complete credit lifecycle from acquisition to collections, rather than isolated point solutions. It is built on India-hosted, compliance-aligned infrastructure with integrated AI decision support, audit-ready governance, modular deployment, and human-in-the-loop controls, enabling faster processing, lower costs, and improved credit quality. |
Revenue |
Revenue |
Banks, NBFCs, HFCs, Fintech lenders, Financial Institutions |
$30 Billion |
$500 Million |
$15 Million |
SaaS |
Our competitors include traditional LOS/LMS and core banking providers such as TCS BaNCS, Nucleus Software, Nelito Systems, and Newgen Software, as well as point-solution and manual service providers |
We acquire customers through direct B2B sales to banks and NBFCs, leveraging founder-led industry networks, live product demos, pilot deployments, and participation in BFSI, fintech industry events and social media |
Our go-to-market strategy focuses on direct B2B enterprise sales to NBFCs, HFCs, and banks, starting with pilot deployments to demonstrate ROI, followed by multi-module expansion. We leverage founder-led networks, industry partnerships, and fintech events to drive adoption and scale. |
Our long-term vision is to become the core AI-driven credit infrastructure layer for financial institutions, powering end-to-end lending across banks, NBFCs, and fintechs with faster, more efficient, and compliant credit decision-making. |
Incorporated |
NA |
Yes |
We are applying to Avinya to gain strategic mentorship, refine our go-to-market in the BFSI sector, and access a strong network of industry leaders and partners. The program will help us strengthen product-market fit, compliance readiness, and accelerate adoption of our AI-driven credit infrastructure platform |
During the programme, we aim to refine our BFSI go-to-market strategy, onboard new institutional clients, strengthen compliance and governance frameworks, and accelerate product adoption. We also seek mentorship and partnerships to scale our AI-driven credit infrastructure platform effectively |
Yes |
Yes |
Core engine |
FinotoAI uses a hybrid AI architecture combining document intelligence, LLMs, and rule-based validation systems.
* Document processing pipelines extract structured data from unstructured financial and property documents
* LLMs are used for contextual understanding and assisted report drafting
* A rule engine validates outputs against credit policies and regulatory requirements
* A workflow orchestration layer ensures human-in-the-loop approvals for critical decisions
This combination ensures both automation efficiency and compliance reliability in regulated environments. |
We are building a proprietary dataset through:
* Structured valuation reports across multiple institutions
* Field-level survey data and validation inputs
* Bank-specific format mappings and workflow patterns
* Feedback loops from drafter/checker corrections
Over time, this creates a domain-specific dataset for credit workflows, which improves model accuracy and institutional customization. |
* Deep integration into workflow (not standalone AI tool)
* Combination of AI + rule engine + human validation
* Institution-specific configurations and data structures
* Accumulated proprietary workflow and validation datasets
* High switching cost once embedded into credit operations
Defensibility comes from data + workflow integration + compliance alignment, not just models. |
We evaluate performance using:
* Data extraction accuracy (field-level precision/recall)
* Validation accuracy (rule compliance success rate)
* Drafting efficiency (time reduction per report)
* Error rate reduction vs manual process
* Human override / correction frequency
Reliability is ensured through human-in-the-loop validation and audit logs. |
* Role-Based Access Control (RBAC)
* Audit logs and activity traceability
* Data encryption (in transit and at rest)
* Secure cloud deployment (India-hosted)
* Controlled API access and vendor permissions
* Data retention and access policies
The system is designed to align with regulated BFSI compliance requirements. |
Yes, key risks include:
* Data privacy regulations
* AI governance in financial decision-making
* RBI/financial sector compliance norms
We mitigate this through human-in-the-loop systems, auditability, and configurable rule-based validation, ensuring AI supports—not replaces—regulated decisions. |
At 10x scale, challenges include:
* Increased variability in document formats and quality
* Higher compute and inference cost
* Model drift across different institutions
* Workflow complexity across multiple configurations
We are addressing this through modular architecture, scalable cloud infrastructure, and continuous model tuning pipelines. |
Currently, we operate with a founder-led product and architecture approach, supported by AI/ML consultants and engineers for model development and implementation.
We plan to build a dedicated in-house AI/ML team as deployments scale, focusing on document intelligence, LLM optimization, and workflow automation. |
* Customer-provided documents (valuation, property, financial)
* Publicly available datasets (where applicable)
* Internally structured datasets from platform usage
All customer data is used with appropriate permissions, and we are building proprietary structured datasets over time. |
* Feedback loops from drafter/checker corrections
* Continuous retraining and fine-tuning
* Institution-specific model tuning
* Expansion of proprietary datasets
* Monitoring performance metrics and error patterns |
Currently focused on India, given regulatory and workflow complexity in BFSI.
The architecture is designed to be globally adaptable in apac & africa enabling expansion into other regulated lending markets over time. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/ziuHNkIR9KDn-Finoto-AI-Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/ziuHNkIR9KDn-Finoto-AI-Pitch-Deck.pdf |
https://youtu.be/2cmd_Sd91cg |
Yes, FinotoAI is a mission-driven startup focused on transforming credit accessibility and efficiency. Our goal is to make lending faster, more transparent, and data-driven by replacing manual, fragmented processes with AI-powered infrastructure. By improving risk assessment and reducing costs, we enable financial institutions to serve underserved segments, including MSMEs and new-to-credit borrowers, thereby supporting financial inclusion and responsible lending at scale. |
No, I do not belong to an underrepresented group. However, I strongly value building a diverse and inclusive team and ecosystem within the fintech and technology space. |
NA |
checked |
| Mar 31, 2026 @ 11:13 AM |
GAURAV GUPTA |
gaurav.gupta@graymobility.ai |
https://www.linkedin.com/in/gaurav-gupta-27410510/ |
https://graymobility.ai/ |
9673113327 |
Gaurav Gupta is an automotive industry professional with over 20 years of experience in vehicle engineering, product development, and mobility systems. An alumnus of Indian Institute of Management Bangalore, he has built a career at the intersection of automotive engineering and next-generation mobility technologies. Nabeel T K is an automotive engineering professional with over 10 years of experience in vehicle development, product engineering, and advanced mobility systems. An alumnus of Indian Institute of Technology Madras, he has built strong expertise in designing and developing high-performance automotive platforms with a focus on engineering reliability and system integration. |
We are long time Industry friends and we've worked together for about 10 years. |
All |
Gaurav Gupta exemplifies vision and execution leadership, bringing strategic thinking, industry insight, and the ability to scale complex EV platforms.
Nabeel T K represents technical leadership and innovation, with deep expertise in Software-Defined Vehicles, embedded systems, and AI. |
GRAY MOBILITY PRIVATE LIMITED |
https://graymobility.ai/ |
BANGALORE |
Gray Mobility builds AI-first, energy-intelligent electric vehicles that ensure reliable, mission-critical mobility starting with smart ambulances. |
Commercial EV adoption is accelerating, but most pilots fail to scale due to downtime and energy uncertainty. Over 60% of fleet burden costs stem from energy-related issues like battery degradation, inefficiency, unexpected maintenance, and operational downtime. Current fleets lack predictive visibility into failures and real usable energy. As a result, EV adoption remains costly, fragile, and operationally unreliable. |
Gray builds electric vehicles as mission-aware, energy-intelligent systems by embedding edge AI directly into Software-Defined Vehicles and fleets.
Its Energy Truth engine models real-world usable energy with degradation-aware, confidence-bound predictions.
Mission-aware vehicles validate every trip before execution, flagging energy and operational risks early.
Predictive intelligence prevents failures before breakdown, shifting maintenance from reactive to behavior-driven.
Fleet-level orchestration and infrastructure-aware routing optimize charge, route, and load decisions across coordinated systems. |
By combining edge AI and energy-intelligent systems, the ambulance aims to improve patient stabilization and lifesaving outcomes by about 20% during emergency transit. |
MVP |
Pilots |
Our target customers are institutional fleet operators, including government EMS agencies, state health departments, corporate hospital chains, and emergency response service providers operating large-scale ambulance fleets. |
$ 16.7 Bn |
$ 3.34 Bn |
$ 668 Mn |
Gray Mobility operates a hybrid B2B revenue model. We generate revenue through the sale or lease of software-defined emergency vehicles to hospitals, fleet operators, and government EMS agencies, combined with recurring SaaS fees for our AI-driven monitoring, diagnostics, and fleet optimization platform. |
Our main competitors include traditional ambulance OEMs like Force Motors and Tata Motors, electric vehicle manufacturers such as Tata EV and Mahindra Electric, and fleet telematics/SaaS providers offering tracking and analytics. However, none combine energy intelligence, mission-aware AI, and predictive reliability into a single integrated platform like Gray Mobility. |
We acquire customers through direct enterprise sales targeting EMS operators, hospitals, and government agencies. We begin with paid pilot deployments (2–10 vehicles) to demonstrate KPIs, then convert into multi-year contracts. We also leverage government tenders, partnerships, and industry networks to scale adoption across institutional fleets. |
Our go-to-market strategy follows a pilot-to-scale enterprise model. We engage EMS operators and hospitals through direct sales, begin with paid pilot deployments to validate KPIs, and convert into multi-year SLA-backed contracts. We leverage government tenders, partnerships, and expand into UAE/MENA and adjacent fleet segments for scale. |
Our long-term vision is to build the energy-intelligence operating system for mission-critical mobility, transforming vehicles into AI-driven, self-optimizing systems. We aim to scale from Smart e-Ambulances to a platform across logistics, cold-chain, and public infrastructure fleets. Over time, we envision enabling autonomous, infrastructure-integrated mobility systems, positioning Gray Mobility as a global leader delivering reliable, sustainable transportation solutions across multiple markets and critical use cases. |
Incorporated |
75 LAKH INR (BOOTSTRAPPED) |
Yes |
We are applying to Avinya to scale from MVP traction to enterprise sales while accelerating R&D. The program’s focus on mobility and deep engineering will help us validate our Smart e-Ambulance through pilots, refine our technology, and access mentorship, infrastructure, and partnerships needed for faster commercialization and growth. |
We aim to advance from prototype to pilot-validated product (TRL 6/7), deploy Smart e-Ambulances with institutional partners, and validate key performance metrics. Additionally, we seek to convert early interest into enterprise sales, accelerate R&D, and achieve regulatory readiness for scalable commercialization across target markets. |
Yes, we are open to hybrid and virtual participation. We are flexible in engaging remotely for regular program activities while also willing to attend key on-site sessions when required. This approach allows us to balance execution on our product development with active participation in the program. |
Yes |
Core engine |
Our AI architecture is a hybrid edge–cloud system embedded in a Software-Defined Vehicle. Edge compute runs energy prediction, mission validation, and failure detection in real time, while cloud systems handle model training and fleet optimization. Continuous data loops, OTA updates, and secure pipelines ensure scalable, adaptive, low-latency performance. |
Our proprietary advantage comes from real-world, mission-critical datasets combining battery behavior, vehicle telemetry, and mission profiles. This enables highly accurate energy prediction and failure models. Continuous fleet data creates a strong data flywheel, improving model performance over time and making our system increasingly predictive and difficult to replicate. |
Our AI is defensible through proprietary data, deep hardware-software integration, and domain-specific models. We generate unique real-world datasets across energy and mission behavior, creating a data flywheel. Our AI runs at the vehicle edge, tightly integrated with systems, making replication difficult without similar data, architecture, and deployment scale. |
We evaluate performance using energy prediction accuracy, mission feasibility precision/recall, failure prediction accuracy and lead time, uptime (>98%), downtime reduction, latency, and false positive/negative rates. These metrics are continuously validated through real-world pilot deployments and fleet data to ensure reliability and scalability. |
We follow a privacy-by-design approach with data minimization and anonymization. All data is encrypted. We implement RBAC, multi-factor authentication, and audit logs. Secure boot and firmware signing protect devices. Systems align with India’s DPDP Act and global standards, ensuring secure, compliant data handling. |
Yes, regulatory risks include automotive homologation (AIS/CMVR standards), compliance for onboard medical systems, and data privacy laws like the DPDP Act. Emerging AI safety regulations and approvals for EMS deployments may also impact timelines. We mitigate this through early regulatory engagement, compliance-first design, and continuous alignment with standards. |
At 10x scale, challenges include data drift, model generalization across diverse conditions, increased latency, and strain on edge-cloud synchronization. Prediction accuracy and failure detection may degrade without continuous retraining. We address this through scalable architecture, edge-first inference, distributed training, and robust data pipelines for consistent performance. |
Yes, we have an in-house AI/ML team specializing in edge AI, predictive modeling, and real-time decision systems. They develop energy prediction, mission-aware, and failure detection models using telemetry data, and focus on embedded deployment, low-latency inference, and continuous learning through fleet data for scalable, reliable performance. |
We use a mix of proprietary and public datasets. Proprietary data includes simulated and early telemetry on battery behavior, vehicle performance, and mission profiles, which we own. We also use licensed/public research datasets for benchmarking. Over time, we will rely increasingly on first-party fleet data, strengthening our owned data advantage. |
We use a data flywheel approach, continuously collecting real-world fleet data for retraining and validation. Models are updated via OTA with edge inference and cloud training. We monitor KPIs, detect drift, run A/B tests, and refine models across conditions, ensuring adaptive, scalable, and reliable performance as deployments grow. |
We are building for both India and global markets. Our initial focus is India due to strong EMS demand and EV adoption, followed by expansion into UAE and MENA regions. Our platform is designed to be scalable and adaptable across geographies, regulatory environments, and diverse mission-critical fleet requirements. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/UVhQ57kZnxlQ-GM_Pitch_Deck_TRL_3-1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/UVhQ57kZnxlQ-GM_Pitch_Deck_TRL_3-1.pdf |
https://youtu.be/hzXqgH_m5fc |
Yes, Gray Mobility is a mission-driven, impact-focused startup addressing critical gaps in emergency response and EV reliability. We aim to improve patient outcomes by enabling predictable, high-uptime ambulance systems with ≥20% potential improvement in stabilization. Our platform also accelerates sustainable mobility by solving energy uncertainty in EV fleets. This aligns with global priorities such as Good Health (SDG 3), Sustainable Cities (SDG 11), and Climate Action (SDG 13). |
No, we do not currently identify as part of an underrepresented group. However, we are committed to building an inclusive and diverse team as we scale, ensuring equal opportunity across backgrounds and disciplines. |
NA |
checked |
| Mar 31, 2026 @ 2:23 AM |
Prakash Chand Jain |
p9340297@gmail.com |
https://www.linkedin.com/in/prakash-chand-jain-coder015675328 |
https://ever-news-flow.lovable.app/ |
9340297104 |
Founder & Full-Stack AI Developer – I am a Computer Science student with strong experience in building scalable web and mobile applications using React.js, Node.js, and cloud technologies. I completed several real-world projects which included an AI-powered learning platform and an event booking system and a civic issue-solving platform. My expertise lies in integrating AI into practical applications, focusing on real-time problem solving, user-centric design, and scalable architectures. I developed essential execution and product-building abilities through my hackathon participation which required me to create valuable solutions within strict time constraints. |
My work as a solo founder has led me to create and develop various projects for two years. My participation in hackathons and my work on personal projects have helped me develop strong execution skills which enable me to transform ideas into functional products. |
one |
My biggest strength is execution speed combined with strong technical depth. I can rapidly build and iterate products from idea to MVP, integrating AI capabilities into real-world applications.
I focus on solving meaningful problems and validating ideas quickly through working prototypes and user feedback. My experience across full-stack development and AI enables me to develop products independently while I need to make fast product iterations based on actual user feedback. |
GAINN |
https://ever-news-flow.lovable.app/ |
Pune |
AI-powered personalized knowledge platform that delivers real-time, relevant insights without information overload |
People today face challenges because the internet offers excessive information yet they cannot discover useful content that matches their needs. Platforms in their current state deliver either standard material to users or they need to perform active searches which results in users experiencing information overload while they retain little and spend excessive time on the process. Intelligent systems need to develop capabilities that enable them to filter, personalize, and deliver accurate information during specific time intervals. |
The AI-powered platform GAINN filters and curates personalized knowledge streams which it delivers to users based on their specific interests and goals and their behavioral patterns.
The platform uses AI models to analyze user preferences which it uses to deliver real-time updates and relevant content and learning materials.
The system transforms information consumption from a passive activity into an intelligent proactive experience which enables users to maintain their knowledge and accelerate their learning process and enhance their decision-making abilities. |
GAINN’s core differentiation lies in its AI-driven personalization engine that continuously learns from user behavior to deliver highly relevant and evolving knowledge streams. Unlike traditional content platforms, GAINN focuses on intelligent filtering and contextual delivery rather than just aggregation. Over time, the system builds a unique data layer around user preferences and interaction patterns, creating a strong personalization moat that improves with usage. This combination of adaptive AI, real-time relevance, and user-specific learning paths makes the platform increasingly valuable and difficult to replicate. |
two |
one |
Primary: Students, developers, and young professionals who consume large amounts of information daily and need personalized, relevant insights. Secondary: Knowledge workers, researchers, and lifelong learners who want to stay updated efficiently without information overload. |
The global digital learning and knowledge consumption market exceeds $300B+, including e-learning, content platforms, and productivity tools. With the rapid growth of AI-driven personalization, this market is expanding significantly. |
The serviceable market includes digitally active learners and professionals using AI-powered tools, estimated at $30B+ across edtech, productivity, and personalized content platforms. |
Initially targeting urban students and professionals in India and global early adopters of AI tools, representing a reachable market of $500M–$1B in the near term. |
The SaaS business operates under a freemium model which provides users access to a free plan that includes restricted personalized content. The premium subscription service delivers advanced AI insights together with enhanced personalization options and productivity capabilities. The company will generate future income through its business-to-business services which will be offered to both enterprises and educational institutions together with its API-based system for accessing the personalization engine. |
The platforms which compete with us operate through Google News and Feedly and standard content aggregation systems. The AI-based competitors of the company include ChatGPT and Perplexity which deliver information to users. The platforms provide search and aggregation functions but GAINN delivers personalized knowledge to users through its active delivery system. |
The first step requires college communities and developer networks to adopt AI technology. The second step requires social media platforms to display content which demonstrates AI-powered insights. The third step enables product development through customer acquisition who use referral and sharing options. The system works together with current operational processes which include daily summaries and notifications. The company will first concentrate on building organic growth through community-based user adoption. |
Phase 1: Launch MVP to early adopters (students, developers) and gather feedback Phase 2: Build strong engagement through personalized daily insights and retention loops Phase 3: Expand through content-driven marketing, partnerships with communities, and referral growth Phase 4: Scale globally with AI-driven personalization as the core differentiator |
To build a global AI-powered knowledge layer that intelligently connects users with the most relevant information in real time.
GAINN aims to become the default interface for learning, decision-making, and staying informed reducing information overload and enabling smarter, faster thinking at scale. |
Not yet incorporated (currently in early-stage MVP phase, with plans to incorporate as we scale) |
Bootstrapped (self-funded) |
one |
Avinya operates as an AI accelerator which maintains its dedication to mentoring and ecosystem development while providing access to top-tier networks. I want to study the structured development program which experienced founders and IIT alumni created because it offers me the chance to improve both our product and business strategy. Avinya provides GAINN with its essential services because it functions as an early-stage AI startup which helps us achieve our growth targets through its delivery of proper mentorship and investor access. |
AI personalization engine needs to be developed into a more powerful system that can handle increased demand. The company needs to validate its products through user testing which requires customers to use its products. The business model requires improvements while the company needs to establish its path to market. The company needs to create its first customers which will help them attract additional funding. The company needs to create essential capabilities which will help them expand their product to international markets. |
Yes |
one |
one |
GAINN operates through a modular AI framework which integrates recommendation systems together with large language models.
The system utilizes LLMs through APIs to create summaries and understand context and provide customized insights.
The Recommendation Engine evaluates user behavior through their interactions with content to determine which materials should be shown to them.
The backend system operates through Node.js which supports scalable APIs together with vector storage that enables contextual content retrieval.
The system uses a React-based interface to deliver customized content to users in real time.
The system creates personalized experiences which change over time because it continuously learns from user actions. |
GAINN builds a proprietary data layer based on user preferences, interaction patterns, and engagement signals. Over time, this creates a unique dataset of personalized knowledge consumption behavior, which improves recommendation accuracy and user experience. |
The defensibility comes from the continuously evolving personalization layer driven by user-specific data.
As more users interact with the platform, the system becomes better at understanding context, preferences, and relevance, creating a feedback loop that is difficult to replicate without similar data and usage patterns. |
The system uses three types of engagement metrics which include click-through rate and time spent and retention metrics. The system uses user feedback to determine content relevance. The system evaluates its recommendations and summaries through accuracy measurements. The system measures its performance through system latency and response time metrics. The company applies A/B testing methods for its product testing process to enhance both user experience and personalized content delivery. |
The system gathers only essential user activity data which it needs to function. The system uses secure storage methods together with encrypted HTTPS communication and token authentication for its operations. The organization does not disclose any customer information to external entities. The system design provides a ready-to-use compliance solution which follows established data protection regulations.
The organization intends to implement GDPR-like standards as its platform expands to international markets. |
Currently minimal, as GAINN focuses on content personalization rather than sensitive domains.
However, as the platform scales, potential risks around data privacy and AI transparency will be proactively addressed through compliance and governance frameworks. |
The primary obstacles which the organization will face at its 10x operational capacity will involve increased data processing requirements and the need to sustain rapid personalized services and the challenge of building an effective recommendation system.
The solution will be achieved through three specific actions which include first optimizing the infrastructure and second developing enhanced caching solutions and third implementing data pipelines and model-serving systems that can handle larger workloads. |
Currently led by a solo technical founder with strong experience in integrating AI systems into real-world applications.
As the product scales, the plan is to build a dedicated AI/ML team to enhance model development and infrastructure. |
Currently using publicly available and API-accessed data sources, along with user interaction data generated within the platform.
As the product evolves, focus will shift toward building proprietary datasets based on user behavior and engagement. |
Continuous learning from user interactions
Feedback loops to refine recommendations
A/B testing different personalization strategies
Gradual integration of fine-tuned models as data grows
The system is designed to improve as more data and usage signals are collected. |
Both starting with India as an initial market, with a clear vision to scale globally as the product and personalization engine mature. |
|
|
|
|
|
|
| Mar 30, 2026 @ 2:12 PM |
Sri Hari Kolusu |
ceo@innovibemobility.com |
https://www.linkedin.com/in/sriharikolusu |
https://innovibemobility.com/ |
9000148333 |
Founder & CEO |
Single Founder |
1 |
Our core strength lies in deep execution capability backed by strong on-ground EV service experience. As a founder, I have led large-scale service operations across 150+ service centers, giving us a clear understanding of real-world EV challenges.
We combine this operational expertise with a strong problem-solving mindset, enabling us to build practical, scalable solutions rather than just theoretical models. Our ability to quickly test, iterate, and deploy in live environments gives us a significant edge in building a reliable EV service and intelligence platform. |
InnoVibe Mobility India Pvt Ltd |
https://innovibemobility.com/ |
Visakhapatnam |
InnoVibe Mobility is an AI-powered EV health and service platform enabling predictive diagnostics, doorstep maintenance, and fleet optimization. |
EV users and fleet operators face frequent breakdowns, unpredictable battery failures, and lack of reliable service infrastructure. Current solutions are reactive, fragmented, and inefficient, leading to high downtime, increased costs, and low trust in EV adoption—especially in emerging cities. |
InnoVibe Mobility provides an integrated AI-powered EV health platform that combines predictive battery analytics, remote diagnostics, and on-demand doorstep service.
Our system collects vehicle and battery data, applies intelligent algorithms to predict failures before they occur, and enables proactive maintenance scheduling. For fleet operators, we offer a centralized dashboard with insights on vehicle health, usage patterns, and cost optimization.
By bridging the gap between technology and on-ground service execution, we reduce downtime, extend battery life, and significantly improve operational efficiency for EV users and fleets. |
Our defensibility lies in the combination of proprietary EV health intelligence, real-time service integration, and deep operational expertise. Unlike traditional service providers, we are building a data-driven platform that continuously learns from vehicle performance and service history. Our early focus on data collection, predictive models, and integrated service delivery creates strong network effects and high switching costs. Additionally, our founder’s experience managing 150+ service centers enables faster execution and scalable deployment across cities, giving us a significant advantage in building India’s EV service intelligence layer. |
MVP |
Revenue |
Our target customers include EV two-wheeler and three-wheeler owners, EV fleet operators (last-mile delivery, logistics, ride-sharing), and OEM/dealer networks. We primarily focus on high-usage commercial EV fleets and urban EV users who require reliable, fast, and predictive service support. |
₹1.5 Lakh Crore+ |
₹15,000 Crore+ |
₹150–300 Crore |
We operate a multi-stream revenue model: Pay-per-service (diagnostics, repair, doorstep service) Subscription plans for EV owners and fleets Fleet management SaaS (AI-driven dashboard & analytics) Battery diagnostics & lifecycle services OEM and dealer partnerships for service outsourcing |
Traditional EV service centers (fragmented, non-digital) OEM service networks (limited reach and scalability) Emerging EV service startups like ReadyAssist, GoMechanic EV, and Battery Smart (focused on specific segments) Our advantage lies in combining AI-driven diagnostics, predictive maintenance, and on-ground service execution into a unified platform. |
Digital marketing (location-based ads, EV-focused campaigns) On-ground activation (campus drives, EV awareness camps) Fleet partnerships (delivery/logistics companies) OEM and dealer collaborations Referral and subscription-based retention model |
We follow a cluster-based city expansion strategy: Launch in high EV-density zones (starting with Visakhapatnam) Partner with fleet operators for bulk onboarding Deploy doorstep service teams for fast response Build local brand trust through awareness campaigns Scale to adjacent Tier 2 cities using a hub-and-spoke model This approach ensures rapid adoption, lower CAC, and strong operational control. |
To build India’s largest AI-powered EV service and intelligence platform, becoming the default infrastructure layer for EV maintenance, battery health management, and fleet optimization.
We aim to expand across 100+ cities, integrate with OEM ecosystems, and enable a data-driven EV future by reducing downtime, extending battery life, and accelerating EV adoption at scale. |
Private Limited |
1000000 |
Yes |
We are building an AI-powered EV health and service platform and see Avinya as a strong fit due to its focus on AI-driven startups. We are specifically looking to strengthen our predictive diagnostics capabilities, refine our technology stack, and gain access to high-quality mentorship and investor networks. Avinya’s ecosystem can significantly accelerate our transition from an operations-led model to a scalable, AI-first mobility platform. |
Build and validate our AI-based predictive diagnostics engine Launch pilot projects with EV fleet operators Strengthen our technology architecture and data pipeline Achieve strong product-market fit in at least one city Prepare for seed fundraising and investor readiness |
Yes, we are open to both hybrid and virtual participation. |
Yes |
Core engine |
Our AI architecture is built around a data-driven EV health intelligence system. We collect vehicle telemetry, battery parameters (voltage, current, temperature), usage patterns, and service history through IoT integrations and manual diagnostics inputs.
The data pipeline is built using cloud infrastructure (AWS/GCP), with real-time data ingestion and storage. We apply machine learning models (time-series analysis, anomaly detection, and predictive models) using Python-based frameworks such as TensorFlow/PyTorch.
The system generates battery health scores, failure predictions, and maintenance recommendations, which are exposed via APIs to our mobile app and fleet dashboard.
Over time, models improve through continuous learning from field data, enabling higher accuracy in predictions and diagnostics. |
We are building a proprietary dataset combining real-world EV battery performance, failure patterns, service records, and usage behavior—especially from Indian road and climate conditions.
This dataset is difficult to replicate as it is tightly integrated with on-ground service operations and fleet usage. Over time, this creates a strong data moat, enabling more accurate predictive models and localized insights compared to global or generic datasets. |
Our defensibility comes from three layers:
Data Advantage – Continuous collection of real-world EV performance and failure data creates a compounding data moat.
Model Improvement Loop – AI models improve with every service interaction, increasing prediction accuracy over time.
Full-stack Integration – Unlike competitors focusing only on software or service, we integrate AI with on-ground execution, making replication difficult.
This combination of data, learning loops, and operational integration creates high entry barriers and long-term defensibility. |
We evaluate our AI models using both technical and business metrics:
Model Metrics:
Prediction accuracy (battery failure prediction accuracy)
Precision & recall for fault detection
Mean Absolute Error (MAE) for health score predictions
Operational Metrics:
Reduction in vehicle breakdowns
Decrease in service turnaround time
Improvement in battery life
Business Impact Metrics:
Fleet uptime improvement (%)
Cost savings for users
Customer satisfaction and retention rates
Continuous monitoring and feedback loops ensure model reliability and real-world effectiveness. |
We use Post-Quantum cryptopgrahy(PQC) algorithms with Quantum Random number generation (QRNG) to make data en decriptble even with the quantum computers. by using PQC and NIST Compliant QRNG we are already compliant with all the security compliences needed. (NIST FIPS, NIST SP-800,NIST SP-90A,B,C ) |
The primary regulatory dependencies are related to EV standards, battery safety norms, and data protection laws. Changes in battery compliance (BIS, AIS standards) or OEM restrictions on data access could impact operations.
We mitigate this by maintaining OEM partnerships, adhering to certified diagnostic processes, and designing our platform to be compliant with evolving regulatory frameworks. Our model remains adaptable to policy changes in the EV ecosystem. |
At 10x scale, potential challenges include data pipeline latency, model performance drift, and infrastructure load.
To address this, we are building a scalable cloud-native architecture with distributed data processing, auto-scaling infrastructure, and model monitoring systems. We also plan to implement MLOps pipelines for continuous training, validation, and deployment to ensure consistent performance at scale. |
Currently, we operate with a lean team supported by external AI/ML consultants and technology partners.
Our focus is on building in-house capabilities in machine learning, data engineering, and EV diagnostics. During the accelerator, we plan to strengthen our core AI team with dedicated engineers specializing in predictive modeling, IoT data processing, and fleet analytics systems. |
We are currently building and using a combination of proprietary and structured datasets, including EV battery parameters, vehicle usage data, service records, and diagnostic reports collected from our operations.
This data is owned by us (with user consent) and forms the foundation of our AI models. Additionally, we may use publicly available and licensed datasets for benchmarking and model validation where required. |
We follow a continuous learning approach:
Real-time data collection from vehicles and service interactions
Feedback loops from diagnostics and repair outcomes
Periodic model retraining and validation
A/B testing of model performance improvements
Monitoring for drift and recalibration of models
This ensures our models become more accurate and reliable as data volume and diversity increase. |
We are initially focused on India, given its rapidly growing EV market and unique operating conditions.
Our long-term vision is to expand into global emerging markets with similar EV adoption patterns, particularly in Southeast Asia, the Middle East, and Africa. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/m7U9p7m080ZU-Predict-Prevent-Power-EV-Performance-30-March-2026.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/m7U9p7m080ZU-Predict-Prevent-Power-EV-Performance-30-March-2026.pdf |
http://na |
Yes, InnoVibe Mobility is a mission-driven startup focused on accelerating EV adoption by solving one of its biggest barriers—reliable and accessible service infrastructure.
By enabling predictive maintenance and reducing unexpected breakdowns, we improve vehicle efficiency, extend battery life, and lower the total cost of ownership for EV users. This directly supports sustainable mobility and reduces environmental impact.
Additionally, our doorstep service model creates local employment opportunities for EV technicians and brings structured service access to Tier 2 and Tier 3 cities, contributing to both economic and environmental impact. |
NA |
NA |
checked |
| Mar 30, 2026 @ 11:39 AM |
dff |
gouravabtikwalal2@gmail.com |
https://www.linkedin.com/feed/ |
http://www.f.com |
4565454335 |
sdsd |
sdsdd |
2 |
sdsd |
wsdsad |
sad |
dsad |
sad |
sds |
sad |
add |
MVP |
Revenue |
sads |
44 |
44 |
33 |
sd |
dsd |
sdsd |
sds |
dsd |
dsd |
43 |
No |
ffd |
sdsd |
sd |
Yes |
Supporting feature |
sd |
sad |
sdsd |
ds |
sad |
ds |
ds |
dsa |
sdsd |
sds |
sdd |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/ooTkCxaZalfo-pitch_deck_guide.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/ooTkCxaZalfo-pitch_deck_guide.pdf |
https://docs.google.com/document/d/1q0IR3lwrsM9cM_s11RJfYKNMrHHpr_m6/edit |
szdxzd |
sdsd |
sd |
checked |
| Mar 30, 2026 @ 12:01 AM |
Suchandana Ghoshal, Sougam Maity |
avistha2024@gmail.com |
http://www.linkedin.com/in/sougam/%20 |
https://avistha.com/ |
09932242598 |
Founder and CTO, Founder and CEO |
We are working for last 3.5 years. We meet in a Company and working in a Team. After that we joined CLiMB fellowship at IIT Kharagpur and working in a team. We completed the Biodesign and developed the product for 2 years and then for last 1 Year we are working on the same product. |
two |
Suchandana Ghoshal : Founder and CTO
CLiMB Fellow at IIT Kharagpur and having 8 Years experience as Healthcare, designer & manager in Renewal Power Sector, Control System, SCADA, PLC, product development, Mechanical design, IOT etc. She has completed BTech & MTech in Electrical and Power Engg. She is expert in the hardware sides and very good in product design and development.
Sougam Maity : Founder & CEO
CLiMB Fellow at IIT Kharagpur with around 17+ Years experience in Healthcare, Startup, IT & IOT domain, AI and NLP etc. Member NASSCOM, Tie, ICC. Completed BTech in CSE from MAKAUT. He has connections and good in marketing, Software and IOT. |
Punyaha Heathcare Solution Pvt Ltd. |
https://avistha.com/ |
Bagnan |
AI & IOT -powered, eco-friendly medical capsule that bridges the last-mile healthcare gap by transforming any local vehicle into an ICU-level emergency transport for safety, cozy and dignity of the patient during relocation. |
1) ALS Ambulance Scarcity 2) The "Last-Mile" Barrier 3) Blind Transit in local vans as doctors cannot monitor the patient during transit 3) Secondary Trauma : Transporting a heart attack /trauma/ heat stroke patient in a bumpy cart, hot van and excessive vibration causes "secondary injuries." 5) Resource Mismatch |
Avistha is an AI & IOT -powered, eco-friendly medical capsule that bridges the last-mile healthcare gap by transforming any local vehicle into an ICU-level emergency transport for safety, cozy and dignity of the patient during relocation.
Environment & vibration control, stretchable. Non-invasive monitoring & analysis of the patient & their medical data through mobile app during relocation.
It is like a "ICU-in-a-Box.", Vehicle-Agnostic : can be placed in boat, van, goods toto, helicopter etc, Vibration Control ( 3 Layer and MPU6050) and Solar-Powered Eco-friendly Cooling (Peltier and PCM), The IoT Telemetry provides continuous data stream, allowing trained staff to "Remote-Supervise" the transit & provide early instructions, The App-Integrated Resource Mapping solves the Coordination Gap. It syncs with hospital backends to ensure the patient is routed to the right hospital bed with the right equipment ready.
The features are as follows :
Eco-Friendly Temperature & Humidity Control
3 Layer of Vibration Control
AI based camera monitoring
IOT and Sensors for Health Data and Capsule monitoring
Essential utilities like Saline Stand, Mini Oxygen Cyllinder |
Unique product for safety of the patient, Can be placed in any vechile, Ecofriendly cooling, Vibration control and Health Data Retrieval |
two |
Pilots |
Patients of Rural and Semi-urban areas |
14100000 |
10000000 |
20000 |
Selling at 90,000 INR or Rent at Rs 500 per day |
Red Health, JCBL, Modularzz, MedCab |
Social Media Marketing, Seminars etc for B2C and Meeting and Collaboration with Channel Partners |
B2B connect with Medical Clinics, CLub etc and create Channel Partners |
Our vision is to revolutionize the rural & semi-urban healthcare through smart, affordable and eco-friendly systems with fast & adequate information during emergency for safety and dignity of the patients.
Our target is to add all the semi-urban and rural areas of India, onboard the best available hospitals and clinics and have Channel partners in all places to help a patient reach safe in to the hospital with dignity. |
Incorporated |
CLiMB fellowship support from ICMR and IIT Kharagpur |
one |
Funding |
Funding and connects |
Yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 28, 2026 @ 5:07 PM |
RAVIKUMAR CHAVADI |
kvbgreenenergies@gmail.com |
https://www.linkedin.com/company/kvb-green-energies/posts/?feedView=all |
http://www.kvbgreenenergies.com |
09545529950 |
The project is led by Ravikumar Chavadi, an experienced engineer with over 20 years in industrial system development. The team includes AI developers, embedded engineers, and mechanical designers working together to build an AI-powered thermal grading system. The team combines strong domain knowledge with emerging AI capabilities to develop a scalable solution. |
We came together through collaborative work on real-world engineering and AI projects under KVB Green Energies. Over the last few months, we have been actively working together on this solution, developing the initial AI pipeline and system concept. This has helped us build a strong understanding of each other’s strengths and a clear execution roadmap. |
1 |
Our biggest strength as a team is the combination of deep industry experience and emerging AI capabilities. With over 20 years of hands-on experience in building and deploying real-world engineering systems through KVB Green Energies, we understand practical challenges at the ground level. At the same time, our team is actively developing AI-based solutions, bringing innovation into traditional systems. This combination allows us to not only design intelligent solutions but also implement them effectively in real industrial environments. |
KVB Green Energies |
www.kvbgreenenergies.com |
Hubli-Dharwad |
“We are building an AI-powered thermal intelligence system for real-time fruit grading, enabling accurate detection of ripeness and defects beyond what the human eye can see.” |
India is one of the largest producers of mangoes, but a significant portion of the produce is lost or undervalued due to inefficient grading systems. Current methods are mostly manual, inconsistent, and depend on visual inspection, which cannot accurately determine internal ripeness or detect hidden defects. As a result, good quality fruits are often misclassified, and defective ones may pass through, leading to economic losses for farmers, reduced export quality, and increased food wastage. |
We are developing an AI-powered mango grading and sorting system that uses thermal and visual imaging to assess both internal and external quality of fruits in real time. The system captures thermal images to analyze internal heat patterns, which indicate ripeness and hidden defects, while RGB images are used to evaluate surface features such as color and visible damage.
Using AI models, we process this data to classify mangoes based on two parameters: ripeness and quality. These are then combined to assign a final grade. The system is integrated with a conveyor mechanism for continuous operation and can be connected to automated sorting systems for large-scale industrial deployment. |
Our solution is unique because it goes beyond traditional visual grading by introducing thermal intelligence to analyze the internal condition of fruits. While existing systems rely only on RGB images, we combine thermal and visual data to assess both ripeness and hidden defects, which are not visible to the human eye. Additionally, we have developed a dual-parameter grading approach by separating ripeness and quality into independent axes, allowing for more accurate and industry-relevant classification. Our system is further strengthened by a self-calibrating thermal mechanism, which ensures consistent performance across different environmental conditions. This combination of multi-modal sensing, intelligent grading logic, and real-time industrial integration makes our solution both unique and difficult to replicate. |
MVP |
Users |
Our primary target customers are mango pack houses, exporters, and large-scale fruit processing units that require consistent and high-quality grading. These businesses need reliable systems to improve sorting accuracy, maintain export standards, and reduce post-harvest losses. Additionally, we target institutional buyers and aggregators who handle large volumes of fruits and require automated solutions for efficiency. In the long term, the solution can be extended to other fruits and agricultural supply chains, expanding the customer base significantly. |
India produces a huge amount of mangoes every year. Many companies need grading systems to sort them properly. So the total market for our solution in India is very large, around ₹2,000 to ₹5,000 crore, and it can grow even bigger when we expand to other fruits. |
Our SAM includes mango pack houses and exporters in India who handle large quantities and need grading systems. This market is around ₹500 to ₹1,000 crore. |
In the next few years, we plan to install our system in around 20 to 50 places like pack houses and exporters. This will give us a business of around ₹20 to ₹50 crore. |
Our revenue model is a combination of hardware sales and recurring service revenue. We generate primary revenue through the sale and installation of AI-based grading systems to pack houses, exporters, and processing units. In addition, we offer a subscription-based analytics platform that provides insights such as grading reports, quality trends, and performance monitoring. We also generate recurring revenue through annual maintenance contracts (AMC), system upgrades, and ongoing support services. In the long term, we plan to leverage the data collected to provide advanced advisory services and expand into multi-crop grading solutions. |
Our competitors are mainly traditional grading machines and AI systems that use normal cameras. These systems can only check the outside of the fruit. Our system is different because we also check the inside using thermal data, which gives better accuracy. |
Our customer acquisition strategy is primarily driven by direct industry engagement and pilot deployments. We leverage our existing network through KVB Green Energies to connect with institutions, food processing units, and large-scale users. We plan to onboard early customers through pilot projects, demonstrating clear value in terms of accuracy, efficiency, and cost savings. Once validated, we will scale through referrals, partnerships with industry stakeholders, and targeted outreach in major mango-producing regions. In addition, we will use digital channels and industry events to increase visibility and reach potential customers. |
Our go-to-market strategy follows a phased approach. In the initial phase, we focus on pilot deployments with selected mango pack houses and exporters through our existing industry network. This allows us to validate performance, demonstrate value, and build strong case studies. In the next phase, we scale through direct B2B sales targeting high-volume users in key mango-producing regions. We leverage referrals, partnerships, and industry connections to accelerate adoption. In the long term, we expand to additional fruits and regions, while building a data-driven platform with recurring revenue through analytics and services. |
Our long-term vision is to build an AI-powered post-harvest intelligence platform for agriculture that enables accurate quality assessment, reduces food wastage, and improves value across the supply chain.
Starting with mango grading, we aim to expand our solution to multiple fruits and crops, creating a scalable system that combines thermal and visual intelligence for real-time decision-making.
In the future, we plan to leverage the data collected to provide predictive insights such as shelf-life estimation, quality forecasting, and advisory support for farmers and industries, ultimately transforming how agricultural produce is graded, valued, and managed. |
We are developing this project under KVB Green Energies, which is already an established company in 2021 |
We have raised ₹10 lakhs in funding from ICAR–Pusa Krushi, New Delhi, for the development of solar tunnel dryer systems under KVB Green Energies. This funding supported the development and validation of our renewable energy solutions. Building on this experience, we are now expanding into AI-based agricultural solutions, leveraging our existing expertise in engineering and deployment. |
Yes |
We are applying to Avinya to accelerate the development and deployment of our AI-based agricultural solution. Beyond funding, we are particularly interested in Avinya’s mentorship, ecosystem, and industry connections, which can help us refine our technology, improve our go-to-market strategy, and scale effectively. We believe Avinya’s focus on AI-driven innovation aligns strongly with our vision of building a data-driven intelligence platform for agriculture, and we see this as an opportunity to transform our prototype into a scalable, real-world solution. |
During the program, we aim to achieve three key milestones. First, we want to strengthen our AI model by building a high-quality, balanced dataset and improving accuracy using advanced techniques. Second, we aim to develop a fully functional real-time prototype by integrating thermal and RGB cameras with a conveyor-based system for industrial use. Third, we plan to conduct pilot deployments with industry partners to validate performance, gather feedback, and prepare for large-scale commercialization. Overall, our goal is to transition from a prototype to a deployable and scalable product. |
yes |
Yes |
Core engine |
Our AI architecture is designed as a multi-stage pipeline for real-time grading. The system begins with data acquisition using thermal and RGB cameras mounted over a conveyor system.
The captured images are processed using a YOLOv8-based segmentation model to detect and isolate individual mangoes. Once segmented, we extract thermal features such as mean temperature, variance, and hotspot characteristics to represent ripeness and quality.
These features are then passed into a machine learning model (currently Random Forest) that classifies mangoes based on ripeness (R-axis) and quality (Q-axis). The outputs are combined to assign a final grade.
The system is optimized for real-time inference and integrated with an automation layer for sorting. In future iterations, we plan to incorporate deep learning models for end-to-end feature learning and improve accuracy through multi-modal data fusion. |
Our key proprietary advantage lies in the thermal imaging dataset we are building for mango grading, which includes labeled data across ripeness and defect categories. This type of dataset is not widely available, especially for thermal-based analysis.
In addition, we are developing a structured R–Q grading framework that maps thermal features to real-world grading standards, creating a unique combination of data and domain knowledge.
As we scale, continuous data collection from real-world deployments will further strengthen our dataset, creating a strong data-driven moat that improves model accuracy and makes the system difficult to replicate. |
Our AI component is defensible primarily due to our proprietary thermal dataset and domain-specific feature engineering. Unlike generic vision models, our system relies on thermal patterns linked to ripeness and internal defects, which require specialized data collection and labeling that is not easily available.
Additionally, we use a structured R–Q grading framework that separates ripeness and quality into independent parameters, creating a more interpretable and industry-aligned model.
As we deploy in real-world environments, continuous data collection and feedback loops will further improve our models, creating a strong data-driven advantage that becomes increasingly difficult for competitors to replicate. |
We evaluate our model performance using both classification accuracy and reliability metrics. For grading performance, we measure accuracy, precision, recall, and F1-score across different ripeness and quality classes to ensure balanced performance.
For the segmentation stage (YOLOv8), we use metrics such as Intersection over Union (IoU) and mean Average Precision (mAP) to evaluate detection and masking quality.
In addition, we assess system reliability through consistency checks under different environmental conditions, including temperature variations and conveyor speeds. We also validate predictions using ground truth verification by physically inspecting sample fruits to ensure real-world accuracy.
This combination of model metrics and real-world validation ensures both accuracy and robustness of the system. |
We take data privacy and security seriously, even at the early stage of development. Our system primarily processes image-based data of agricultural produce, which does not involve sensitive personal information.
All data collected is securely stored with controlled access, and we follow best practices such as encryption and access management to prevent unauthorized use.
For deployments, we ensure that customer data remains confidential and is used only for improving system performance. We also design our system to comply with relevant data protection guidelines and are prepared to adopt stricter compliance standards as we scale. |
At present, our solution has minimal regulatory risk as it primarily involves non-invasive imaging and AI-based analysis of agricultural produce, without handling sensitive personal data or restricted materials.
However, as we scale, potential regulatory considerations may include compliance with food quality standards, industrial equipment safety norms, and data handling guidelines.
We are mindful of these aspects and plan to align with relevant standards and certifications required for deployment in food processing and export environments. Our system is designed to be adaptable to regulatory requirements, ensuring smooth adoption across different markets. |
As we scale 10x, the primary challenges will be related to data variability, real-time processing, and system robustness.
First, increased deployment across different environments will introduce variations in temperature, lighting, and mango varieties, which may affect model accuracy if not properly handled.
Second, scaling to high-throughput operations will require optimization of our AI pipeline to maintain low latency and real-time performance.
Third, managing large volumes of data and ensuring consistent labeling and quality will become critical.
To address these challenges, we are focusing on building a robust data pipeline, implementing calibration mechanisms for environmental variations, and optimizing models for edge deployment. We also plan to continuously retrain our models using real-world data to improve performance at scale. |
Yes, we have an in-house AI/ML team consisting of engineers and students working on computer vision, machine learning, and embedded systems. The team is actively involved in developing the core AI pipeline, including YOLOv8-based segmentation, thermal feature extraction, and classification models for ripeness and quality assessment.
The team has hands-on experience in working with image data, model training, and real-time system integration. They are also exploring multi-modal AI approaches by combining thermal and RGB data for improved accuracy.
This is complemented by strong industry and engineering expertise from KVB Green Energies, enabling us to build and deploy practical, scalable solutions. |
Currently, we are primarily using a self-collected dataset of thermal images of mangoes, which we have generated during the prototype stage. The dataset initially started with a limited number of samples and has been expanded using data augmentation techniques.
This dataset is owned by us and is being continuously improved with better labeling across ripeness and quality categories.
As we scale, we plan to build a larger, high-quality proprietary dataset through real-world data collection during pilot deployments. We are not dependent on external licensed datasets for our core model development. |
We plan to continuously improve model performance through a data-driven feedback loop. As we deploy the system in real-world environments, we will collect more diverse and high-quality data across different mango varieties, conditions, and defect categories.
This data will be regularly annotated and used to retrain and fine-tune our models, improving accuracy and generalization.
We will also monitor model performance using key metrics and identify failure cases, which will be specifically targeted for improvement. Additionally, we plan to incorporate advanced techniques such as multi-modal learning (thermal + RGB) and deep learning-based feature extraction to further enhance performance over time. |
We are building for both India and global markets. We are initially focusing on India, as it is the largest producer of mangoes and provides a strong opportunity for validation and early adoption.
Once the solution is proven and scaled locally, we plan to expand to global markets, particularly regions with high fruit production and export demand. The core technology is scalable and can be adapted to multiple crops and geographies, making it suitable for international deployment. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/g1jKGmPvUQvt-AI-powered-thermal-intelligence-for-real-time-fruit-grading.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/g1jKGmPvUQvt-AI-powered-thermal-intelligence-for-real-time-fruit-grading.pdf |
https://1drv.ms/v/c/76E621E6E8A4E36F/IQDKJS_fbTGATICTWss_pVIrARIlOGS43b1XzIUh_nPu2Og?e=VivVpw |
Yes, we are building a mission-driven and impact-focused startup. Our goal is to reduce post-harvest losses, improve quality assessment, and increase value realization for farmers and agricultural supply chains.
A significant portion of agricultural produce is lost or undervalued due to inefficient grading systems. By introducing AI-based grading, we aim to bring transparency, consistency, and efficiency into the system. This helps farmers and suppliers receive fair pricing for high-quality produce while reducing food wastage.
In the long term, we aim to build a data-driven platform that not only improves grading but also provides insights such as shelf-life prediction and quality trends, contributing to a more sustainable and efficient agricultural ecosystem. |
Yes |
Self / Direct Application |
checked |
| Mar 28, 2026 @ 1:20 PM |
Pushpa Latha, Bhaskar Dudhi Venkatesh |
operations@hcsbharath.com |
https://www.linkedin.com/in/pushpa-latha-0a077821a/ |
http://www.hcsbharath.com |
97390002734 |
Pushpa Latha — Founder & Managing Partner. Leads product vision, operations, and social impact strategy. Brings experience in program leadership and social enterprise operations. Bhaskar Dudhi Venkatesh — Co-founder & Technology Lead. Leads AI systems, enterprise platform development, and program delivery. Has built scalable AI/ML platforms for large enterprise and government clients. |
We first met in 2010 as co-workers in the technology industry and have been working together for over 15 years. Our professional collaboration evolved into a personal partnership — we got married in 2018. In 2024, we founded Hamrith Consulting Service LLP, combining our complementary strengths to tackle a problem we both felt deeply about. Pushpa brings the product vision and operational leadership, while Bhaskar brings enterprise AI/ML expertise. Having worked together across both professional and personal contexts for over a decade and a half, we share a deep trust, aligned values, and an efficient working dynamic that allows us to move fast and make decisions decisively. |
All |
Our 15+ years of working together — first as colleagues, then as life partners, and now as co-founders — gives us an unmatched level of trust, alignment, and speed in decision-making. We complement each other perfectly: Bhaskar brings deep expertise in building scalable AI/ML platforms for enterprise and government clients, while Pushpa brings product vision, operational leadership, and the ability to navigate complex stakeholder ecosystems like state governments, ICDS, and NGOs. We don't just divide tasks — we think together, challenge each other, and move decisively. This combination of technical depth and ground-level execution capability is rare in India's ECD space, and it's what allows us to build the technology and take it to market ourselves. |
Hamrith Consulting Service LLP |
www.hcsbharath.com |
KGF, Karnataka |
HCS ECD-AI Connect is India's first AI-powered platform for early detection of developmental delays in children aged 0–6, designed to work within the ICDS ecosystem. |
India has over 160 million children aged 0–6 and 1.4 million Anganwadi workers, yet there is zero AI-based developmental tracking in the public childcare system. The ICDS currently tracks only nutrition and survival — not developmental intelligence. As a result, developmental delays are identified only at age 6–8, far beyond the critical 0–3 year intervention window when 90% of brain development occurs. Late detection costs 7–10x more to remediate and leads to irreversible developmental gaps. Anganwadi workers are overloaded with paper-based manual records, making real-time tracking impossible. India essentially loses the most important developmental window in a child's life because no system is designed to catch it. |
ECD-AI Connect is an AI-powered platform that flags developmental delays before age 3 using longitudinal child data — not one-time snapshots. It auto-generates age-appropriate, personalized activity plans for each child and caregiver. The platform is voice-first and multilingual (Hindi, Kannada, Telugu, Tamil, Marathi, Urdu, and more), purpose-built for low-literacy Anganwadi workers. It includes government-ready dashboards for district- and state-level analytics, is ICDS-interoperable, and deployable across states. The platform is mobile-first and offline-capable, enabling frontline workers to screen, track, and intervene — all from a single app. |
ECD-AI Connect is the only AI infrastructure in India purpose-built for early childhood developmental tracking within the ICDS framework. Our defensibility rests on four pillars: first, there is no direct competitor offering AI-driven developmental screening in India's public ECD space — we have clear first-mover advantage. Second, our ICDS integration readiness is a significant moat, as building for government interoperability requires deep domain understanding that generic EdTech or health-tech platforms lack. Third, our voice-first, multilingual design (8 languages) makes us uniquely accessible to low-literacy frontline workers — a user segment most tech solutions ignore. Fourth, as we scale across states and accumulate longitudinal child development data, our AI models become increasingly accurate and harder to replicate, creating a compounding data moat. |
MVP |
Pilots |
We serve four customer segments: B2G (State Governments) — Anganwadi digital transformation and ICDS upgrade through district and state-level licensing contracts. B2B (Private Preschools) — SaaS subscription model targeting 15,000+ preschools nationwide for AI-powered developmental tracking. B2NGO (NGOs & CSR Programs) — impact measurement, pilot programs, and developmental data access for organizations working in early childhood development. B2C (Caregivers) — a parent-facing developmental tracking app for direct consumer use. |
₹2,800+ Crore — India's total ECD, GovTech, and EdTech market covering 160M children (0–6), 1.4M Anganwadi workers, and the ICDS budget of ₹20,000 Crore (2024–29). |
₹800+ Crore — digitisation-ready segments including state government pilots, 15,000+ private preschools, and NGO/CSR programs targeting early childhood development. |
₹80 Crore — 5-year serviceable target through 5 state pilots, 2,000 preschools, 50 NGOs, and district licensing contracts by FY2030. |
Our revenue model is multi-stream: SaaS subscriptions for state governments (district/state-level licensing contracts for ICDS integration), SaaS subscriptions for private preschools (per-institution annual licensing), analytics and data subscriptions for NGOs and CSR partners (impact measurement and research data access), and a freemium-to-paid model for the parent-facing caregiver app. Our revenue target is ₹7 Crore by FY2030 with a minimum net profit margin of 7.5%. |
There is no direct AI infrastructure competitor in India's ECD space. The two indirect alternatives are: ICDS (Manual) — the existing government system which tracks nutrition and survival but has no AI developmental detection, no predictive insights, no longitudinal tracking, and no digital tools for frontline workers. EdTech Apps — consumer-facing education apps that offer partial milestone tracking and caregiver guidance but lack AI-driven risk detection, government dashboards, ICDS integration, and voice-first low-literacy design. ECD-AI Connect is the only platform combining all these capabilities in a single ICDS-compatible solution, giving us clear first-mover advantage. |
We acquire customers through five channels: government tenders and direct engagement with state ICDS departments, direct business development outreach to private preschool chains, CSR program partnerships with top corporates, NGO referrals and impact-driven partnerships, and incubator and startup network connections (including JITO and similar ecosystems). Our customer acquisition cost starts at ₹0 for grant/CSR-funded government pilots, moves to ₹800–1,200 per institution during the scale phase, and drops to ₹400 per institution at national scale. |
Our GTM is structured in three phases. Phase 1 — Pilot (2026): 5 district Anganwadi pilots in Karnataka and Andhra Pradesh, 3 NGO partnerships for impact measurement, and screening of 500 children, all grant/CSR funded at zero CAC. Phase 2 — Scale (2027–28): state-level government SaaS contracts, onboarding of 500 private preschools, CSR partnerships with top 20 corporates, at a CAC of ₹800–1,200 per institution. Phase 3 — National (2029–30): expansion to 5+ states and 2,000+ preschools, national ICDS integration partnership |
To make ECD-AI Connect India's early childhood intelligence infrastructure — the national standard for AI-powered developmental tracking within ICDS. If India can track payments in real time, we can track child development in real time. Our vision is to ensure that no child's developmental delay goes undetected during the critical 0–3 year window. Long-term, we aim to expand beyond India to become the developmental AI platform for the Global South, replicating across Africa and Southeast Asia, addressing a $7B+ global opportunity in AI for child health and development. |
Hamrith Consulting Service LLP, incorporated in 2024 |
15 Lakhs (promoter equity invested by founders). No external funding raised to date. |
Yes |
We are applying to Avinya because ECD-AI Connect is at a critical inflection point — our platform is built and validated, and we now need the right ecosystem support to move from MVP to scaled deployment. Avinya offers exactly what we need at this stage: strategic mentorship from experts who understand the intersection of deep-tech and public systems, access to pilot partnerships with state governments and NGOs, connections to investors aligned with social impact and health-tech, and technical guidance to refine our AI engine for field-level performance. We are not just looking for funding — we are looking for a launchpad that can help us become India's early childhood intelligence infrastructure, and Avinya's focus on translating innovation into real-world impact aligns perfectly with our mission. |
We want to achieve four key outcomes during the programme. First, secure 2–3 formal pilot partnerships with state ICDS departments or NGOs to deploy ECD-AI Connect in real Anganwadi settings and validate our screening engine with 500+ children. Second, refine our AI engine (v2.0) and mobile app with technical mentorship — particularly around offline-first architecture, voice-first UX for low-literacy users, and predictive model accuracy. Third, develop a robust commercialisation roadmap with mentorship on government procurement processes, SaaS pricing for B2G and B2B segments, and IP filing strategy. Fourth, build investor relationships to prepare for our next funding round, with a clear pitch backed by field-validated traction and impact data from the pilots. |
Yes — we are fully open to hybrid or virtual participation and can adapt to the programme's format. |
Yes |
Core engine |
ECD-AI Connect is built on a full-stack AI architecture with AI as the core engine, not just an add-on. The frontend is React 18 with Tailwind CSS, communicating with a FastAPI (Python) backend powered by Motor (async MongoDB driver). The AI layer uses OpenAI GPT-5.2 for developmental screening — analyzing caregiver and frontline worker responses against CDC/WHO developmental milestones across 6 domains (cognitive, motor, language, social-emotional, adaptive, health) to generate risk assessments (Low/Moderate/High) and personalized intervention plans. Our flagship feature is a multi-dimensional speech analysis engine that evaluates children's speech across 4 scientific domains: Linguistic Analysis (vocabulary count, Mean Length of Utterance, grammar complexity, age-normed percentile ranking), Acoustic Analysis via librosa (speech rate, pause patterns, pitch variability), Pragmatic Analysis (turn-taking, topic maintenance, narrative ability, social communication), and Behavioral Analysis (engagement level, attention span, emotional expression). Speech-to-text is handled by OpenAI Whisper and text-to-speech by Google Cloud TTS, supporting 8 Indian languages. The system runs on a hierarchical multi-tenant architecture (National > State > District > Taluk > Village) with JWT-based authentication and role-based access control, deployed on Kubernetes and Docker. |
Our proprietary data advantage is threefold. First, we are building India's first longitudinal developmental dataset for children aged 0–6, capturing multi-dimensional speech and screening data across 8 Indian languages — this dataset simply does not exist anywhere else. Second, our age-normed developmental benchmarks calibrated for Indian children across linguistic and cultural contexts are proprietary — existing global tools like ASQ or MCHAT are not normed for India's diverse population. Third, as we scale across districts and states through our hierarchical multi-tenant architecture, every screening enriches our models with region-specific, language-specific developmental patterns, creating a compounding data moat that no competitor can replicate without deploying at similar scale. |
Four layers of defensibility. First, our multi-dimensional speech analysis engine screens across 4 scientific domains simultaneously (linguistic, acoustic, pragmatic, behavioral) — replicating 80% of a certified speech-language pathologist's initial screening. Single-domain tools miss conditions that our cross-validation catches, reducing false negatives. Second, our 8-language speech analysis with native script support and language-specific developmental norms is extremely hard to replicate — each language requires its own acoustic models, vocabulary benchmarks, and MLU norms. Third, our ICDS-interoperable hierarchical architecture (National > State > District > Taluk > Village) with strict data isolation is purpose-built for India's government infrastructure — a design choice that takes deep domain knowledge and cannot be bolted onto generic platforms. Fourth, our growing longitudinal dataset of Indian children's developmental trajectories across languages and regions becomes increasingly valuable and irreplicable over time. |
We track several key metrics: high-risk identification accuracy (target >85%, currently calibrating against clinical assessments), screening completion rate (target >80%), average screening time (target <10 minutes, currently ~7 minutes), false negative reduction through multi-domain cross-validation, admin response time to high-risk alerts (target 50). For the speech analysis engine specifically, we evaluate domain composite scores using weighted scoring across all 4 dimensions (linguistic, acoustic, pragmatic, behavioral), track longitudinal progress trends across multiple sessions per child, and benchmark against CDC/ASHA developmental milestones for age-normed accuracy. |
Data privacy and security are foundational to our architecture. All personal data is encrypted at rest and in transit. Formal consent is required before any screening is conducted. We maintain a complete audit trail logging all user actions (500+ audit log records and growing). Access is controlled through strict role-based access control (RBAC) with JWT-based token authentication with expiry — a Super Admin sees national data, a State Admin sees only their assigned state, a District Admin sees only their districts, and frontline workers see only their assigned children. Our multi-tenant architecture ensures state-wise data separation. We are building toward compliance with India's Digital Personal Data Protection Act (DPDPA) and plan to implement consent PDF downloads and enhanced data governance as we scale. |
The primary regulatory considerations are India's Digital Personal Data Protection Act (DPDPA 2023), which governs how we collect, store, and process children's personal and health data — we are designing for compliance from the ground up. Additionally, since our AI performs developmental screening (not diagnosis), we need to clearly position the platform as a screening and decision-support tool, not a medical diagnostic device, to avoid classification under medical device regulations. Integration with ICDS will require adherence to government data sharing protocols and security standards. We do not foresee these as blockers — rather, our ICDS-interoperable architecture and consent management are being built to meet these requirements proactively. |
At 10x scale, the key stress points would be: first, API costs — our reliance on OpenAI GPT-5.2 and Whisper means per-screening costs could become significant at millions of screenings, requiring us to explore fine-tuned open-source models or negotiate enterprise API pricing. Second, MongoDB performance under concurrent load from multiple states and districts would need sharding and read-replica optimization. Third, our speech processing pipeline (audio upload, Whisper transcription, librosa analysis, GPT analysis) is sequential and compute-heavy — at scale, this needs to be parallelized with queuing systems. Fourth, offline mode is still in development (3–5 days effort) and is critical for rural deployment — without it, 10x scale across India's connectivity-challenged regions would fail. Fifth, WhatsApp integration is currently blocked pending Meta Business API credentials but is essential for reaching caregivers at scale. |
Yes. Bhaskar Dudhi Venkatesh, our Co-founder & Technology Lead, has extensive experience building scalable AI/ML platforms for large enterprise and government clients. He architected the entire ECD-AI Connect platform — including the multi-dimensional speech analysis engine, the GPT-5.2 integration for developmental screening, the hierarchical multi-tenant data architecture, and the real-time risk notification system. As we scale, our immediate hiring plan includes 3 additional AI engineers to strengthen the team across NLP, speech processing, and model optimization. |
We use a combination of licensed and proprietary datasets. Licensed/public sources include CDC developmental milestones, WHO growth standards, and ASHA (American Speech-Language-Hearing Association) speech-language benchmarks for age-normed developmental references. Our proprietary datasets include real-world screening data from our Andhra Pradesh pilot covering 1,000 screened children, along with multi-dimensional speech analysis sessions, longitudinal child profiles, and age-normed benchmarks calibrated for Indian children across 8 languages — all generated through our platform and fully owned by HCS. This pilot dataset from 1,000 children gives us a validated foundation for model training and calibration specific to Indian developmental patterns. As we scale through additional pilots across Karnataka and other states, this proprietary dataset will grow into India's largest AI-ready developmental dataset for children aged 0–6. |
Through four mechanisms. First, longitudinal data accumulation — every screening session enriches our dataset with real-world developmental patterns across ages, languages, and regions, enabling continuous model retraining and accuracy improvement. Second, clinical validation loops — we plan to benchmark our AI screening results against certified developmental pediatrician and SLP assessments during pilot programs to calibrate accuracy and reduce false negatives. Third, feedback integration — frontline workers and caregivers provide ground-truth feedback on AI-generated intervention plans, which feeds back into model refinement. Fourth, expanding from GPT-5.2 API dependency toward fine-tuned, domain-specific models trained on our proprietary Indian developmental dataset, which will improve accuracy for Indian linguistic and cultural contexts while reducing cost and latency. |
Both. Our immediate focus (2026–28) is India — deploying across state ICDS systems, private preschools, and NGO programs. Our long-term vision (2029–30) includes Global South replication across Africa and Southeast Asia, addressing a $7B+ global opportunity in AI for child health and development. The platform's multilingual architecture and hierarchical multi-tenant design are already built to support multi-country deployment. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/uLXl5ak2E8Z3-HCS-ECD_AIConnect_AVINYA_AI_Accelerator-–-Cohort-2_PitchDeck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/uLXl5ak2E8Z3-HCS-ECD_AIConnect_AVINYA_AI_Accelerator-–-Cohort-2_PitchDeck.pdf |
https://youtu.be/9-WWcAKKeYA |
Absolutely. ECD-AI Connect is fundamentally a mission-driven startup. Our core mission is to ensure that no child in India loses the critical 0–3 year developmental window simply because the system wasn't designed to catch delays early. Today, 170 million children under 6 in India have zero access to AI-powered developmental screening, and delays are detected only at school age when intervention costs 7–10x more and outcomes are significantly worse. We are building the missing intelligence layer for India's public childcare infrastructure — not to replace ICDS, but to make it smarter. Our design choices reflect this mission: voice-first for low-literacy Anganwadi workers, 8 Indian languages so no child is excluded by language, offline-capable for rural India, and ICDS-interoperable so it works within the system rather than outside it. Every technical decision we make is measured against one question — does this help us reach more children, earlier? We believe early childhood development is not a premium service but a public right, and our goal is to make AI-powered screening as universal as Aadhaar or UPI. |
Yes — HCS is a women-led startup. Our Founder & Managing Partner, Pushpa Latha, leads the company's product vision, operations, and strategic direction. Women-led deep-tech startups remain significantly underrepresented in India's AI and health-tech ecosystem, and we are committed to demonstrating that women founders can build and scale critical national infrastructure. |
Self-referral / found online. |
checked |
| Mar 27, 2026 @ 12:45 PM |
Aneel Kumar Bonu and Prashanth Reddy |
aneel@talentspotify.com |
https://www.linkedin.com/in/aneel-kumar-bonu/ |
https://www.talentspotify.com |
8247716036 |
Founder & CEO: 23+ years of experience in HRTech, enterprise SaaS, and talent management solutions; led multiple product builds and GTM initiatives across India and global markets. Co-Founder: 14+ years of experience in HRTech, specializing in product engineering, AI/ML applications, and scalable platform development. |
We have been working together for over 12 years, having collaborated across multiple HRTech product initiatives.
Our partnership is built on complementary strengths across business strategy, product development, and AI-led innovation, with a proven track record of long-term execution. |
1 |
Deep domain expertise in HRTech combined with strong in-house capabilities across AI/ML, behavioral science, and full-stack engineering.
Proven ability to translate complex talent problems into scalable AI-driven products, with a strong focus on real-world enterprise adoption.
Balanced founding team with both business and technical leadership, enabling faster execution and product-market fit.
We combine domain depth with execution velocity, having already built and deployed AI-driven talent solutions with early customer validation. |
TalentSpotify Private Limited |
https://www.talentspotify.com |
Bengaluru |
We help companies retain top talent by turning performance into real-time rewards and eliminating bias in employee evaluations using AI. |
Performance evaluations in most organizations are subjective, inconsistent, and influenced by managerial bias.Employees lack real-time recognition and visibility into how their performance translates into rewards.This leads to disengagement, high attrition of top performers, and ineffective talent decisions. |
TalentSpotify is an AI-powered talent platform that transforms performance management into a continuous, data-driven system.
Our AI voice agent facilitates employee-manager interactions, capturing real-time performance signals and detecting cognitive and behavioral biases.
We convert KPIs into gamified, real-time rewards, creating transparent and motivating feedback loops.
Seamlessly integrates with existing HRMS/ERP systems, enabling organizations to improve retention, engagement, and performance—without increasing compensation costs. |
AI Voice-based Performance Layer: Unlike traditional HR tools, we capture live performance conversations, not just static inputs.Bias Detection Engine: Proprietary models analyze behavioral and cognitive biases in evaluations, improving fairness and decision quality.KPI-to-Rewards Engine: Direct linkage between performance and real-time rewards creates a continuous motivation system. |
Revenue |
Revenue |
Mid-sized companies (50–500 employees) across IT, GCCs, and services sectors looking to improve retention and performance management. Enterprises (1000+ employees) for AI voice-based performance review pilots and deployments |
$15 billion |
$6 billion |
$120 million |
SaaS subscription (per employee per month pricing). Enterprise licensing for AI voice agent deployments. Implementation and consulting services for customization and integrations |
Peoplebox.ai, Profit.co (OKR and performance management platforms). Darwinbox, Keka (HRMS platforms with performance modules) |
Channel partners, industry events and Founder-led sales and referrals within HR networks |
Start with mid-sized companies via pilot programs focused on retention and performance improvement. using microsites like AI HR Score as a lead magnet to drive inbound interest. Convert pilots into annual SaaS contracts with upsell to enterprise deployments |
To become the default AI-powered talent intelligence layer for organizations globally
Enable fair, real-time, and bias-free performance management across all employee segments
Make performance conversations accessible via voice and local languages, removing barriers of technology and literacy
Build a system where every employee understands and benefits from their performance in real time
Our vision is to move organizations from periodic reviews to continuous, AI-driven talent decisions. |
Private Limited |
INR 15 lacs |
Yes |
To refine our product-market fit for our AI voice-based performance platform and validate it across larger enterprise use cases. To leverage Avinya’s mentorship and network for scaling GTM, especially in enterprise and GCC segments.To accelerate our transition from early revenue to a repeatable and scalable sales model |
Close 5–10 enterprise pilot deployments for our AI voice-based performance solution. Convert at least 3–5 pilots into annual SaaS contracts. Strengthen our GTM strategy with a repeatable acquisition funnel (AI HR Score → Pilot → Conversion).Refine product positioning and pricing for mid-market and enterprise segments |
Yes, we are open to both hybrid and virtual participation and can actively engage across formats. |
Yes |
Core engine |
Multi-layer AI architecture combining speech-to-text (STT), LLM-based reasoning, and behavioral analytics
AI voice agent captures employee-manager conversations in real time
NLP models analyze sentiment, intent, and performance signals
Bias detection models identify cognitive and behavioral biases in evaluations
KPI-to-reward engine maps performance signals to real-time incentives
Built using a combination of LLM APIs, fine-tuned models, and scalable cloud infrastructure with integration into HRMS/ERP systems
The complete stack is hosted in AWS |
Proprietary dataset of employee-manager interaction data (voice + text) from performance conversations
Structured performance signals linked to KPIs, rewards, and outcomes
Behavioral and bias patterns across roles, industries, and organizations
Continuous feedback loops improve model accuracy and personalization over time.
This creates a compounding data advantage that improves with every interaction and is difficult to replicate. |
Proprietary interaction data + behavioral insights (not publicly available datasets)
Deep workflow integration into performance management systems
Continuous learning loops from real organizational usage
Domain-specific models tuned for HR and performance evaluation use cases
High switching cost once embedded into enterprise workflows |
Accuracy of transcription and intent detection
Precision/recall of bias detection models
Correlation between AI insights and manager/HR validation
User engagement metrics (conversation completion, feedback usage)
Business outcomes: improvement in retention, performance ratings consistency
Continuous human-in-the-loop validation for critical decisions |
Compliant with India DPDP Act and global data protection standards
Explicit consent-based data collection for employees and candidates
Data encryption at rest and in transit
Role-based access controls and audit logs
Option for enterprise clients to host data in-region (India/GCC) |
Potential concerns around employee data privacy and AI-driven decision transparency
Mitigated through consent mechanisms, explainable AI outputs, and human oversight
Designed to assist decision-making, not replace human judgment |
Real-time processing latency for voice interactions
Model inference costs at scale
Data pipeline and storage scaling challenges
Addressed through model optimization, batching, and scalable cloud infrastructure |
Yes, with expertise across NLP, speech processing, behavioral analytics, and full-stack AI systems
Team combines data scientists, ML engineers, and product engineers with HRTech domain knowledge |
Proprietary datasets from customer interactions (voice and text performance reviews) (one of the social enterprise we are using as pilot)
Synthetic and anonymized datasets for model training and testing
Public NLP datasets for baseline model benchmarking |
Continuous learning from new interaction data
Human-in-the-loop feedback from HR teams
Regular model fine-tuning and evaluation cycles
Expansion of domain-specific datasets across industries |
India-first approach focusing on mid-sized and enterprise customers
Expanding to GCC and global markets with multilingual and voice-first capabilities
Designed for diverse workforce environments and distributed teams |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/mZij4n0VCfLT-TS-Pitch-11Mar26-2.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/mZij4n0VCfLT-TS-Pitch-11Mar26-2.pdf |
https://youtu.be/HF0clsWkotQ |
Yes. TalentSpotify is focused on making workplace performance evaluation fair, transparent, and accessible to all employees.
We address systemic issues like bias in performance reviews and lack of recognition for high-performing employees.
By enabling voice-based interactions and multilingual access, we ensure that employees across different backgrounds, including those with limited digital proficiency, can participate effectively.
Our goal is to create equitable workplaces where performance—not perception—drives growth and rewards.
We aim to reduce bias in talent decisions and improve retention of deserving employees at scale. |
NA |
Email |
checked |
| Mar 27, 2026 @ 10:44 AM |
Ganesh Bhat |
indraliganesh@gmail.com |
https://www.linkedin.com/in/ganeshindrali/ |
https://www.wamimedia.com/ |
8970777835 |
Ganesh Bhat – Founder & Product Lead Responsible for product vision, technology, and overall execution of the WAMI ecosystem. Background in building digital platforms and leading product development. Krishna Murthy Bhat – Director Provides strategic support and business guidance, with experience in operations and management. |
The team is locally connected through Udupi and Dakshina Kannada, with strong regional understanding. We have been working together over the past few months on building WAMI and aligning on its vision and execution. |
1 |
Our biggest strength is our deep understanding of Tier 2/3 Indian markets, combined with the ability to build and execute quickly with limited resources. We have strong local networks, cultural understanding, and a clear focus on solving real grassroots problems through technology. |
Dzentral Media Private Limited (WAMI Media) |
https://www.wamimedia.com/ |
Udupi, Karnataka, India |
WAMI is a district-first digital platform delivering hyperlocal news, events, and government information in regional languages. |
Citizens in Tier 2/3 districts lack a reliable, structured digital platform for accessing hyperlocal news, community updates, and government information in their own language. Existing platforms focus on national content, leaving a gap at the district and taluk level where real daily information needs exist. |
WAMI Media is a mobile-first platform that aggregates and delivers district-level news, local events, and government information in regional languages. It combines a personalized content feed, community-driven updates, and verified information sources, enabling citizens to stay informed, businesses to gain visibility, and governments to communicate effectively at the grassroots level. |
WAMI’s key differentiation lies in its district-first approach, vernacular focus, and integrated ecosystem. Unlike traditional media platforms, WAMI operates at the taluk and district level, building a localized content and distribution network. Its defensibility comes from early establishment of local trust, user-generated content pipelines, language-based personalization, and network effects as it scales across districts, making it difficult for national players to replicate. |
MVP |
Users |
Primary (B2C): Citizens in Tier 2/3 cities (starting with Udupi & Dakshina Kannada) consuming local news, events, and government information in regional languages Secondary (B2B): Local SMEs, shops, service providers for ads, promotions, and hiring Tertiary (B2G): Government departments for awareness campaigns and public communication |
TAM: ₹50,000+ Cr (media + local ads + classifieds + jobs) |
SAM: ₹5,000–₹8,000 Cr |
SOM: ₹20–₹50 Cr achievable scale |
Local advertisements (SMEs, shops) Sponsored content (articles/videos) Event promotions & listings Classifieds (jobs, rent, services) Government awareness campaigns Future: subscriptions, business tools, and premium visibility |
Indirect: Instagram, YouTube, ShareChat (attention platforms) Direct (partial): Public apps, Dailyhunt, local news portals Offline: Local newspapers, WhatsApp groups |
Ground-level marketing in districts College partnerships & student ambassadors Local reporters & content creators WhatsApp sharing loops SME onboarding (shops bring users) Hyperlocal social media pages |
Launch in Udupi & Dakshina Kannada Build strong local content + trust + daily usage Onboard SMEs & events for monetization Expand district-by-district across Karnataka Replicate model across states with regional branding |
To build India’s district-first digital infrastructure platform connecting citizens, businesses, and governments — evolving from a media platform into a hyperlocal super app for Bharat, enabling content, commerce, jobs, services, and public communication at the grassroots level. |
Private Limited Company (Dzentral Media Private Limited) |
Bootstrapped (₹3 Lakhs Mudra Loan Approved) |
Yes |
We are building WAMI, a district-first digital platform focused on delivering hyperlocal news, public information, and community engagement in regional languages. We are applying to Avinya to gain structured mentorship, refine our product-market fit, and build a scalable growth strategy. We believe Avinya’s ecosystem can help us transition from an early-stage MVP to a strong, execution-driven company with the right guidance in product, distribution, and fundraising. |
During the programme, we aim to successfully launch WAMI in Udupi and Dakshina Kannada, achieve strong user adoption, and validate our hyperlocal content and engagement model. We want to build a repeatable district expansion playbook, improve retention through product iterations, and establish early revenue streams through local ads and classifieds. Additionally, we aim to strengthen our team, technology, and investor readiness. |
Yes |
Yes |
Supporting feature |
WAMI leverages a modular AI architecture combining:
Content processing layer for summarization and translation (English ↔ regional languages)
Recommendation engine for district-level personalized feeds
Classification models for categorizing news, events, and user-generated content
The system is designed to evolve from rule-based logic to ML-driven personalization as data scales. |
WAMI is building a district-level structured data layer, including:
Hyperlocal news and events
User-generated reports
Local business and government data
Vernacular engagement patterns
This creates a unique regional language + district behavior dataset, which is currently underserved in India. |
Hyperlocal data moat (district-level, not city-level)
Vernacular-first content across multiple Indian languages
User-generated + verified content loop
Continuous accumulation of behavioral and engagement data
Over time, this builds a network + data advantage difficult to replicate. |
Content relevance (CTR, dwell time)
Engagement metrics (session time, retention)
Recommendation accuracy (click-through rate)
Language accuracy (translation quality checks)
Content moderation accuracy (false positives/negatives) |
Minimal data collection (privacy-first approach)
Secure APIs with authentication
Data stored in compliant cloud infrastructure
No sensitive personal data is shared with third parties
Planned compliance with India DPDP Act and global standards as we scale |
Content moderation and misinformation regulations
Data protection laws (India DPDP Act)
Platform liability for user-generated content
We mitigate this through moderation systems, disclaimers, and verification layers. |
Cold-start problem in new districts
Data imbalance across regions
Increased moderation complexity
Infrastructure scaling for real-time recommendations
We plan to address this through distributed systems, better data pipelines, and model optimization. |
Currently, AI development is handled by the core tech team with experience in:
Backend systems
Data pipelines
API integrations
We plan to onboard dedicated AI/ML engineers as we scale. |
Public news datasets (aggregated and transformed)
Government open data sources
Internally generated content (owned)
User-generated content (owned with consent) |
Feedback loops from user engagement
Continuous data collection from districts
A/B testing of recommendation systems
Incremental model upgrades
Human-in-the-loop moderation |
Primarily India (Bharat-focused), with long-term expansion potential to other vernacular-first emerging markets. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/OotQdromUcbB-WAMI_DzentralMediaPrivateLimited.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/OotQdromUcbB-WAMI_DzentralMediaPrivateLimited.pdf |
http://NA%20(MVP%20under%20development%20–%20demo%20available%20on%20request) |
Dzentral Media Private Limited is building WAMI, a district-first digital platform aimed at solving the problem of information access in India’s Tier 2 and Tier 3 regions. Millions of citizens lack structured access to verified local news, government schemes, and community updates in their own language.
WAMI addresses this by delivering hyperlocal, language-first content, improving awareness of public services, enabling local businesses to grow, and creating digital infrastructure at the grassroots level.
Our mission is to empower citizens, strengthen local economies, and bridge the digital information gap across Bharat. |
NA |
Self-applied |
checked |
| Mar 27, 2026 @ 12:09 AM |
Mountford Vijay Thomas, Mercy Thomas |
vijay.thomas@safeblocksolutions.com |
https://www.linkedin.com/in/mountford-thomas/ |
https://safeblocksolutions.com/ |
9341557357 |
1. Mountford Vijay Thomas, currently working as CEO of SafeBlock Solutions LLP, which is a software development company focussed on solving real world problems with focus on circular economy, reduce carbon footprint and provide opportunity to large group of people from different background to earn revenue from our business model. I come with more than 2 decades of experience in IT having worked at CXO level and managing large teams and projects globally. 2. Mercy Thomas, Director & Managing Partner at SafeBlock Solutions LLP provides guidance and mentorship for shaping the ideas that helped in building our first product called V2A (Value To All), which is an Online Marketplace for buying and selling second hand products with Trust. She is a Mathematics Graduate from MG University Kerala and is 75 years old |
The 2 founders who built SafeBlock Solutions LLP are related to each other by relation of Mother & Son. I am 46 years old so my association with my mother is as old as my age. The DNA of this company is based on building product which helps individuals to use their time and skill to earn a legitimate source of income and help society to improve longitivity of products with trust and transparency using human touch and technology. |
1 |
We are a small team of 7 members consisting of Product Manager and software developers with an aim to build a world class product with utility which can be used by anyone in the world with an objective of providing transparent and trustworthy transactions, stored on public blockchain. The business model is self sustaining and designed to propogate to any geography with minimum human intervention for its setup. |
SafeBlock Solutions LLP |
https://safeblocksolutions.com/ |
Bengaluru |
A software development company focussed on developing products to solve real world problems and benefit members of society with the aim of providing a revenue source for anyone and everone who gets associated with our project. |
We are solving problems related to Trust and financial frauds which exist in Online Marketplace in recommerce market using technology and human touch, both of which are the USP's of our project. |
We built our product, V2A (Value To All), with unique features. Some of which are :
1. Privacy Protection - In order to protect sellers from spam calls, their personal information such as mobile number and address is kept confidential from Public. Hence sellers can list their used products without worrying about spam calls.
2. Validated Product : Every second hand product that is listed on this platform is physically verified by an individual, known as validator, whose responsibility is to provide correct information to potential buyers and ensuring the listed products are genuine with description matching with what seller has mentioned for their product.
3. Transparent Bids - In order to get best price for a second hand product, the platform provides transparent bidding process which is visible to all potential buyers thus providing transparency and ensuring the product value is determined by market value.
4. Safe Payment : Once seller identifies the highest bidder of his product, the buyer makes the payment through company managed escrow account to ensure buyers payment is protected.
5. Transaction Audit : All successful transactions are recorded on a public blockchain for transparency, which is accessible to buyer and seller within the app. |
Physical Validation of product and Escrow Payment mechanism is very unique to our solution and differentiates us from our competitors. |
Product-Market Fit |
Users |
Working professionals in the age group of 24 to 45 in Tier 1 cities where migratory population is high |
$58 billion |
$5.9 billion |
$3.5 billion |
Platform fee + Premium listing + Ads based on subscriber base |
OLX, Facebook Marketplace |
Primarily through Digial Marketing through social media platforms and Influencer based marketing |
Targeting Tier 1 cities such as Bangalore, Chennai, Mumbai, Delhi, Hyderabad |
In 5 years, this platform should scale across pan India and capture 10% of recommerce market in India. Once proven in India, then roll it across global cities especially ones with high migratory population. The app should become a revenue generating source for people who volunteer to become validators, so that they get paid handsome commission for the time and skill utilized in validating seller products. |
LLP incorporated in Oct 2024 at Bengaluru |
NA |
Yes |
One to get funding, especially for marketing, research and operations. Second reason is to get expert views in improving the product and business model. |
Learn various options to generate revenue and achieve breakeven in best possible time, maybe 18 to 24 months. |
Hybrid Participation preferred |
No |
Not applicable |
NA |
NA |
NA |
NA |
Transaction Data stored on Public Blockchain with encryption for provacy, compliance and security |
Not anything I can think of in the current setup |
NA |
NA |
NA |
NA |
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/IMtFKbQmjiRw-Pitch-Deck-2-Business-Model.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/IMtFKbQmjiRw-Pitch-Deck-2-Business-Model.pdf |
https://www.youtube.com/shorts/IfCH25xNoSg |
I am building an impact focused startup which not only solves real world problem and helps to build circular economy but it also provides opportunity for employment/revenue generation for multiple users who get associated with our app. The Validator concept was built specially to cater to this need, which not only adds value to a second hand product by verifying products quality but also provides opportunity to individuals to become validators and earn income through commission on every validated product that gets sold on our platform. |
NA |
NA |
checked |
| Mar 26, 2026 @ 9:05 PM |
Varun U, Shilpa V P, Sarang T S |
varun@testmyskills.ai |
https://www.linkedin.com/in/varun-umesh/ |
https://testmyskills.ai |
+91 8296024720 |
Varun U (Co-Founder & CEO): Founded Zysk Technologies (2015). Scaled from 2 → 75 people, ₹5Cr revenue. Built EdTech products used by 100K+ students. Leads business, product strategy and R&D. BE, JSS Academy of Technical Education. Shilpa V P (Co-Founder & CTO): 10 years in tech. Founding member at Zysk Technologies. Built TestMySkills and its AI engine from scratch. 0 → 25K users in 13 months. BE in CS, Hassan Engineering College. T S Sarang (Founding Member): Founded Inovact. Scaled to 13K users and profitability. Drives ops, GTM, and digital presence at TestMySkills. Grew social media from 300 → 35,000 followers in 4 months. BE in EC, BMSCE. |
Shilpa and Varun have been working together for over 12 years. They were colleagues at their previous company, and when Varun decided to start Zysk Technologies, they chose to continue building together. Varun's role has always been solutioning and strategy, while Shilpa has been the execution specialist. They grew from a two-member team to 75 people, hitting roughly half a million dollars in revenue over 10 years.
Through it all, they've shared a common drive: building something that helps people get jobs. That need has only grown stronger in today's market. So they started building TestMySkills, a solution that solves the skill problem, not just the symptom. We've reached over 30K users and positively impacted the majority of their learning journeys.
With Sarang joining the team, bringing his experience of building communities and scaling to 13K users at his previous social network venture independently, we're more confident than ever about making a real dent. |
All |
Every founder on this team has independently built and scaled a product before. Varun and Shilpa co-founded Zysk Technologies, scaling it to 85 people and half a million dollars in revenue, building EdTech products used by 100K+ students. Shilpa then built TestMySkills and its AI engine from scratch, reaching 25K users in 13 months. Sarang built a student social network to 13K users and profitability, and has grown TestMySkills' social media from 300 to 35,000 followers in 4 months.
Our biggest strength: we combine deep EdTech domain expertise, technical execution speed, and community-building ability in one team. Varun brings strategy and R&D, Shilpa brings technical depth, and Sarang brings GTM and audience growth. We've proven we can build, ship, and grow with minimal capital. |
TestMySkills |
https://www.testmyskills.ai |
Bengaluru |
TestMySkills.ai is an AI-powered adaptive assessment platform that identifies what a learner doesn't know and builds a personalized mastery plan to fix it. |
Current testing systems measure completion, not competence, leaving learners unaware of their weak spots, employers spending lakhs retraining bad hires, and educators losing a third of their week to test creation and grading. |
TestMySkills diagnoses exactly what a learner doesn't know and builds a personalized mastery roadmap. Our platform covers four verticals: competitive exam prep, skill assessments, AI-powered interview prep, and certifications. All powered by one adaptive AI engine that generates assessments for any skill, any exam, any interview in seconds. |
Three barriers protect us. First, our proprietary adaptive assessment MASTER framework combines psychometrics, knowledge engineering, and ML, with a patent in process. Second, our agentic AI engine launches new assessment domains in weeks using machine-readable knowledge graphs, while competitors take years. Third, every assessment taken makes our engine smarter. We already have 35,000+ learners worth of data that a new entrant would start from zero. |
Revenue |
Revenue |
Students preparing for competitive exams, job seekers, working professionals, engineering colleges (T&P cells), HR firms, and corporates. |
₹1,585Cr (India) |
₹143.3Cr |
₹1.5Cr |
B2C subscriptions starting at ₹99, B2B SaaS subscriptions for colleges, HR firms, and corporates, workshops and bootcamps, and assessment-as-a-service pricing at ₹100-500 per candidate. |
ALEKS (math-only), Carnegie Learning (years per domain), Duolingo/Anki (concepts forget independently), Testbook/Unacademy (same questions for everyone). Our competitors solve one piece. We built the full stack. |
B2B brings the users, B2C monetizes them. We partner with colleges, HR firms, and corporates, offering them free job postings, placement support, and AI-powered interviews. In return, they bring their students, job seekers, and employees onto the platform. Once on the platform, learners pay for credits starting at ₹99 to access assessments, interview prep, and certifications. On top of this, we drive direct signups through social media. |
Start with engineering colleges and exam prep students in Karnataka, expand to HR firms and corporates in top metros, then scale nationwide. B2B partnerships as the primary distribution channel, supported by organic social media growth. |
We're building India's default infrastructure for skill verification. Starting with assessments in Karnataka, scaling to every exam, every skill, & every interview nationwide. The vision: any learner, anywhere, can prove what they actually know, & not just what degree they hold. TestMySkills for Assessments, the way Paytm is for Payments and Swiggy is for Food Delivery. |
Private Limited |
We have received a grant of INR 15L from Startup India Seed Fund Scheme |
Yes |
Avinya's combination of structured mentorship, IIT alumni network, and deep industry connections is exactly what we need at this stage. We have a working product with 35,000+ users and 3 paid B2B pilots, but we need help accelerating B2B deal closures with colleges and corporates, expanding beyond Karnataka, and refining our growth strategy. Access to ACSEL's network of industry leaders and Karnataka government support aligns directly with our goal of becoming India's default assessment infrastructure. |
Close 10 college partnerships and 8 corporate/HR firm contracts. Refine our B2B sales playbook with mentor guidance. Expand from Karnataka to at least 2 additional metros. Strengthen our product roadmap with input from industry experts. Position ourselves for our ₹2Cr pre-seed raise with investor introductions through the ACSEL network. |
Yes, we are open to virtual participation. |
Yes |
Core engine |
Multi-agent assessment engine built on LangChain and LangGraph, with model providers including OpenAI, Anthropic, Gemini, and Perplexity. The engine has 6 levels: orchestration (routes requests to the right assessment flow), deep context and research (understands syllabi, skills, job descriptions), use-case subgraphs (exam-specific, interview-specific, certification-specific generators), performance analytics (tracks patterns, errors, speed), the MASTER Framework (maps knowledge states, identifies gaps, generates mastery roadmaps), and personalization (adaptive difficulty, learner profile, recommendations). Backend: Nest.js. Frontend: Next.js. Databases: PostgreSQL and Neo4j for knowledge graphs. Cloud: AWS. |
35,000+ learners worth of assessment response data including answer patterns, skill gaps, learning rates, decay rates and performance trends across competitive exams, skill assessments & interview preparation. Every assessment taken feeds back into our engine, making it smarter. This data powers our adaptive models and is not available to any new entrant. |
Three layers. First, our proprietary MASTER Framework combines psychometrics, knowledge engineering, and ML, with a patent in process. Second, our multi-agent system generates assessments for any domain in weeks using machine-readable knowledge graphs, while competitors take years per domain. Third, our compounding data from 35,000+ learners continuously improves our adaptive models. This is a science moat, not replicable by adding AI features to a content platform. With this our learners will be able to improve their skill at 1/3rd the speed compared to traditional edtech applications. |
Assessment quality: accuracy of generated questions validated against actual exam papers, difficulty calibration using Item Response Theory parameters. Learner outcomes: test completion rate, & score improvement across repeated assessments, time-to-mastery reduction. Engine reliability: response latency, error rates, and model output consistency across providers. |
Our privacy-first proctoring uses risk-scoring models to flag anomalies without video or audio capture, meeting India's data protection rules and GDPR-style standards out of the box. All data is hosted on AWS with encryption at rest and in transit. Role-based access controls for learner, faculty, and HR users. No biometric data is collected or stored. |
India's Digital Personal Data Protection (DPDP) Act is tightening data handling requirements. However, our architecture is already built for compliance since we don't capture video, audio, or biometric data. Stricter regulations actually benefit us, as competitors relying on invasive proctoring will need to rebuild their systems. |
At 10x scale, the main pressure points would be knowledge graph query latency as concurrent users spike, and AI inference costs increasing across multiple model providers. We're mitigating this by designing our architecture to be model-agnostic (switching providers based on cost and speed), using caching layers for frequently accessed knowledge graph paths, and leveraging AWS elastic infrastructure for traffic spikes during peak exam & hiring seasons. |
Yes. Varun (CEO) leads R&D on the MASTER Framework, including IRT, Neo4j knowledge graphs, and RL architecture. Shilpa (CTO) built the entire AI engine from scratch, including the multi-agent assessment system on LangChain and LangGraph. Our development team of 6 supports platform and AI feature development. We also have advisory support from Dr. Raghavendra Rao (Behavioral Science Expert). |
All datasets are owned. Our primary datasets include: 35,000+ learner response records (owned, generated on-platform), curated question banks across competitive exam syllabi (created in-house and through AI generation), and knowledge graphs mapping concept relationships across exam domains (built in-house using Neo4j). No third-party licensed datasets. |
Every assessment taken feeds response data back into our engine. We use this to recalibrate question difficulty parameters, refine knowledge graph relationships, and improve adaptive path recommendations. As our user base grows, our IRT models become more precise. We also continuously evaluate and integrate new model providers to optimize for cost, speed, and accuracy. |
Starting with India, specifically Karnataka as our beachhead market. Our domain-agnostic engine is designed to scale to any exam, skill, or interview domain globally. The long-term vision is to become a global skill verification infrastructure, starting here, scaling everywhere. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Fo8mTmlyOY7w-TestMySkills-ACSEL-Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Fo8mTmlyOY7w-TestMySkills-ACSEL-Pitch-Deck.pdf |
https://drive.google.com/file/d/1FSa4sP2rrOiR3buojv4smqhwd7H-P-rj/view?usp=sharing |
Both. Our mission is to make competence visible and accessible for every learner, regardless of their background or budget. The impact we're driving is closing the gap between completion and competence across India's education and employment ecosystem. We do this by making quality assessments accessible starting at ₹99 for competitive exam aspirants, job seekers, and working professionals, giving them a credible way to prove what they can actually do, not just what degree they hold. Everything we build is driven by one question: does this help someone prove what they truly know and get where they deserve to be? |
NA |
IT BT Dept, Govt. of Karnataka |
checked |
| Mar 26, 2026 @ 6:57 PM |
Swarasya VL, Sujay P, Prithiv Palani |
swarasya.vl@schijnenn.com |
https://www.linkedin.com/in/swarasya-v-l-b52730207/ |
https://schijnenn.com/ |
9480258896 |
1. Swarasya V L - Co-Founder & Business Management Lead B.E. in Computer Science | 8+ years at Mercedes-Benz R&D India Leads overall business strategy, partnerships, and institutional collaborations. Drives program management, client engagement, and organizational governance. Brings strong technical foundation combined with business leadership to drive sustainable growth. 2. Prithiv Palani - Co-Founder & Technology Driver B.Tech in Computer Science | 9 years (Mercedes-Benz R&D + leading startups) Defines and executes the core technology roadmap and digital innovation strategy. Architects scalable software platforms and advanced engineering solutions. Serves as the principal technical enabler for high-impact, mission-critical programs. 3. Sujay Prabhu - Co-Founder & Operations / Business Lead B.E. in Computer Science | 8 years experience across Java R&D, Byju’s, 7Edge Leads operations, administration, and internal process optimization. Manages business development, client relationships, and commercial execution. Ensures operational excellence, compliance, and effective stakeholder coordination. |
The team has a strong foundation built on both professional collaboration and trust. Prithiv and Swarasya worked together at Mercedes-Benz for over 4 years, where they collaborated closely on immersive engineering and design-related workflows. Sujay joined the team with a complementary skill set, and the existing trust and alignment within the team helped in forming the venture.
This combination of prior professional experience and strong personal trust enables us to work efficiently, make quick decisions, and execute effectively as a team. |
All |
Our team’s biggest strength is the combination of deep domain understanding and fast execution. Having worked closely in automotive environments, we understand real design and engineering workflows—not just from a theoretical perspective, but from hands-on experience.
At the same time, we have the ability to move quickly from idea to implementation, as demonstrated by building an MVP and progressing to a paid pilot with Maruti Suzuki in a short timeframe.
This blend of industry insight, technical capability, and execution speed allows us to build solutions that are both practical and impactful. |
SCHIJNENN DIGITAL PRIVATE LIMITED |
https://schijnenn.com/ |
Bengaluru |
Schijnenn is a deep tech product company focused on creating AI powered immersive and interactive experiences without any complex toolchains or coding. |
Automotive workflows today are fragmented and slow : Design conversion takes time and requires multiple tools, while maintenance relies heavily on manuals and experience, leading to delays and errors. With the shift toward digital and connected systems, there is a clear need for faster, more integrated workflows. Our solution addresses this gap by enabling rapid design generation, seamless CAD-to-interactive conversion, and real-time guided maintenance. |
Our solution combines RealEdg and Sketch to 3D to create a seamless, end-to-end workflow for automotive design and validation.
RealEdg is our core platform that directly imports native CAD data and automatically converts it into interactive, immersive experiences without the need for intermediate tools, enabling real-time visualization and collaboration.
Complementing this, Sketch to 3D acts as an AI Design Co-Pilot that converts rough sketches and prompts into accurate 3D models within minutes, accelerating early-stage design exploration.
Together, they enable teams to move from idea to interactive validation instantly eliminating delays, reducing iterations, and improving decision-making. |
We differentiate by integrating AI-driven design, direct CAD-to-experience conversion through RealEdg, and immersive interaction into a single platform eliminating the need for multiple tools and significantly accelerating workflows |
Revenue |
Revenue, Pilots, Testimonials |
Automotive OEMs and engineering teams are our primary customers, along with Tier 1 suppliers. Through Sketch to 3D, we also target fashion, construction, and architecture sectors, and through our AR solutions, maintenance teams in automotive and infrastructure. |
$180 billion + |
$350-600 million |
$35-45 billion |
Our revenue model is primarily enterprise SaaS-based, combined with pilot engagements and customization services. We begin with paid pilot projects with OEMs and enterprise customers to validate use cases and demonstrate ROI. These pilots typically convert into annual enterprise licenses based on number of users, features, and deployment scale. For our core platforms—RealEdge and Sketch to 3D—we offer subscription-based pricing, with tiers depending on usage, integrations, and advanced capabilities. Additionally, we generate revenue through customization, integration, and deployment services, especially for enterprise clients with specific workflow requirements. In the long term, we plan to expand into platform-based pricing with usage-based components as adoption scales |
Vizcom, Meshy AI |
We acquire customers through enterprise partnerships, accelerator programs like the Maruti Suzuki Accelerator, and a pilot-led approach that converts into long-term contracts, supported by referrals and industry networks |
We follow a pilot-led enterprise GTM strategy starting with OEM partnerships, deploying paid pilots to prove ROI, and scaling through a land-and-expand approach, supported by industry references and partnerships |
Our long-term vision is to build a unified AI-driven platform that powers the entire lifecycle of products from concept design to engineering validation to real-world maintenance.
We aim to become the core digital layer that connects ideas, data, and real-time experiences, enabling organizations to move faster, make better decisions, and operate more efficiently.
Starting with automotive, we plan to expand across industries such as construction, architecture, and manufacturing, where similar challenges exist in design and operational workflows.
Ultimately, we envision a future where creating, visualizing, and interacting with complex systems becomes instantaneous and intuitive driven by AI and immersive technologies. |
Incorporated on 15.01.2025 and DPIIT recognised. CIN - U62099KA2025PTC196877, DPIIT Number - DIPP218079 |
NA |
No |
We are applying to Avinya to collaborate on next-generation mobility solutions, validate our AI-driven platform in new environments, and scale through strategic industry partnerships |
We aim to validate our solution in real-world use cases, refine and scale our product, and build long-term enterprise partnerships leading to larger deployments |
Yes |
Yes |
Core engine |
Our AI architecture is designed as a modular, multi-stage pipeline that converts unstructured inputs into structured, usable 3D outputs.
1. Input & Preprocessing Layer:
We accept sketches, images, and text prompts, which are preprocessed using computer vision techniques such as edge detection, feature extraction, and normalization to capture design intent.
2. Generative AI Layer:
We use generative models to interpret inputs and generate structured 3D representations. This includes diffusion-based and neural implicit techniques to create geometry aligned with design constraints and style.
3. Design Alignment & Refinement Layer:
Outputs are refined using constraint-based optimization and domain-specific rules (e.g., automotive design patterns) to ensure feasibility, proportions, and consistency.
4. Geometry Structuring Layer:
The generated outputs are converted into structured 3D formats (meshes/surfaces) that can be used in downstream engineering workflows.
5. RealEdg Integration Layer:
The AI outputs are integrated with our RealEdge engine, where native CAD data and generated models are optimized and converted into real-time interactive assets.
6. Feedback & Learning Loop:
We incorporate human-in-the-loop feedback and usage data to continuously improve model accuracy and performance over time. |
Our proprietary data advantage comes from a combination of domain-specific data, workflow data, and continuous feedback loops.
1. Through our engagement with Maruti Suzuki via the Maruti Suzuki Accelerator, we are exposed to real-world automotive design and engineering workflows, enabling us to understand patterns, constraints, and practical requirements that are not available in public datasets.
2. As users interact with our platform, we generate valuable workflow data such as sketch-to-output mappings, design iterations, and decision patterns which helps improve our models over time.
3. Our integration with CAD and real-time interaction systems allows us to build structured datasets around geometry processing, optimization, and usage behavior, creating a strong feedback-driven data loop.
Together, this creates a compounding data advantage that improves model accuracy, aligns outputs with real industry needs, and becomes increasingly difficult to replicate. |
Our AI component is highly defensible due to our unique combination of game-engine-based architecture and deep automotive domain expertise.
Unlike most competitors, our core system is built on Unity, allowing us to natively bridge CAD data with real-time interactive environments. This gives us a significant advantage in generating accurate, real-time-ready 3D outputs rather than just static or approximate models.
Additionally, our team’s prior experience at Mercedes-Benz enables us to understand complex automotive design and engineering workflows at a granular level. This allows us to combine CAD precision with game development techniques to produce outputs that are both accurate and usable in real-world scenarios.
This combination of a unique technical foundation and domain expertise, along with continuous learning from real-world usage, creates a strong barrier for competitors who typically operate either in AI visualization or CAD domains but not both. |
We evaluate model performance and reliability using a combination of geometric accuracy, usability, and workflow efficiency metrics.
1. Geometric Accuracy Metrics:
We measure how closely the generated 3D outputs align with expected shapes using metrics such as shape similarity, dimensional accuracy, and surface deviation from reference models.
2. Design Consistency & Validity:
We evaluate whether outputs follow domain-specific constraints (e.g., proportions, symmetry, and design language), ensuring they are not just visually correct but practically usable.
3. Usability Metrics:
We track how often generated outputs are directly usable in downstream workflows such as integration into CAD or RealEdge without major rework.
4. Iteration Reduction:
We measure reduction in the number of design iterations required to reach a final output, which directly reflects model effectiveness.
5. Time-to-Output:
We track how quickly users can move from input (sketch/CAD) to a usable result, as speed is a critical performance indicator.
6. User Feedback & Acceptance Rate:
We incorporate qualitative and quantitative feedback from users, including acceptance rates of generated outputs and satisfaction scores.
7. System Performance Metrics:
We monitor latency, rendering performance, and stability in real-time environments to ensure a smooth user experience. |
We ensure data security through encryption, access controls, and secure deployments (cloud/on-premise). We maintain strict data privacy with client ownership and consent-based usage, and align with industry standards like ISO 27001 and SOC 2. |
Regulatory risks are mainly around data security and use in safety-critical workflows. We mitigate these through compliance with standards, secure deployments, and positioning the system as a decision-support tool. |
At 10x scale, challenges include increased latency, maintaining model accuracy across diverse data, system reliability under high concurrency, and integration complexity. We mitigate these through optimization, modular architecture, and scalable infrastructure. |
Yes, we have an in-house AI/ML capability supported by a team with expertise across computer vision, generative AI, and 3D data processing.
Our team combines strong technical skills in areas such as image understanding, generative modeling, and geometry processing with deep domain knowledge in automotive design and engineering. This allows us to build models that are not only technically robust but also aligned with real-world workflows and constraints.
Additionally, our experience in integrating AI with real-time systems and platforms like Unity enables us to deliver outputs that are directly usable in interactive environments, rather than just theoretical models.
This combination of AI expertise and domain understanding is a key strength in building practical, high-impact solutions |
We use a combination of publicly available, synthetic, and proprietary datasets to train and refine our models.
Public datasets are used for foundational training in areas such as image understanding and 3D representations, ensuring a strong base for our AI models.
We also generate synthetic data to augment training, especially for controlled scenarios and edge cases where real-world data may be limited.
Most importantly, we build proprietary datasets through platform usage such as sketch-to-output mappings, workflow interactions, and feedback loops which help us continuously improve model performance in a domain-specific manner.
For enterprise engagements, including our work with Maruti Suzuki, all data usage follows strict confidentiality and ownership agreements, and customer data is not used for training without explicit consent.
Overall, our approach ensures a strong balance between licensed/public data and continuously growing proprietary data |
We continuously improve model performance through a combination of data, feedback, and iterative optimization.
1. We leverage a feedback loop from real-world usage capturing sketch-to-output mappings, user edits, and acceptance rates to refine our models and align them with practical requirements.
2. We expand and enhance our datasets using a mix of proprietary data, synthetic data generation, and domain-specific augmentation to improve robustness across diverse use cases.
3. We follow a modular model approach, allowing us to fine-tune and upgrade specific components (e.g., geometry generation, design alignment) without impacting the entire system.
4. We incorporate domain-specific constraints and rules, especially from automotive workflows, to improve accuracy and usability beyond generic AI outputs.
Finally, we continuously optimize performance through benchmarking, A/B testing, and monitoring key metrics such as accuracy, latency, and user acceptance rates |
We are building for both India and global markets.
India is our primary starting point, where we are working closely with leading OEMs like Maruti Suzuki to validate our solution in real-world use cases and establish strong product-market fit.
At the same time, the problem we are solving inefficiencies in design and maintenance workflows is global in nature. Our platform is designed to be scalable and adaptable across geographies, enabling us to expand into international markets, particularly in automotive and industrial sectors.
Our approach is to build and validate in India, and then scale globally with a strong, proven foundation. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/J0SwzZRtjWGH-Schijnenn-Digital.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/J0SwzZRtjWGH-Schijnenn-Digital.pdf |
https://drive.google.com/file/d/1EOp1YEzBOyG8Q3MGYSdsGaElWwVXQm9z/view?usp=sharing |
Yes, we are building a mission-driven startup focused on improving how complex products are designed, validated, and maintained. Our mission is to make these workflows faster, more accessible, and less dependent on manual effort and specialized expertise.
By enabling AI-driven design and real-time interactive guidance, we help teams reduce inefficiencies, accelerate innovation, and improve overall productivity. This has a direct impact on large-scale industries by reducing time-to-market, minimizing errors, and optimizing resource utilization.
Additionally, our AR-based maintenance solutions democratize access to knowledge by enabling even less experienced technicians to perform complex tasks with higher accuracy, contributing to skill enhancement and workforce efficiency.
As we expand into sectors like construction and infrastructure, we see a broader impact in improving operational efficiency and supporting scalable, technology-driven growth. |
Yes, our team includes members from underrepresented backgrounds in tech. We believe this diversity brings unique perspectives to problem-solving and product design, and it strengthens our ability to build inclusive, practical solutions for a wide range of users |
IIT Madras Incubation Network |
checked |
| Mar 26, 2026 @ 6:14 PM |
Vadthyavath Nikshitha |
nikshitha1312@gmail.com |
https://www.linkedin.com/in/v-nikshitha-860655258/ |
https://gozy.online/ |
9347556415 |
Nikshitha is the co-founder leading product vision, design, and overall execution of Gozy. She has a strong focus on building user-centric digital experiences and has hands-on experience in web/app development, allowing rapid prototyping and iteration. Sandeep is the co-founder responsible for technology and system architecture. He brings technical expertise in development, problem-solving, and building scalable solutions, ensuring the platform is robust and efficient. |
We met during our academic journey and connected over a shared interest in technology and building impactful solutions. We have been working together for the past few months, collaborating closely on ideation, development, and refining Gozy. |
2 |
Our biggest strength is the combination of product vision and strong technical execution. We complement each other well while one focuses on user experience and product direction, the other ensures solid technical implementation. This allows us to move fast, iterate quickly, and build scalable solutions with a clear user focus. |
Gozy Private Limited |
https://gozy.online/ https://portfolio.gozy.online/ |
Hyderabad |
Gozy is a superapp that unifies food, travel, shopping, tickets, and payments into one seamless, AI-powered, reel-driven platform. |
Today’s users rely on 30–60 different apps to manage everyday needs like food ordering, travel booking, shopping, payments, and entertainment. This leads to fragmented experiences, constant app-switching, lost information (tickets, receipts), and notification overload. Existing platforms solve individual problems but fail to provide a unified, intuitive experience, making digital life inefficient and cluttered. |
Gozy consolidates multiple essential services into a single intelligent platform. Users can discover, decide, and transact through a seamless reel-based interface similar to scrolling content but with instant actions like ordering food, booking travel, or purchasing products. The platform automatically stores tickets, receipts, and confirmations, while AI-powered systems ensure content authenticity and personalized recommendations. This creates a smooth, end-to-end user journey without leaving the app. |
Gozy stands out through its reel-to-transaction model, where discovery and purchase happen in the same interface, significantly reducing friction. Its multi-service integration across verticals creates strong network effects, making the platform more valuable as users engage with multiple services. Additionally, the AI-based content verification system builds trust by filtering misleading or synthetic content. Combined with auto document storage and a unified ecosystem, this creates high user retention and strong defensibility. |
MVP |
Signups |
Gozy targets digitally active users who frequently use multiple apps for daily needs. Our primary segments include students and Gen Z users, urban professionals, frequent travelers, online shoppers, and food enthusiasts. Initially, we are focusing on college campuses to build high engagement and dense user networks before expanding to broader urban markets. |
$175B+ |
$4B+ |
$2M |
Gozy follows a multi-stream revenue model including: Commission on transactions (food orders, bookings, shopping) Featured listings and promotions for businesses Advertising through reel-based discovery Premium services and subscription features (future) Data-driven insights for partner businesses |
Our competitors include vertical-specific platforms like food delivery apps, travel booking platforms, e-commerce marketplaces, and payment apps. However, these operate in silos. Gozy differentiates itself by combining all these services into a single, integrated superapp with content-driven discovery and AI-powered trust systems. |
We acquire users through campus ambassador programs, referral incentives, social media marketing (especially reel-based content), and partnerships with local businesses. Our content-driven discovery model also enables organic growth through shareable and engaging user experiences. |
We are launching within college campuses to build a dense and highly engaged user base. By focusing on a specific ecosystem, we can drive frequent usage and strong network effects. Once validated, we will expand city by city, onboarding local vendors and scaling through partnerships and digital marketing. |
Gozy aims to become a Personal Digital OS the only app users need to manage their daily lives. We envision a future where discovery, decision-making, and transactions happen seamlessly within one intelligent platform that learns user behavior, anticipates needs, and simplifies digital experiences at scale. |
Incorporated startup, recognized under DPIIT (Startup India) |
NA |
Yes |
We are applying to Avinya to accelerate Gozy’s growth with the right mentorship, network, and strategic guidance. As an incorporated and DPIIT-recognized startup, we are now focused on moving from MVP to real user traction. Avinya’s ecosystem can help us refine our product, strengthen our go-to-market strategy, and scale efficiently with the right support. |
During the programme, we aim to launch our beta, onboard active users within campuses, and build strong early traction. We want to refine our product based on real user feedback, establish key partnerships, and prepare for our first round of funding with clear metrics and growth validation. |
both |
Yes |
Core engine |
Gozy uses AI across multiple layers to power discovery, personalization, and trust. The system combines recommendation algorithms, computer vision, and LLM-based processing. Reel-based content is analyzed using computer vision models to understand context (food, travel, products), while NLP models process metadata, captions, and user interactions. A recommendation engine personalizes content and services for each user. The backend is built using scalable cloud infrastructure with APIs for model inference and real-time data processing. |
Gozy builds a unique dataset from user interactions across multiple services (food, travel, shopping, tickets) combined with reel-based engagement data. This cross-domain behavioral data allows us to understand user intent better than single-service platforms, creating a strong data network effect over time. |
Our defensibility comes from the combination of multi-service data, reel-based interaction signals, and AI-driven trust systems. As more users engage, our models improve through richer behavioral data that competitors in single verticals cannot replicate. Additionally, our AI-based content verification system adds a trust layer that differentiates us from traditional discovery platforms. |
We evaluate performance using engagement metrics (CTR, watch time, conversion rate), recommendation accuracy, and user retention. For AI trust systems, we track precision/recall in detecting misleading or synthetic content. System reliability is monitored through latency, uptime, and consistency of recommendations. |
We follow a privacy-first approach by minimizing data collection, anonymizing user data, and securing it through encryption. Sensitive data is handled with strict access controls. We are aligned with Indian data protection guidelines and ensure transparency in how user data is used. |
Potential risks include evolving data privacy regulations and compliance requirements for digital payments and e-commerce. We mitigate this by designing flexible systems that can adapt to regulatory changes and by ensuring compliance with relevant standards. |
At 10x scale, challenges may arise in real-time recommendation latency and model inference efficiency. To address this, we are designing our system with scalable cloud infrastructure, model optimization, and distributed processing to maintain performance at scale. |
Currently, the founding team is building and integrating AI capabilities, with strong technical expertise in development and system design. As we scale, we plan to expand with dedicated AI/ML specialists. |
We use a combination of publicly available datasets, platform-generated data from user interactions, and curated content datasets. As the platform grows, our primary advantage will come from proprietary user interaction data. |
We use continuous feedback loops from user interactions, A/B testing, and model retraining. As more data is collected, models are refined to improve personalization, accuracy, and trust detection. |
We are initially focused on India, starting with campus ecosystems, and plan to expand globally once the model is validated. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/OiczWjzKXHk8-pitchdeck_Gozy.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/OiczWjzKXHk8-pitchdeck_Gozy.pdf |
https://drive.google.com/file/d/1dxDsQv_NQ3y4LsoidnFtX5N_BGUsHebk/view |
Yes. Gozy is built with the mission of simplifying digital life by reducing fragmentation and making everyday services more accessible, efficient, and trustworthy. Today’s users—especially students and young professionals—struggle with managing multiple apps, scattered information, and inefficient digital journeys.
Gozy addresses this by creating a unified platform that not only improves convenience but also enhances digital inclusion. We also aim to empower small and local businesses by giving them a content-driven discovery platform where they can reach users organically without heavy marketing spend.
Our AI-powered trust system further contributes to a safer digital ecosystem by reducing misleading or synthetic content, ensuring users make informed decisions. Overall, Gozy is designed to make digital experiences simpler, more transparent, and more equitable. |
Yes – woman co-founder (Nikshitha) and team from an emerging startup ecosystem beyond major metro hubs. |
NA |
checked |
| Mar 26, 2026 @ 2:53 PM |
Prathwi Jain |
prathwi@roavai.com |
https://www.linkedin.com/in/prathwi-jain-raj1729?utm_source=share&utm_campaign=share_via&utm_content=profile&utm_medium=ios_app |
http://www.roavai.com |
+918618035867 |
Co-founder |
Na |
2 |
First principal thinking. Passionate about technology |
Roavai |
www.roavai.com |
Manglore |
Sparking human curiosity with robotics |
Customer Problem: Children average high screen time, spiking to 6 hours daily by age 10. Existing smartphones lack safety and lead to passive consumption, hindering curiosity and cognitive growth. • Technology Limitations: Current robotic solutions rely heavily on cloud computing, resulting in slow edge use, severe privacy risks, and internet dependency. Furthermore, existing hardware solutions are highly expensive. |
• 100% Offline Edge AI: Ensures zero latency and absolute data privacy by processing all interactions locally.
• Curiosity "Hooks": Transforms passive screen time into active learning by ending answers with engaging questions.
• Autonomous Play: Built-in object detection and indoor navigation encourage physical, room-scale interaction.
• Proprietary Lean Tech: Custom Roav RTE and Mixture of Experts (MoE) architectures run complex AI efficiently on low-cost chips.
• Wini Parent App: Provides parents with total control over playtime limits and tracks daily educational progress. |
• For Children (Cognitive Growth & Engagement): Wini replaces passive, directionless screen time with interactive, screen-free educational play. Instead of simply providing flat facts like a standard search engine, Wini ends every answer with a conversational "hook". This compels the child to ask more questions, actively deepening their discovery and natural curiosity. • For Parents (100% Peace of Mind & Privacy): Because Wini's intelligence runs entirely on the edge, it guarantees absolute data privacy—voice and visual data never leave the device to go to the cloud. Additionally, the Wini Parent App gives parents total tension-free control by allowing them to set playtime limits and track their child's daily educational progress |
MVP |
Testimonials |
• B2C (Primary): Tech-aware working parents of 3-to-7-year-olds in India, the US, UK, and EU. Behaviorally, they are "price-conscious" middle-class buyers who want safe, screen-free STEM tools but are priced out of the $300+ luxury robot market. • B2B/B2G (Secondary): Organizations focused on childcare, special education, or accessibility. |
2.3 billion |
650 million |
5 million |
Hardware plus software |
Miko,loona |
Kisoks mall , online influencer, therapy clinic , school session |
Na |
To become a global leader in privacy-first, edge-native consumer robotics, expanding from educational companions to broader household AI assistance. |
Yes |
0 |
Yes |
To successfully transition Wini from a TRL-4 prototype to a mass-manufactured, commercial-ready product, we are seeking the following strategic support from Avinya |
Assistance with securing AVINYA Seed support and navigating our upcoming Startup India Seed Fund Scheme (SISFS) application to secure our ₹1 Crore runway. We also seek warm introductions to early-stage, deep-tech venture capital networks. |
Yes |
Yes |
Core engine |
Na |
Na |
Na |
Na |
Na |
Na |
Na |
Na |
Na |
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/7TBTWIWSUdRf-Roavai_Elevate_Pitch-Deck-2b.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/7TBTWIWSUdRf-Roavai_Elevate_Pitch-Deck-2b.pdf |
http://na |
To build robotics and AI systems that are deeply beneficial and integrate naturally into everyday life.
We focus on first-principles engineering, careful iteration, and human-centered design — ensuring that intelligence is deployed responsibly and transparently.
By making robotics accessible, trustworthy, and adaptive, we aim to expand access to knowledge, assistance, and opportunity for everyone.
Ultimate goal of ROAVAI is to elevate the human experience and enable us to explore the universe to its fullest potential. |
No |
Na |
checked |
| Mar 25, 2026 @ 10:07 PM |
Dimple Sharma |
dimple@compliancesutras.ai |
https://www.linkedin.com/in/dimplesharma111/ |
https://regtechcomply.ai/ |
+91 8928466952 |
- Dimple is a CA and Finance Professional having worked with GE, Genpact in Finance. - Abhishek is also CA and with 2 decades of working in Banks, NBFCs and Fintech in India, Singapore, UK, Sri Lanka and handling DIFC Compliances for Indian Bank |
- Corporate working
- CA
- Professional domain expertise
- Related |
2 |
- Combined working experience of 25+ years
- Domian expertise |
Compliance Sutras Technologies Private Limited |
https://regtechcomply.ai/ |
Bengaluru - rural |
Compliance Sutras is a women-founded, AI-SaaS-based RegTech company transforming regulatory compliance for the Banking, Financial Services, and Insurance (BFSI) sectors. |
Incomplete & inappropriate methods adopted by Financial Entities to comply with the regulatory & compliance requirements leading to heavy penalties by regulators, business disruption and reputational damage. |
At Compliance Sutras, we offer an AI-powered RegTech platform that automates regulatory compliance and operational risk management.
Our solution addresses the core compliance needs of BFSI institutions, including:
Risk and Control Self-Assessment (RCSA): Automated tools that help assess and mitigate risks.
Incident and Breach Reporting: Enables timely reporting of breaches to regulatory bodies, ensuring swift action.
Regulatory Circular Tracking: A centralized system that tracks and updates regulatory circulars for ongoing compliance.
Our platform supports end-to-end compliance, including risk assessments, incident reporting, and regulatory tracking, helping institutions stay compliant with regulatory frameworks while mitigating operational risks. |
- We aim to reduce compliance processing times for enterprises by up to 80% through automation and streamlined workflows. - Our platform is designed to lower compliance costs and minimize manual errors, directly improving operational efficiency for clients. - By enabling faster onboarding and real-time regulatory updates, we help financial institutions and fintechs accelerate customer acquisition and maintain regulatory readiness. - We support measurable improvements in ESG reporting and green finance compliance, helping organizations meet sustainability and net-zero targets. - Our solution enhances transparency and auditability, strengthening trust with regulators and stakeholders. |
MVP |
Pilots |
Banks, NBFCs, Fintech, Insurance, and Securities firms. Our platform is designed for key decision-makers in financial institutions, including: CXOs: Strategic leaders who need an overarching view of compliance risks. CCOs: Primary users responsible for tracking regulatory requirements and managing compliance risks. CROs: Responsible for assessing, monitoring, and mitigating operational risks using our AI-enabled tools. |
USD 38 BN |
USD 1.3 BN |
USD 650 Mn |
Standalone SaaS Platform: AI-enabled compliance automation for banks, NBFCs, fintech, and insurance firms. SaaS + Consulting: Integration of expert advisory services with the platform for tailored compliance strategies. Complete Digital Transformation: Comprehensive solutions, including cross-border regulatory implementations and training. Revenue generation is anchored in a dual structure of one-time implementation fees and recurring subscription-based income, ensuring sustained client engagement. Our model aligns with evolving regulatory demands, leveraging AI and machine learning to offer real-time insights, predictive analytics, and risk mitigation |
Complinity, Alphaplus technologies, Lexi-nexis, ServiceNow, Alphaplus, AuditBoard |
Networking, References, WOM |
Demo, Pilot, UAT, Live - B2B |
- Best RegTech Company from Bharat - catering all regulatory needs at ONE Stop global integated solution. |
Incorporated |
Bootstrapped |
Yes |
Funding |
Funding, References and Growth |
Yes |
Yes |
Core engine |
Gemma 3 Open Source trained with regulatory data set - ComplanceSutrasOS model is our internal trained model for all regulatory needs. |
- Regulatory training data set |
- Backed by 20+ years of regulatory exposures, it meets the regulators' expectations |
Our ComplianceSutras OS Model is propriety model for Regtech {Based on the UAT Results from Axis Bank implementation}
Accuracy Score: 90% (±3% confidence interval).
Here are the following Metric Assessment:
• Total Obligations Extracted - 247 obligations (IDs 1-247)
• Circular Structure Coverage - Parts A-F + 26 Annexures
• Core Chapters Covered ~90% of substantive regulatory content |
As per DPDP Act/Rules |
None |
Processing Power to match the volume of data to be handled. Nothing else. |
Yes |
Yes. All open source. All internal training instructions and training knowledge is experienced based prop know-how. |
Yes, Indeed |
Both. Build in Bharat for the World.
Target - India, Dubai, Africa and global. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/FnwR8oboAvi8-RegComply_2025_V.1_NPCI-compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/FnwR8oboAvi8-RegComply_2025_V.1_NPCI-compressed.pdf |
https://youtu.be/foWjxx7Q-tA |
Mission - To democratize the regulatory matter with simple, easy and profound solve
Impact - Reduction in cost, manpower and efficency plus by 80% |
NA |
Karnataka Startup Mission |
checked |
| Mar 25, 2026 @ 9:39 PM |
Dimple Sharma |
dimple@compliancesutras.ai |
https://www.linkedin.com/in/dimplesharma111/ |
https://regtechcomply.ai/ |
+91 8928466952 |
- Dimple is a CA and Finance Professional having worked with GE, Genpact in Finance. - Abhishek is also CA and with 2 decades of working in Banks, NBFCs and Fintech in India, Singapore, UK, Sri Lanka and handling DIFC Compliances for Indian Bank |
- Corporate Background
- CA
- Working professionals
- Related |
two |
- Corporate working experince close to 25+ years together
- Professional domian expertise |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 25, 2026 @ 9:32 PM |
Sagar mainkar , Abhilash anandan , Suralal Suresh , Vijay Lingapuram , Anish agarwal |
info@theprojectskin.com |
https://www.linkedin.com/in/sagar-dilip-mainkar-7160252b0/ |
https://theprojectskin.com/ |
9071144458 |
Our team combines expertise in AI, dermatology, marketing, finance, and business execution. * Sagar Mainkar (CEO) – Over a decade of experience in product and building health & wellness brands. * Suralal Suresh (CTO/AI & Technology Lead) – Experienced in AI and technology, leading the platform’s architecture and development. * Abhilash (CMO/Marketing & Branding) – Experienced in marketing, brand building, and growth strategies. * Dr. Vijay (Head of R&D & Dermatology) – Dermatologist leading research, clinical validation, and product efficacy. * Anish Agarwal (CFO) – Expert in financial strategy, operations, and scaling business growth. Together, we combine technology, clinical expertise, and business acumen to build a scalable, high-impact skincare platform. |
Our core team came together through a shared vision of transforming personalized skincare. Sagar, Suralal, Abhilash, Dr. Vijay, and Anish connected over industry networks and mutual interest in health tech. We have been working closely as a team for over two years, collaborating on product development, AI innovation, and clinical validation to bring Project Skin from concept to active beta users. |
2 |
Our biggest strength is the combination of domain expertise, technical capability, and execution speed. We have deep knowledge across skincare, AI, marketing, and business operations, allowing us to build solutions that are clinically relevant and user-centric. At the same time, we execute quickly, moving from AI prototype to 250+ skin scans and validation with 100+ active beta users. This blend of expertise and rapid execution enables us to deliver a scalable, high-impact skincare platform efficiently. |
Nureya Beauty and Wellness Pvt Ltd |
https://theprojectskin.com/ |
Bangalore , Karnataka |
AI-powered personalized skincare platform that analyzes skin data to deliver clinically relevant insights and tailored treatment recommendations. |
Consumers struggle to identify the right skincare solutions due to generic products, inconsistent advice, and lack of personalized guidance, leading to ineffective results and low trust. |
We provide an AI-powered platform that analyzes individual skin conditions using images and user data to deliver personalized skincare recommendations. The system combines computer vision with dermatology expertise to assess concerns like acne, pigmentation, and sensitivity, then suggests the most effective products and routines. Over time, it learns from user feedback and outcomes, continuously improving accuracy and personalization, creating a data-driven, trusted skincare journey. |
Our defensibility comes from proprietary data, deep personalization, and a closed feedback loop. We are building a unique dataset of user skin images, concerns, and outcomes, creating a strong data moat and improving model accuracy over time. The system continuously learns from user feedback and results, making recommendations more precise with scale. We combine AI with dermatology-backed logic, ensuring clinical relevance, and deliver an end-to-end experience from diagnosis to tracking, driving retention and trust. |
MVP |
Testimonials |
Our target customers are individuals seeking effective, personalized skincare solutions—primarily urban adults aged 18–45 who are health-conscious, tech-savvy, and frustrated with generic products. We also target dermatology clinics and wellness brands looking for data-driven insights to improve treatment outcomes and customer engagement. |
For Project Skin, the Total Addressable Market (TAM) is the global skincare and wellness segment seeking personalized solutions. Consumer segment: Urban adults (18–45) using skincare products—over $150B annually, with personalized/AI-driven skincare growing rapidly. Professional segment: Dermatology clinics and wellness brands integrating AI diagnostics—an additional $10–15B market. Combined, the TAM exceeds $160B, driven by rising demand for personalized, effective skincare solutions. |
For Project Skin, the Serviceable Available Market (SAM) focuses on segments we can realistically reach in the near term: Urban adults in India (18–45) seeking personalized skincare solutions, estimated at $2–3B annually, and dermatology clinics and wellness brands in India adopting AI-driven diagnostics and personalized treatment tools, representing $200–300M. Combined, the SAM is around $2.2–3.3B, the portion of the global market we can serve with our current go-to-market strategy. |
For Project Skin, the Serviceable Obtainable Market (SOM) represents the portion of the SAM we can realistically capture in the first few years. Targeting urban Indian adults (18–45) and early-adopter dermatology clinics, our initial SOM is estimated at $50–70 million based on expected adoption rates, marketing reach, and pilot traction. |
Our revenue model combines direct-to-consumer subscriptions, product partnerships, and B2B licensing. * D2C Subscriptions: Users pay for personalized skincare plans and AI-driven recommendations. * Product Partnerships: Commission or revenue share with skincare brands when users purchase recommended products through our platform. * B2B Licensing: Offer our AI diagnostics and analytics platform to dermatology clinics and wellness brands for a subscription or usage fee. This multi-channel approach ensures diversified, recurring revenue and scalability as we grow. |
For Project Skin, the Serviceable Obtainable Market (SOM) represents the portion of the SAM we can realistically capture in the first few years. Targeting urban Indian adults (18–45) and early-adopter dermatology clinics, our initial SOM is estimated at $50–70M, based on expected adoption rates, marketing reach, and pilot traction. |
We acquire customers through a mix of **digital marketing, partnerships, and clinical channels**. * **Digital Marketing:** Targeted social media campaigns, influencer collaborations, and content marketing to reach urban adults seeking personalized skincare. * **Partnerships:** Collaborations with dermatology clinics, wellness centers, and skincare brands to integrate our AI platform and gain credibility. * **Direct Engagement:** Webinars, workshops, and events to educate users on personalized skincare and drive trial adoption. This multi-channel approach ensures both rapid user acquisition and strong trust in our platform. |
Our go-to-market strategy focuses on rapid adoption through digital campaigns targeting urban consumers, strategic partnerships with dermatology clinics, and brand-led marketing to build trust and credibility, enabling scalable growth. |
Our long-term vision is to build a global category leader in AI-driven personalized skincare. We aim to transform how consumers and dermatology clinics approach skincare by providing data-backed, clinically validated recommendations that improve outcomes and build trust. By combining proprietary skin data, AI insights, and clinical expertise, we plan to create a scalable platform with strong network effects, high user retention, and multiple monetization opportunities across B2C and B2B channels, positioning Project Skin as the go-to solution for personalized skincare worldwide. |
incorporated o 01/09/2025 |
0 |
Yes |
We are applying to Avinya to leverage their **expert mentorship, industry network, and access to resources** that can accelerate Project Skin’s growth. Avinya’s support will help us refine our AI platform, validate our clinical and consumer strategies, and scale rapidly while building strong investor and partner connections. This aligns perfectly with our goal of becoming a leading AI-driven personalized skincare platform. |
During the program, we aim to **accelerate product development, validate our AI models, and scale user adoption**. Specifically, we plan to: * Refine our AI diagnostics and recommendation engine with expert guidance. * Strengthen clinical validation and regulatory compliance. * Expand our go-to-market strategy and grow early adopter traction. * Build strategic partnerships and investor readiness to support rapid scaling. These milestones will position Project Skin for sustainable growth and market leadership. |
yes |
Yes |
Core engine |
Project Skin’s AI architecture is designed for scalable, accurate, and personalized skin analysis. It consists of three core layers: 1. Data Ingestion Layer: Collects high-quality skin images and user metadata through mobile app uploads, ensuring standardized inputs for analysis. 2. Processing & Analysis Layer: Uses computer vision models to detect skin conditions such as acne, pigmentation, and sensitivity. Dermatology-informed algorithms map features to severity and treatment recommendations. 3. Personalization & Feedback Layer: Combines AI predictions with user feedback and historical outcomes to continuously improve accuracy and tailor product and routine suggestions. The architecture is modular, allowing easy integration with partner platforms and future expansion into predictive analytics, treatment outcome tracking, and B2B clinic solutions. |
Project Skin’s proprietary data advantage comes from our growing dataset of real user skin images, conditions, and treatment outcomes, collected through our platform and validated with dermatologists. This data allows our AI models to continuously improve in accuracy and personalization. Unlike generic skincare platforms, our dataset combines clinical insights with real-world usage, creating a strong barrier to entry and enabling highly targeted, effective recommendations that competitors cannot easily replicate. |
Project Skin’s digital twin feature creates a virtual representation of each user’s skin using AI-driven analysis of images and historical data. This digital twin allows us to simulate treatment outcomes, predict how different products or routines will affect the user’s skin over time, personalize recommendations based on the user’s unique skin profile and response patterns, and track progress to adapt suggestions dynamically. This feature strengthens engagement, improves accuracy of recommendations, and creates a defensible data-driven advantage over competitors. |
We evaluate Project Skin’s AI model performance and reliability using a combination of accuracy, precision, recall, and clinical validation metrics.
Accuracy measures the percentage of correct skin condition predictions against dermatologist-labeled data.
Precision and recall ensure the model correctly identifies skin concerns without over- or under-detection.
F1 score balances precision and recall to evaluate overall prediction quality.
AUC-ROC assesses the model’s ability to distinguish between different skin conditions.
Clinical validation involves dermatologists reviewing model recommendations to ensure medical relevance and safety.
User feedback loop continuously monitors outcomes and feedback to refine predictions and maintain reliability over time.
This combination ensures our AI is both technically robust and clinically trustworthy. |
Project Skin handles data privacy, compliance, and security through strict protocols and industry-standard practices.
Data privacy: User data is anonymized and encrypted during collection, storage, and transmission. Personal identifiers are separated from skin images and analytics.
Compliance: We follow applicable regulations such as GDPR and India’s data protection guidelines, ensuring informed consent and user control over their data.
Security: Our systems use end-to-end encryption, secure cloud storage, and regular vulnerability assessments to prevent unauthorized access.
Access control: Role-based permissions ensure only authorized team members and clinical partners can access sensitive data.
Continuous monitoring: Security audits, penetration testing, and AI model monitoring maintain ongoing protection and compliance.
This approach ensures that user trust, legal compliance, and data integrity are maintained at every stage. |
Yes, there are potential regulatory risks, primarily related to data privacy, medical device classification, and skincare claims.
Data privacy: Non-compliance with laws like GDPR or India’s data protection regulations could lead to fines or restrictions.
Medical device regulations: If our AI recommendations are classified as a medical device, additional approvals or certifications may be required.
Product claims: Making unverified efficacy claims could attract scrutiny from consumer protection authorities.
We mitigate these risks by working closely with regulatory experts, ensuring clinical validation of AI outputs, following strict data protection protocols, and keeping all marketing and product claims evidence-based. |
If we scale 10x, the main challenges in our AI system would be data processing, model latency, and feedback loop management.
Data processing: A sudden surge in image uploads and user data could strain our ingestion and preprocessing pipelines.
Model latency: Real-time analysis might slow down if infrastructure is not scaled efficiently, affecting user experience.
Feedback loop management: Continuous learning relies on high-quality labeled data; a rapid influx of new users could introduce noisy data and reduce model accuracy if not managed properly.
Integration with partners: Scaling across clinics and brands requires robust API handling and monitoring to prevent downtime or errors.
We plan to address these by optimizing cloud infrastructure, implementing scalable pipelines, and maintaining strict data quality checks, ensuring the AI remains reliable even at higher volumes. |
Yes, we have an in-house AI/ML team led by Suralal Suresh, our CTO. The team has expertise in computer vision, deep learning, predictive modeling, and AI-driven personalization. They are responsible for developing and optimizing our skin analysis models, integrating dermatology-informed algorithms, managing continuous learning from user feedback, and ensuring the AI system scales efficiently while maintaining accuracy and reliability. |
We are currently using a combination of proprietary and licensed datasets. Proprietary data is collected through our platform, including over 250 skin scans and user feedback, validated by dermatologists, and forms the core of our AI training. Licensed data includes selected publicly available dermatology image datasets and research datasets to augment rare skin conditions and improve model robustness. By combining owned and licensed data, we ensure both high accuracy and comprehensive coverage across diverse skin types and conditions. |
We plan to continuously improve model performance through iterative learning, user feedback, and data expansion. The AI system collects ongoing user data, treatment outcomes, and skin images, which are validated by dermatologists to maintain accuracy. Regular model retraining and hyperparameter optimization ensure the system adapts to new skin types, conditions, and trends. We also monitor performance metrics like accuracy, precision, and recall and use real-world feedback to fine-tune predictions, ensuring the AI becomes more precise and personalized over time. |
We are building for both India and global markets. Our initial focus is on urban Indian consumers and dermatology clinics to validate the platform and refine personalization. Once scaled, we plan to expand globally, targeting markets with high demand for personalized skincare solutions, leveraging our AI, proprietary data, and clinical validation to establish a competitive advantage worldwide. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/AtU7IAfSNofa-Project-Skin-Pitch-Deck-190326-India.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/AtU7IAfSNofa-Project-Skin-Pitch-Deck-190326-India.pdf |
https://drive.google.com/drive/folders/14x455vdXw5dNVaNfZkCW5keWfMJnv5Ca?usp=drive_link |
Project Skin is a mission-driven, impact-focused startup with a clear market opportunity. Our goal is to provide personalized, data-backed skincare solutions that improve outcomes, reduce trial-and-error, and increase user trust. At the same time, we empower dermatologists and wellness professionals with AI tools that enhance treatment effectiveness and patient engagement. By combining proprietary data, AI, and clinical expertise, we create a scalable platform with strong retention, network effects, and growth potential. |
No, our core team does not identify as part of an underrepresented group in tech or startups. |
The startup Economist |
checked |
| Mar 25, 2026 @ 8:39 PM |
Sagar mainkar , Abhilash anandan , Suralal Suresh , Vijay Lingapuram , Anish agarwal |
sagar@theprojectskin.com |
https://www.linkedin.com/in/sagar-dilip-mainkar-7160252b0/ |
https://theprojectskin.com/ |
9071144458 |
Sagar product , Abhilash marketing Suralal Technology , Dr vijay Head of Dermatology Anish CFO |
8 Months |
two |
Our biggest strength is the combination of deep user insight and fast execution.
We’ve spent significant time understanding real skincare behavior by working closely with users and practitioners, which helps us build solutions based on actual needs rather than assumptions. This leads to stronger product relevance and user trust.
At the same time, we execute quickly. From building the AI prototype to completing 250+ scans and validating with 100+ active beta users, we’ve demonstrated the ability to move efficiently from idea to real-world validation.
This combination of customer-first thinking and rapid iteration gives us a strong advantage in building a scalable, high-impact product. |
Nureya Beauty and Wellness Pvt Ltd |
https://theprojectskin.com/ |
Bangalore , Karnataka |
AI-powered personalized skincare platform that analyzes skin data to deliver clinically relevant insights and tailored treatment recommendations. |
Consumers struggle to identify the right skincare solutions due to generic products, inconsistent advice, and lack of personalized guidance, leading to ineffective results and low trust. |
We are building an AI-powered platform that analyzes individual skin conditions through image-based diagnostics and user data to deliver personalized skincare recommendations.
The system combines computer vision with dermatological insights to assess concerns like acne, pigmentation, and sensitivity, and then maps users to the most effective products and routines.
Over time, the platform learns from user feedback and outcomes, continuously improving accuracy and personalization, creating a data-driven, trusted skincare journey rather than trial-and-error. |
Our defensibility comes from proprietary data, deep personalization, and a closed feedback loop. We are building a unique dataset of user skin images, concerns, and outcomes, creating a strong data moat and improving model accuracy over time. The system continuously learns from user feedback and results, making recommendations more precise with scale. We combine AI with dermatology-backed logic, ensuring clinical relevance, and deliver an end-to-end experience from diagnosis to tracking, driving retention and trust. |
two |
Pilots |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 25, 2026 @ 8:00 PM |
Mutturaj Goudra |
mutturajshramic@gmail.com |
https://www.linkedin.com/in/mutturaj-goudra-2a27653a1/ |
https://www.shramic.com/ |
9535455717 |
I am the CEO and Founder of this Startup I am currently pursuing BE AIML branch from the JSS Academy Of Technical Education. |
We met through our academic and professional network, primarily during our engineering journey and through shared interest in solving real-world problems using technology. Over time, we aligned on a common vision of building impactful solutions in the agriculture and rural employment sector, which led to the formation of our startup.
Our collaboration started informally through discussions, research, and small projects, and has now evolved into a structured team working together on Shramic Networks. We have been working together for several months, during which we have built strong coordination, clear role distribution, and a shared commitment towards our mission.
Our team brings together diverse skills across AI/ML, software development, operations, product design, and community building, allowing us to effectively collaborate and execute across multiple domains. |
two |
Our biggest strength as a team is our shared vision combined with complementary skill sets. Every member brings expertise in areas such as AI/ML, software development, product design, operations, and community building, which allows us to approach problems in a well-rounded and practical way.
What truly sets us apart is our execution mindset. We focus on building, testing, and iterating quickly rather than just discussing ideas. Even at an early stage, we operate with strong ownership where each member is responsible for outcomes.
We also maintain clear communication and alignment, which helps us move fast without confusion. Our ability to adapt, learn quickly, and stay committed to solving real problems in the agriculture and rural sectors makes us a focused and resilient team. |
SHRAMIC NETWORKS PRIVATE LIMITED. |
https://www.shramic.com/ |
Hanamsagar |
Building an AI-powered digital ecosystem that connects farmers, workers, and service providers to create fair pricing, employment, and growth opportunities in rural India. |
We are solving the lack of transparent pricing, reliable employment, and access to services in the agriculture sector, where farmers and rural workers often face inefficiencies, exploitation, and limited market access due to disconnected systems and information gaps. |
We are building an AI-powered digital ecosystem that connects farmers, workers, service providers, and experts on a single platform to streamline agricultural operations. Our solution uses intelligent pricing models, data-driven matching, and real-time insights to ensure fair wages, efficient resource allocation, and better decision-making. |
We focus on agriculture and construction because they are among the largest yet most underserved sectors in rural and semi-urban India, where inefficiencies in pricing, labor access, and service availability directly impact livelihoods. Both sectors face common challenges such as informal pricing, lack of transparency, and fragmented demand and supply. By targeting these areas, we address real-world problems where digital solutions can create immediate and meaningful impact. Our approach is to build a strong foundation in these domains, develop reliable pricing and matching systems, and create a connected ecosystem that improves efficiency and trust. |
two |
one |
Our target customers are farmers, rural and semi-urban workers, and service providers in the agriculture and construction sectors. This includes small and medium farmers, skilled and unskilled laborers, equipment owners, contractors, and local service providers who face challenges in pricing, job access, and resource availability. Additionally, we also serve agri-businesses, nurseries, and individuals or organizations seeking reliable workforce, services, or market connections within these sectors. |
₹20+ lakh crore combined market across agriculture and construction sectors in India, considering labor, services, equipment, and transactions. |
₹2–3 lakh crore market focusing on digitally accessible agriculture and construction services, including labor marketplaces, equipment access, and service platforms in target regions. |
₹500–1000 crore achievable in the next 5–7 years by focusing on selected states and scaling through a network-driven platform approach. |
We follow a multi-stream revenue model including commissions on transactions such as job bookings, equipment rentals, and services, along with subscription plans for service providers and businesses. Additionally, we generate revenue through lead generation, premium listings, and paid expert consultations on the platform. |
Our main competitors are mostly indirect and fragmented platforms that address only parts of the problem, such as agri marketplaces, labor listing platforms, and equipment rental services. Some notable players include platforms like Ninjacart and DeHaat in agriculture, and Urban Company or local contractor networks in services. However, these solutions operate in silos, whereas we are building an integrated ecosystem that combines pricing, employment, services, and expert access in a single platform. |
We acquire customers through a combination of on-ground and digital strategies. This includes community-level outreach in rural and semi-urban areas, partnerships with local networks, farmer groups, and contractors, as well as referral programs to build trust and drive organic growth. We also leverage digital channels such as social media, targeted campaigns, and early user success stories to increase adoption, while ensuring strong local engagement to create network effects and long-term retention. |
We follow a cluster-based go-to-market strategy, starting with selected regions where we can build strong supply and demand simultaneously. Our focus is on creating high local density by onboarding farmers, workers, and service providers within specific clusters to ensure quick matching and visible value. We prioritize on-ground execution through partnerships with local communities, organizations, and early adopters to build trust and drive initial traction. Once strong network effects are established in a region, we expand gradually to nearby areas, replicating the model in a scalable way. |
Our long-term vision is to build a unified Shramic Community that serves as a digital backbone connecting farmers, workers, service providers, and businesses across agriculture and construction sectors. We aim to organize and formalize these largely unstructured markets by bringing transparency, efficiency, and trust into everyday transactions.
At the core of our vision is an AI-driven ecosystem that enables fair and dynamic pricing, reliable employment opportunities, and seamless access to services, equipment, and expert support. By leveraging data and intelligent systems, we want to eliminate inefficiencies such as price exploitation, labor mismatch, and lack of market access that currently affect rural and semi-urban communities.
We envision creating a scalable, network-driven platform where each participant whether a farmer, worker, contractor, or service provider benefits from better opportunities, increased income stability, and improved decision-making. Over time, this ecosystem will evolve into a trusted infrastructure layer that powers workforce and service coordination, ultimately improving livelihoods and contributing to inclusive economic growth at scale. |
YES |
NA |
one |
We are applying to Avinya because it aligns strongly with our mission of building impactful, scalable solutions that address real-world problems at the grassroots level. The focus on innovation, mentorship, and ecosystem support makes it an ideal platform for us to refine our product, validate our approach, and accelerate our growth. At this stage, we are building a complex, high-impact system in the agriculture and construction sectors, and guidance from experienced mentors, along with access to the right network, will help us avoid critical mistakes and move faster with clarity. We also see Avinya as an opportunity to collaborate, learn, and strengthen our execution while staying aligned with our long-term vision of creating a large-scale, inclusive digital ecosystem for rural and semi-urban communities. |
During the programme, we aim to achieve three key outcomes: First, to refine and validate our pricing engine and core platform by improving accuracy, scalability, and real-world reliability through expert mentorship and structured feedback. Second, to build strong early traction by successfully deploying our solution in selected regions, onboarding users, and demonstrating clear value through measurable outcomes like better pricing, faster job matching, and increased efficiency. Third, to strengthen our product, technology, and business model, including architecture, go-to-market strategy, and revenue streams, so we are well-prepared for scaling and future fundraising. Overall, our goal is to move from a developing system to a validated, execution-ready platform with real impact on the ground. |
Yes, we are open to both hybrid and virtual participation. We are flexible and committed to actively engaging in the programme in whichever format is most effective, while ensuring consistent progress and full participation from our team. |
one |
one |
Our AI architecture is designed as a multi-layered, modular system that combines causal inference, graph intelligence, temporal modeling, LLMs, and autonomous agents to enable accurate and fair pricing in agriculture and construction markets.
At the foundation, we start with data ingestion and preprocessing, where structured data (workers, jobs, wages, skills, transactions) and unstructured data (job descriptions, profiles) are cleaned and standardized.
The system is then built across six key layers:
Causal Intelligence Layer (Phase 1)
This layer identifies the true drivers of wages using causal inference models instead of simple correlations. It generates causal multipliers that ensure pricing decisions are fair and explainable.
Graph Intelligence Layer (Phase 2)
We model the ecosystem using a hypergraph knowledge graph that captures complex multi-entity relationships between workers, jobs, employers, and skills. This layer generates embeddings that represent real-world connections.
Temporal Intelligence Layer (Phase 3)
This layer analyzes worker career trajectories over time using sequence models like LSTMs and Transformers. It predicts fair wages based on both historical patterns and current context.
Privacy Layer (Phase 4)
We implement federated learning so models can learn from multiple stakeholders without sharing raw data, ensuring privacy and scalability across employers.
Language Intelligence Layer (Phase 5)
We integrate LLMs to understand unstructured text such as job descriptions and worker profiles. This helps capture hidden signals and also generates human-readable explanations for pricing decisions.
Autonomous Decision Layer (Phase 6)
We use multi-agent systems where AI agents representing workers and employers can negotiate wages based on model predictions, enabling a self-optimizing marketplace.
All these layers are integrated into a unified pipeline:
Data → Causal + Graph + LLM → Temporal Prediction → Agent-based Execution → User Interface
From a technology perspective, the system leverages Python-based ML/DL frameworks, graph neural networks, transformer models, federated learning frameworks, and LLM stacks such as HuggingFace and LangChain.
Overall, our architecture is designed to be explainable, scalable, privacy-preserving, and adaptive, ensuring that pricing decisions are not only accurate but also trustworthy and continuously improving with data. |
We are building a proprietary data advantage through the continuous collection of real-world, ground-level data from agriculture and construction ecosystems. This includes transaction-level data such as wages, job types, worker skills, equipment usage, location-specific demand, and time-based patterns, which are currently highly fragmented and largely unavailable in structured form.
As users interact with our platform, we generate high-quality, contextual, and dynamic datasets that capture relationships between workers, employers, tasks, and outcomes. This allows us to continuously improve our pricing models and matching algorithms.
Additionally, our system combines structured data with unstructured insights such as job descriptions, feedback, and behavioral signals, creating a richer dataset that is difficult to replicate.
Over time, this creates a strong data network effect where more usage improves accuracy, and higher accuracy attracts more users, building a scalable and defensible advantage. |
Our AI component is defensible because it combines multiple advanced approaches such as causal inference, graph-based learning, temporal modeling, and language understanding, making it more accurate and context-aware than traditional systems.
It continuously improves using real-world, ground-level data generated from platform usage, creating a strong data advantage that is difficult for competitors to replicate.
Additionally, the system focuses on explainability and fairness, ensuring that every pricing decision is transparent and trustworthy, which helps build long-term user confidence and adoption. |
We evaluate model performance and reliability using a combination of predictive accuracy, fairness, robustness, and real-world impact metrics.
For prediction quality, we use metrics like MAE (Mean Absolute Error), RMSE (Root Mean Squared Error), and R² to measure how close our predicted prices are to actual outcomes.
To ensure fairness and consistency, we track bias and variance across different worker groups, locations, and job types, along with stability of predictions under similar conditions.
For reliability, we perform cross-validation, out-of-sample testing, and sensitivity analysis to ensure the model performs well on unseen data and is not overly dependent on specific variables.
We also evaluate causal validity by checking whether identified factors truly influence outcomes rather than just correlating with them.
At the system level, we monitor real-world metrics such as job match success rate, acceptance rate of recommended prices, negotiation convergence, and user satisfaction, ensuring that the model is not only accurate but also practical and trustworthy in real scenarios. |
We handle data privacy, compliance, and security through a combination of architecture-level safeguards, strict data practices, and regulatory alignment.
First, we follow a privacy-by-design approach, where sensitive data is minimized, anonymized, and only collected when necessary. Personally identifiable information (PII) is separated and protected using encryption both at rest and in transit.
Second, we implement secure access controls and authentication mechanisms, ensuring that only authorized systems and users can access specific data. All activities are logged and monitored to prevent misuse.
Third, our architecture supports privacy-preserving learning techniques such as federated learning, where models can learn from distributed data without requiring raw data to be centrally shared.
We also ensure compliance with relevant data protection standards and best practices, and we are building our systems to align with evolving regulations in India.
Overall, our focus is to ensure that data is handled responsibly, securely, and transparently, while maintaining user trust and system integrity. |
Yes, there are potential regulatory risks, as we operate in sectors like agriculture, labor, and digital platforms, which are influenced by evolving policies in India.
These include changes in labor laws, platform worker regulations, and compliance requirements around wages, contracts, and worker classification. There may also be regulations related to data protection and privacy, especially as digital platforms handling user data come under stricter frameworks.
Additionally, in agriculture, state-level policies, subsidies, and market regulations can impact how services, pricing, or transactions are structured. In construction, compliance with local labor and contractor regulations is also important.
However, we view this as manageable by designing our platform to be compliant-first and adaptable, ensuring transparency in pricing, proper documentation, and alignment with legal standards. We also plan to continuously monitor policy changes and work with advisors to stay compliant as we scale. |
At 10x scale, the main challenges would come from data complexity, system performance, and model consistency across regions.
First, as data volume and diversity increase, data quality and consistency can become an issue. Noisy, incomplete, or highly localized data may affect model accuracy if not properly standardized and validated.
Second, the graph and temporal models can face computational bottlenecks, as relationships and sequences grow exponentially. This can impact training time, latency, and real-time predictions if not optimized.
Third, maintaining model generalization across different geographies becomes challenging. Pricing dynamics in one region may not directly apply to another, requiring better localization and adaptive models.
Fourth, in a distributed setup, federated learning and synchronization can become slower and more complex, especially with many participants contributing updates.
Finally, ensuring consistent and explainable outputs at scale is critical. As the system grows, small inconsistencies can reduce user trust if not monitored.
To address this, we plan to invest in scalable infrastructure, better data pipelines, regional model tuning, and continuous monitoring to maintain performance and reliability as we grow. |
Yes, we have an in-house AI/ML team with expertise across multiple domains required to build our system end-to-end.
The team works on machine learning, deep learning, and data engineering, including building predictive models, handling large-scale datasets, and designing data pipelines. We also have capabilities in causal inference, graph-based modeling, and time-series analysis, which are core to our pricing engine.
In addition, the team has experience with LLMs and NLP, enabling us to work with unstructured data like job descriptions and user inputs, as well as generate explainable outputs.
On the engineering side, we have strong skills in backend development, APIs, cloud deployment, and MLOps, ensuring models are not just built but also deployed, monitored, and scaled effectively.
Overall, our team combines research-level understanding with practical implementation, allowing us to build a robust, scalable, and production-ready AI system. |
urrently, we are using a combination of synthetic data, publicly available datasets, and internally structured datasets to build and validate our system.
In the early stage, we generated synthetic datasets to simulate real-world scenarios such as worker profiles, job types, wages, and transactions. This helps us test different phases of the pricing engine and validate model behavior without privacy risks.
We also use publicly available data sources related to agriculture, labor trends, and regional factors to enrich our understanding and provide baseline signals where needed.
Alongside this, we are building our own internally owned datasets by structuring and standardizing data models for workers, jobs, skills, and transactions. As we move forward and onboard users, this will evolve into proprietary, real-world data owned by us.
So currently, the data is a mix of synthetic (self-generated), public (open-source), and internally developed datasets, with a clear transition toward fully owned proprietary data as the platform scales. |
We plan to continuously improve model performance through a combination of data, feedback loops, and iterative model refinement.
First, as the platform grows, we will collect real-world, high-quality data from transactions, user interactions, and outcomes, which will be used to retrain and improve models regularly.
Second, we incorporate feedback loops such as price acceptance rates, job completion outcomes, and user feedback to evaluate how well the model performs in real scenarios and adjust accordingly.
Third, we follow an iterative experimentation approach, where we continuously test model improvements using A/B testing, validation pipelines, and performance benchmarking before deploying updates.
We also focus on feature enhancement, adding new signals such as regional trends, behavioral patterns, and contextual factors to improve accuracy.
Finally, we maintain strong monitoring and MLOps practices, including performance tracking, drift detection, and periodic retraining, ensuring the model stays reliable and adapts to changing market conditions over time. |
We are currently focused on India, as it offers a large, high-impact opportunity in agriculture and construction with significant inefficiencies to solve.
Our approach is to build and validate the system deeply in the Indian context, including regional pricing dynamics, language diversity, and ground-level workflows.
At the same time, the core architecture and models are designed to be scalable and adaptable, allowing us to expand to other global markets with similar characteristics once we achieve strong product-market fit in India. |
|
|
Yes, we are a deeply mission-driven and impact-focused startup.
Our core mission is to uplift rural and semi-urban communities by solving fundamental problems in agriculture and construction such as unfair pricing, lack of reliable employment, and limited access to services. These challenges directly affect livelihoods, income stability, and overall economic growth for millions of people.
We are building a platform that brings transparency, efficiency, and trust into these sectors by enabling fair pricing, better job matching, and seamless access to resources and expertise. Our goal is not just to digitize existing systems, but to organize and formalize unstructured markets in a way that creates long-term, sustainable impact.
Beyond business outcomes, we measure success through real-world improvements such as increased worker earnings, better opportunities for farmers, reduced inefficiencies, and stronger local economies.
Ultimately, we aim to build a scalable ecosystem that drives inclusive growth and empowers underserved communities at scale. |
Yes, we come from a rural and semi-urban background and are building in sectors that are often overlooked in mainstream tech. Our journey is closely connected to the communities we are trying to serve, which gives us strong on-ground understanding and insight into real problems. This perspective helps us design solutions that are practical, inclusive, and accessible, especially for users who are typically underrepresented in digital platforms. |
Not referred – we discovered the program through our own research and applied directly. |
|
| Mar 25, 2026 @ 6:55 PM |
SUBRAMANI ARUMUGAM |
ceo@analyticsavenue.in |
https://www.linkedin.com/in/sivani-srinivasan/ |
http://analyticsavenue.in/ |
9677298268 |
CEO & DATA SCIENCE |
We initially connected through our shared professional interests in AI and data analytics, where our complementary skill sets aligned naturally. Over time, this evolved into a collaborative working relationship. We have been working together for over [X months/years], during which we have successfully collaborated on multiple projects, strengthening both our execution capability and strategic vision. |
one |
Our biggest strength as a team lies in the synergy between our expertise in data analytics and our understanding of the EdTech ecosystem. We combine technical capabilities with a strong focus on outcome-driven learning, enabling us to build solutions that bridge the gap between theory and real-world application. Each team member brings complementary skills across AI, analytics, product development, and strategy, allowing us to execute efficiently. Our collaborative approach, adaptability, and hands-on experience with real projects help us continuously refine our offerings and deliver measurable value to both learners and enterprises in a scalable and impactful way. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 25, 2026 @ 6:49 PM |
Vikram B N, Vinay kumar R, Amit Agrawal, Hemant Sharma |
vikram@qualifyme.ai |
https://www.linkedin.com/in/vikrambn/ |
https://www.qualifyme.ai |
9886132601 |
Vikram CEO & Co-Founder, Vinay CTO & Co-Founder, Amit CPO & Co-Founder, Hemant CFO & Co-Founder |
Vinay and I previously co-founded Avadhuta Technologies, where we’ve been working closely on building and delivering software products. In parallel, Amit and Hemant are the founders of VNurt, a company focused on HR recruitment services and product development.
We came together through this partnership, combining our strengths—our experience in building technology and their deep expertise in recruitment. What started as a collaboration to solve hiring inefficiencies evolved into building QualifyMe together.
Since then, we’ve been working closely as a team across product, engineering, and hiring workflows, with strong alignment on both execution and long-term vision. |
All |
Our biggest strength is the combination of deep domain expertise and strong execution capability.
On one side, Amit and Hemant bring hands-on experience in recruitment through VNurt—they understand how hiring actually works on the ground, the inefficiencies in screening, and what recruiters truly care about. On the other side, Vinay and I bring strong product and engineering experience from Avadhuta Technologies, with a focus on building scalable, AI-driven systems.
This allows us to not just identify real problems, but also build and iterate on solutions quickly. We’re able to move from insight → product → feedback → improvement in tight loops, which has been critical in shaping QualifyMe.
Additionally, we have clear ownership across functions and a high level of alignment on vision, which helps us execute fast without friction. |
QualifyMe Technologies |
https://www.qualifyme.ai |
Bengaluru |
AI-powered, simulation-based assessments that help companies identify the right candidates faster—without guesswork. |
Hiring today relies on low-signal resumes and interviews, making it hard to accurately assess real skills and predict job performance. |
We provide AI-powered, simulation-based assessments that evaluate candidates through real-world tasks instead of resumes and guesswork.
Recruiters can set up role-specific evaluation flows, and candidates are assessed across structured simulations, coding rounds, and AI-driven interviews. Our system analyzes responses to generate clear, explainable insights on skills, role fit, and performance—helping teams make faster and more accurate hiring decisions. |
Our defensibility comes from how we combine real-world simulations with structured, explainable AI evaluation. Instead of generic tests, we break down each role into granular, evaluatable signals and assess candidates through task-based simulations that mirror actual job scenarios. This creates a much higher-quality signal compared to resumes or standard assessments. On top of that, our AI doesn’t just score—it provides reasoning, impact analysis, and role-fit insights, making the output usable for decision-making, not just filtering. Over time, this approach compounds: as more candidates are assessed, we build a proprietary dataset linking simulation performance to hiring outcomes, strengthening our evaluation models and creating a strong data moat. |
Revenue |
two |
Startups and mid-sized companies that hire at scale—especially for fresher and early-career roles—where assessing real skills quickly and accurately is critical. |
Global recruitment and assessment market, including hiring software and services, is ~$300B+. |
Companies actively using or open to adopting AI-driven hiring and assessment tools—estimated at ~$15–20B globally. |
Initial focus on startups and mid-sized companies hiring at scale in India and similar markets—~$200–500M. |
SaaS subscription model with tiered plans (Basic, Professional, Enterprise), priced based on number of assessments and features. Additionally, we generate revenue through usage-based pricing for high-volume hiring and offer recruitment support services through our partner network. |
Our competitors fall into three categories: 1. Assessment platforms (direct competitors): HackerRank, Codility, TestGorilla — focused on standardized testing, primarily for technical or aptitude screening. 2. AI interview and screening tools: HireVue, Pymetrics — automate interviews and behavioral assessments but often lack role-specific simulations. 3. Traditional hiring methods (indirect competitors): Resumes, manual screening, and recruiter-led interviews, which still dominate hiring despite being low-signal. |
Outbound-led sales (cold emails and LinkedIn outreach) targeting hiring managers and TA heads, supported by inbound channels like LinkedIn content, website conversions, and referrals. We also leverage partnerships (e.g., VNurt) to access companies already hiring at scale, and use pilot assessments to demonstrate value and drive conversions. |
We follow a focused, outbound-led go-to-market strategy targeting hiring managers and TA heads in startups and mid-sized companies. We start with highly targeted cold outreach (email + LinkedIn), positioning simulation-based assessments as a way to improve screening quality and reduce hiring time. Instead of pushing a full product upfront, we offer pilot assessments for active roles to demonstrate immediate value. Alongside outbound, we build inbound through LinkedIn content, case studies, and website funnels that highlight measurable outcomes (e.g., better signal, faster shortlisting). Partnerships play a key role—through VNurt, we tap into companies already hiring at scale and integrate directly into their workflows. Once value is proven, we expand via a land-and-expand model, growing usage across roles and teams within the same company. |
To become the default infrastructure for hiring—where every candidate is evaluated based on real-world performance, not resumes.
We aim to build a system where companies can simulate any role, assess candidates at scale with high accuracy, and continuously improve hiring decisions using data. Over time, this evolves into a comprehensive talent intelligence layer that not only helps companies hire better but also identifies skill gaps, predicts performance, and guides workforce planning. |
Private Limited Company |
NA |
one |
We’re applying to Avinya because it aligns closely with what we need at this stage—focused mentorship, access to strong datasets, and deep ecosystem support to scale an AI-first product. QualifyMe sits at the intersection of AI and real-world decision-making, where access to high-quality data, industry insights, and the right guidance can significantly accelerate product maturity and accuracy. Avinya’s structured program and access to Karnataka’s AI ecosystem, including IIT alumni networks and industry partners, provide exactly that leverage. Additionally, the opportunity to work closely with experts and tap into real-world problem statements helps us sharpen our solution and expand into adjacent high-impact sectors over time. |
1. Strengthen our AI evaluation layer, 2. Validate at scale with real companies, 3. Refine product and GTM for scale |
Yes |
one |
one |
Our AI architecture is built as a modular pipeline that mirrors the hiring workflow—from requirement understanding to final candidate evaluation.
1. Input structuring layer
Job descriptions are broken down into granular, evaluatable signals (skills, responsibilities, expectations). This creates a structured blueprint for assessment.
2. Simulation & question generation
Using LLMs, we generate role-specific tasks, scenarios, and questions aligned to each signal—ensuring assessments reflect real-world job conditions.
3. Candidate interaction layer
Candidates go through multi-step assessments (MCQs, subjective responses, coding, and conversational interviews via an AI interface).
4. Evaluation engine
Responses are analyzed using LLM-based evaluators with predefined rubrics. Each response is scored, reasoned, and mapped back to the original job signals (matched / partial / not matched).
5. Aggregation & insights layer
All signals are combined into a structured report with skill-level insights, role-fit scoring, and explainable reasoning for each decision.
6. Feedback loop (learning layer)
Assessment data and outcomes are continuously used to refine prompts, evaluation criteria, and scoring models—improving accuracy over time.
The system is designed to be explainable, configurable, and extensible—allowing us to adapt to different roles, industries, and hiring workflows. |
Our primary data advantage comes from structured, ground-up evaluation data tied to real job requirements.
Unlike traditional platforms that rely on generic test scores or resumes, we capture:
Granular signal-level data (each skill/requirement mapped to candidate performance)
Simulation-based responses (how candidates perform in real-world scenarios)
Evaluation reasoning (why a candidate was marked fit or not)
This creates a high-quality, labeled dataset linking job requirements → candidate responses → evaluation outcomes.
As we scale, this evolves into a proprietary dataset that improves:
Accuracy of role-fit predictions
Quality of generated assessments
Benchmarking across roles and industries
Over time, this data forms a strong moat, as it is difficult to replicate without both the simulation layer and continuous hiring interactions. |
Our defensibility comes from how our AI is structured, trained, and continuously improved—not just the use of LLMs.
1. Structured evaluation framework (not generic AI usage)
We decompose roles into granular, evaluatable signals and map every question, response, and score back to these signals. This creates consistency and reduces randomness—something most competitors lack.
2. Simulation-based data layer
Our AI is trained on real-world task performance, not just resumes or MCQs. This produces a richer, harder-to-replicate dataset tied to actual job scenarios.
3. Proprietary evaluation + reasoning data
We don’t just store scores—we capture why a candidate was evaluated a certain way. This builds a unique dataset of responses, rubrics, and reasoning that improves model reliability over time.
4. Closed feedback loop
Every assessment feeds back into improving prompts, rubrics, and scoring logic. This compounding loop increases accuracy and creates a data moat as usage grows.
5. System-level integration
Our advantage is not a single model, but the end-to-end pipeline—from job understanding → simulation generation → evaluation → insights. Replicating this requires both strong AI and deep hiring workflow integration.
Together, this makes our AI more consistent, explainable, and continuously improving—creating a defensible edge beyond just using off-the-shelf models. |
We evaluate model performance and reliability through a combination of quantitative metrics, human benchmarking, and real-world outcomes:
1. Human-aligned evaluation
We benchmark AI evaluations against expert/recruiter judgments to measure agreement, consistency, and drift over time.
2. Signal-level accuracy
Since every assessment is broken into granular signals, we track precision/recall at the skill level—not just overall scores—making errors more visible and fixable.
3. Consistency & stability checks
We run repeated evaluations on the same inputs and slight variations to ensure scoring and reasoning remain stable.
4. Rubric and prompt validation
Each role-specific rubric is tested across multiple candidate responses to ensure it produces fair, unbiased, and role-relevant evaluations.
5. Outcome-based validation
Where possible, we correlate assessment results with downstream outcomes (shortlisting decisions, interview performance, hiring success) to validate real-world effectiveness.
6. Continuous monitoring loop
We track edge cases, failure modes, and recruiter feedback to iteratively improve prompts, scoring logic, and evaluation criteria.
This multi-layered approach ensures the system is not just accurate in isolation, but reliable and trustworthy in real hiring scenarios. |
We take a privacy-first, enterprise-grade approach across data handling, compliance, and system security:
1. Data minimization & control
We only collect data necessary for assessments and give companies full control over candidate data, including access, retention, and deletion.
2. Secure data handling
All data is encrypted in transit (TLS) and at rest. Access is strictly controlled through role-based permissions and audit logs.
3. Isolation & infrastructure security
Customer data is logically isolated, with secure cloud infrastructure, regular backups, and monitoring to prevent unauthorized access.
4. AI data protection
Candidate data is not used to train external models. We use controlled pipelines and, where required, privacy-safe configurations with model providers.
5. Compliance readiness
We align with standard data protection practices (such as GDPR principles) and are building toward formal compliance as we scale.
6. Responsible AI practices
We design evaluations to be explainable, reduce bias through structured rubrics, and continuously monitor outputs for fairness and reliability.
This ensures that both companies and candidates can trust the platform with sensitive hiring data. |
Yes—primarily around AI in hiring, data privacy, and bias/fairness.
1. AI regulation in hiring
Emerging laws (e.g., EU AI Act–like frameworks) may classify hiring AI as high-risk, requiring transparency, auditability, and human oversight.
2. Data privacy regulations
Laws like GDPR and India’s DPDP Act impact how candidate data is collected, processed, stored, and deleted.
3. Bias and discrimination risks
Automated evaluations must be demonstrably fair and non-discriminatory, with explainable outputs and audit trails.
4. Cross-border data handling
If serving global customers, data residency and transfer restrictions can apply.
Mitigation:
We’ve built the system with explainability, human-in-the-loop options, structured evaluations, and strict data controls, making it easier to adapt to evolving regulations and compliance requirements. |
At 10× scale, the core risks are less about model capability and more about consistency, cost, and control:
1. Evaluation consistency
LLM-based scoring can drift across large volumes and varied roles. Without tighter standardization, outputs may become less reliable.
2. Prompt and rubric scalability
Manually crafted prompts and rubrics won’t scale linearly across many roles and industries—we need more automated, self-improving systems.
3. Latency and cost
High-volume assessments (especially simulations and evaluations) can significantly increase inference costs and response times.
4. Edge cases and failure modes
With more diverse candidates and roles, rare edge cases and unexpected responses will increase, exposing gaps in evaluation logic.
5. Data quality management
As data grows, maintaining high-quality, well-labeled datasets (especially for feedback loops) becomes harder.
How we’re addressing this:
Standardizing evaluation frameworks and scoring schemas
Building automated rubric generation and validation
Optimizing model usage (caching, batching, smaller models where possible)
Adding guardrails and fallback mechanisms
Strengthening data pipelines and continuous monitoring
This ensures we can scale while maintaining reliability and unit economics. |
Yes—we have an in-house team focused on applied AI and product engineering.
The team’s strength is in building production-grade LLM systems, including prompt engineering, evaluation frameworks, and workflow orchestration (using tools like n8n). We have experience designing end-to-end AI pipelines—from structuring inputs (job requirements) to generating simulations, evaluating responses, and producing explainable outputs.
On the engineering side, we bring strong expertise in scalable backend systems, API design, and cloud infrastructure, ensuring the AI components are reliable and performant in real-world usage.
Importantly, our approach is highly applied and product-driven—we focus on integrating AI deeply into hiring workflows, continuously iterating based on real user data and feedback rather than building isolated |
We primarily use a combination of proprietary and controlled third-party data:
1. Proprietary data (owned)
Candidate assessment responses from simulations, interviews, and tests
Structured evaluation data (scores, reasoning, signal-level mappings)
Job descriptions and role-specific evaluation frameworks created on the platform
2. Synthetic and generated data
LLM-generated scenarios, questions, and sample responses used to bootstrap and refine evaluation frameworks
3. Public/third-party data (non-sensitive)
General role descriptions and skill benchmarks from publicly available sources, used for baseline structuring (no sensitive or personal data)
We do not use candidate data to train external models, and all proprietary data is controlled within our system with clear ownership and access boundaries. |
We improve model performance through a closed feedback loop built on real usage data:
1. Continuous data collection
Every assessment generates structured data—responses, scores, and reasoning—mapped to specific job signals.
2. Human-in-the-loop validation
We benchmark and periodically review AI evaluations against recruiter decisions to detect gaps and refine scoring logic.
3. Prompt and rubric optimization
We iteratively improve prompts, evaluation rubrics, and scoring criteria based on observed errors and edge cases.
4. Outcome-based learning
We track downstream outcomes (shortlisting, interview success, hiring decisions) to align evaluations with real-world performance.
5. Model and pipeline optimization
We experiment with model combinations, fine-tuning (where needed), and cost-performance tradeoffs to improve accuracy and efficiency.
6. Monitoring and guardrails
We continuously track consistency, drift, and failure modes, with fallback mechanisms to maintain reliability.
This creates a compounding system where more usage directly improves accuracy, consistency, and overall decision quality. |
Both—starting with India as our primary market, and expanding to global markets as we scale.
India provides a strong initial base with high-volume hiring and clear inefficiencies in screening. At the same time, our product is designed to be role-agnostic and globally applicable, allowing us to expand into other markets with minimal changes once we achieve strong validation and repeatability. |
|
|
|
|
|
|
| Mar 25, 2026 @ 6:27 PM |
Vipin Ahuja | Rajkuamar Ahuja |
info@vaidikutpaad.com |
https://www.linkedin.com/in/vipin-ahuja-37323716/ |
https://www.vaidikutpaad.com |
9891270785 |
Founder & CEO| B.E. Electronics DAVCET MDU Haryana| 12+ Years of Experience in Automation & Special Purpose Machine Development |
Other Founder is My Father acting as a sleeping partner |
1 |
Business Management alignment with Vision |
Samvidam Pvt. Ltd. |
https://www.vaidikutpaad.com |
Gurugram |
Reviving Indian Heritage with Scalable Innovation |
Problems in Current Appliances in Market: Disregard to Indian Traditional Principles, Biproducts with Less Nutrient Value, High Speed, Less Features |
Vaidik Mathani(Auto Bi Directional)
Vaidik Hawankund(Smoke Controlled) |
Vaidik Mathani: Automatic Bi-Direction Rotation,Variable Speed, Different Wooden Churner/Whisk, Automatic Stop, DC Motor(Upto 70% Low Power Consumption), Adjustable Height, Home/Commercial Purpose Variants , More Nutrients in Butter/Buttermilk /Ghee | Vaidik Hawankund: Smoke Controlled with Changable Fan Speed,Electric Lighter for ignition, Rechagable Battery |
Revenue |
Revenue |
Mass Market :Health Conscious People Traditional Sanatan Culuture Believer. e.g: Patanjali, Isckon,Shantikunj Follower| Niche Market: Dairy Entrpreneurs(SHG,Small Farms)/Small Mills Vaidik Bilona Ghee Manufacturers Temple Kitchens(Mass Prasad Preparation) Ayurvedic Brands |
660000 Cr. |
210000 Cr. |
2100 Cr. |
Revenue Streams: Direct Product Sales(Online/Offline), Spare Parts/VariousWhisks/Samidha Sale, Institutional Partnerships(Patanjali/Isckon/Shanti Kunj/Temples etc.), Export Partnerships |
Some MSME Players |
Revenue Channels: Amazon/Flipkart/Indiamart,Own Website via SEO, Social Media(Facebook/Instagram/You Tube/Whatsapp), Trade Expo |
Short Term: D2C From Marketplaces, Own Website, Social Media Marketing |
Long Term: B2B through Dealership & Distributorship, VaidikUtpaad Outlets with Franchisee Model
In Long Term We want to be in Heritage Inspired Appliances like Cold Pressed Oil Maker/Flour Maker Appliance etc. |
Private Limitred |
9.5 Lacs Grant |
Yes |
For Funding & GTM Mentorship |
GTM Strategy/Networks & Funding |
Yes |
Yes |
Automation layer |
Microcontroller Based |
|
Development Takes Time |
By Testing in various test parametrs |
Signing NDA with Employees/Vendors |
BIS Certification |
|
Yes/ Sensor/ Automation/Programming |
|
By testing/ Customer Feedback based improvements |
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/UJf4quy26pIT-Consumer-Products_I1_Samvidam-Pvt.-Ltd.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/UJf4quy26pIT-Consumer-Products_I1_Samvidam-Pvt.-Ltd.pdf |
https://youtu.be/44Bfz9ZVMJE |
HEALTH IMPACT
Promotes Natural and Holistic Lifestyles.
Potential for Preventive Health Benefits through Ayurvedic Principles.
ECONOMIC & SOCIAL IMPACT
Empowering Rural Artisans and Local Supply Chains.
Strengthening India’s Positioning in Global Heritage Exports.
MARKET REACH
Growing Demand for Authentic Indian-Wellness & Heritage Products.
Expanding Global Interest in Ayurved Inspired and Eco-Conscious Products.
CONSUMER IMPACT
Affordable access to Trusted, Certified Traditional Products.
Builds Cultural Pride and Consumer Trust through Authenticity and Transparency. |
NA |
NA |
checked |
| Mar 25, 2026 @ 3:45 PM |
Harsh Chauhan |
Legendharsh904@gmail.com |
https://www.linkedin.com/in/harsh-chauhan-a3531730a/ |
http://NA |
9012981324 |
Harsh Chauhan – Founder & CEO: Background in Electronics & Communication Engineering with experience in AI, embedded systems, and assistive technology development. Leads product vision, AI–hardware integration, and overall strategy. Lalita Chauhan – Co-Founder: Background in operations and coordination, focusing on partnerships, execution, and community engagement. Core Team: Includes members with expertise in AI/ML, hardware development, UI/UX, and product design, contributing to development and scaling of Signify. |
The team was formed through academic collaborations, shared interest in innovation, and startup activities. We have been working together for approximately 1 year, collaborating on projects, prototype development, and startup execution. |
All |
Our biggest strength is our ability to combine strong technical expertise with real-world problem understanding. We work closely with users, allowing us to build practical, user-centric solutions rather than just theoretical products. Our team is highly adaptable, collaborative, and execution-focused, enabling us to rapidly prototype, iterate, and deliver impactful innovations like Signify. |
Impech Solutions |
NA |
Meerut |
Signify ( Voice of Voiceless people ) |
Signify is an AI-powered assistive device that uses computer vision and machine learning to recognize sign language gestures and convert them into real-time voice and text, enabling independent, fast, and accessible communication in everyday environments. |
Signify is an AI-powered assistive device that uses computer vision and machine learning to recognize sign language gestures and convert them into real-time voice and text, enabling independent, fast, and accessible communication in everyday environments. |
Signify is unique because it combines AI-based computer vision, edge computing, and embedded hardware into a single portable device for real-time sign language translation. It works offline, ensuring low latency, privacy, and reliability. Our proprietary sign language dataset and optimized AI models, along with hardware–software integration, create strong technical barriers to entry and scalability. |
Idea |
Users, Revenue |
Primary: People with speech and hearing impairments. Secondary: NGOs, special schools, hospitals, corporates, and government institutions working on accessibility and inclusion. |
300 |
100 |
63 |
Hybrid B2C, B2B, and B2G model through device sales, institutional partnerships, government procurement, training services, and future AI subscription upgrades. |
No direct competitors. |
Through NGO partnerships, institutional collaborations, pilot programs, government channels (GeM), and awareness campaigns in disability-focused networks. |
Start with NGOs and institutions for bulk deployments, expand through government programs, and scale globally with multilingual support and partnerships. |
Start with NGOs and institutions for bulk deployments, expand through government programs, and scale globally with multilingual support and partnerships. |
Incorporated as Impacech Solutions Private Limited (Registered in India) |
4 lakh |
Yes |
We are applying to Avinya to gain structured mentorship, refine our product-market fit, and accelerate the development of Signify. The program’s focus on early-stage innovation and impact-driven startups aligns with our mission to build scalable assistive technology. We also seek access to expert guidance, networks, and growth opportunities. |
We are applying to Avinya to gain structured mentorship, refine our product-market fit, and accelerate the development of Signify. The program’s focus on early-stage innovation and impact-driven startups aligns with our mission to build scalable assistive technology. We also seek access to expert guidance, networks, and growth opportunities. |
Yes |
Yes |
Core engine |
Signify uses a multimodal AI architecture combining computer vision and deep learning. A camera captures gesture inputs, which are processed using CNN/Transformer-based models for sign recognition. The output is converted into text and speech using NLP and TTS modules, optimized for edge deployment. |
We are building a custom sign language dataset with real user participation, capturing diverse gestures, environments, and regional variations, improving real-world accuracy. |
Our proprietary dataset, optimized edge-AI models, and hardware–software integration create strong defensibility. Continuous learning from real-world usage further strengthens our competitive advantage. |
Gesture recognition accuracy
Latency (response time)
Precision & recall
Real-world usability feedback
Error rate in live environments |
Signify processes data offline on-device, ensuring user privacy. No sensitive data is stored or transmitted. Future systems will follow data protection and accessibility compliance standards. |
Minimal. Only standard electronic device certifications and accessibility compliance are required for large-scale deployment. |
Challenges may include dataset diversity, model generalization, and hardware performance under scale. We plan to address this through continuous data collection, model optimization, and scalable architecture. |
Yes, we have an in-house AI/ML team with expertise in computer vision, deep learning, and edge AI deployment. The team works on gesture recognition models, dataset creation, model optimization, and integration with embedded hardware for real-time performance. |
We use a combination of public datasets ( WLASL, RWTH-PHOENIX) and proprietary datasets created in-house with community participation. Our owned dataset is continuously growing and tailored for real-world usage. |
We improve performance through continuous data collection, model retraining, real-world feedback, and iterative testing. Techniques like data augmentation, fine-tuning, and optimization for edge devices ensure better accuracy and reliability over time. |
We are building for both India and global markets, starting with Indian Sign Language (ISL) and expanding to international standards like ASL for broader scalability. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/d9YxQOiedRD0-impech_signify.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/d9YxQOiedRD0-impech_signify.pdf |
https://drive.google.com/file/d/13okDvxuTpjUEiuzXq2iA4-mtEzqmcCuJ/view?usp=sharing |
Yes, Signify is a mission-driven startup focused on solving communication barriers for people with speech and hearing impairments. Our goal is to enable inclusive communication using AI-powered technology, improving access to education, employment, healthcare, and public services while promoting dignity, independence, and equal opportunities. |
NA |
Self |
checked |
| Mar 25, 2026 @ 3:29 PM |
Aniket Patil |
medtech360solutions@gmail.com |
https://www.linkedin.com/in/aniket-patil-150632238/ |
https://www.medifor7.com/ |
7219785868 |
CEO |
The team came together through academic and innovation ecosystem interactions, particularly during engineering studies and participation in startup and incubation programs. We have collaborated over the past few years while working on technology-driven projects, research initiatives, and the development of MediFor7. This journey includes working together on prototype development, problem validation, and engaging with mentors, incubators, and healthcare stakeholders. |
1 |
Our biggest strength as a team is the combination of strong technical expertise and a shared commitment to solving a real-world healthcare problem. We bring together skills in hardware development, embedded systems, research, and innovation, along with hands-on experience in the startup ecosystem. Additionally, our ability to stay persistent, adapt quickly, and execute under constraints has helped us progress from idea to prototype while continuously improving the solution based on real-world feedback. |
MediFor7 |
https://www.medifor7.com/ |
Pune |
MediFor7 is an IoT-enabled smart medication management platform that ensures patients take the right medicine at the right time through real-time monitoring and adherence tracking. |
MediFor7 addresses medication non-adherence, a major healthcare issue where patients—especially elderly and chronic cases—miss or incorrectly take medicines, leading to poor health outcomes, increased hospitalizations, and higher healthcare costs, while caregivers lack real-time monitoring visibility. |
MediFor7 offers a smart medication device integrated with IoT and a cloud-based platform that provides automated reminders, tracks medication intake using sensors, and enables real-time monitoring through a digital dashboard for caregivers and healthcare providers. |
MediFor7 stands out through its integrated hardware-software approach, real-time adherence analytics, and focus on affordability for Indian and emerging markets, supported by patented designs and healthcare-centric functionality that creates strong technological and operational defensibility. |
MVP |
Pilots |
The primary customers of MediFor7 are hospitals, clinics, and home healthcare providers that manage patients requiring long-term medication adherence support. The end users include elderly individuals, chronic disease patients, post-discharge patients, and caregivers who need reliable systems to ensure timely and accurate medication intake. |
The Total Addressable Market for MediFor7 includes all individuals requiring daily medication adherence support in India, estimated at around ₹1.6 lakh crore, covering elderly populations, chronic disease patients, and healthcare institutions across urban and rural regions. |
The Serviceable Available Market focuses on urban and semi-urban patients with access to smartphones and healthcare services, estimated at approximately ₹40,000 crore, including hospitals, clinics, and digitally connected patient segments. |
The Serviceable Obtainable Market is initially targeted at around ₹120 crore, aiming to reach approximately 1 lakh users within the first 3 years through a combination of B2B and B2C distribution channels. |
MediFor7 follows a hybrid revenue model including one-time device sales, annual maintenance contracts (AMC), subscription-based services for analytics and alerts, and SaaS-based offerings for clinics and hospitals for patient monitoring and workflow integration. |
The main competitors include traditional pill organizers, mobile reminder applications, and smart medication devices such as MedMinder, Hero Health, and other global IoT-based adherence solutions, though many lack integrated monitoring and affordability for emerging markets. |
Customer acquisition is driven through hospital and clinic partnerships, pilot deployments, referrals, digital marketing, caregiver awareness campaigns, and collaborations with pharmacies, NGOs, and healthcare distributors. |
The go-to-market strategy focuses on B2B partnerships with hospitals and clinics for bulk deployments, supported by B2C channels through pharmacies and online platforms. It also includes pilot programs, ecosystem partnerships, and targeted outreach in major cities to build early traction. |
The long-term vision is to build a comprehensive, India-first digital medication adherence ecosystem that integrates homes, clinics, and hospitals, enabling large-scale remote patient monitoring and improving healthcare outcomes for millions of patients globally. |
Pvt Ltd |
10 Lakh |
Yes |
I am applying to the Avinya Cohort under ACSEL to accelerate the development of MediFor7 by leveraging structured incubation, expert mentorship, and a strong innovation ecosystem. The program’s focus on impactful, scalable solutions aligns closely with our goal of addressing medication non-adherence in healthcare. I believe Avinya can provide the right guidance in refining our product, strengthening our business model, and connecting with relevant stakeholders, including healthcare institutions and investors, which is critical at our current stage. |
During the programme, I aim to refine the MediFor7 prototype into a robust, market-ready product through technical mentorship and validation support. I also want to establish strong pilot partnerships with hospitals and healthcare providers, improve our go-to-market strategy, and prepare for early-stage funding. Additionally, I seek to strengthen |
Yes |
Yes |
Supporting feature |
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/HoRWmbVOUWsc-MediFor7_PitchDeck_SSP_ACIC.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/HoRWmbVOUWsc-MediFor7_PitchDeck_SSP_ACIC.pdf |
https://drive.google.com/file/d/1kBsXkpZL9rBCUE8Dk5UPv4SFJQxJnqqq/view?usp=drive_link |
Yes, MediFor7 is a mission-driven and impact-focused startup aimed at solving the critical healthcare challenge of medication non-adherence, which significantly affects elderly and chronic disease patients. The startup is focused on improving patient outcomes, reducing avoidable hospitalizations, and easing the burden on caregivers through an IoT-enabled smart medication management system. By ensuring timely and accurate medication intake, MediFor7 directly contributes to better health, increased patient safety, and more efficient healthcare delivery. The solution is designed to be affordable and scalable, particularly for India and emerging markets, where access to continuous healthcare monitoring is limited. Our mission is to build a connected healthcare ecosystem that empowers patients, supports caregivers, and enables data-driven decision |
NA |
NA |
checked |
| Mar 25, 2026 @ 3:28 PM |
Sadakathulla Shah , Riyaz Mohamed |
shah@tinkedge.com |
https://www.linkedin.com/in/sadakathulla-shah-42646b61/ |
https://tinkedge.com/ |
9606932923 |
xxx |
xxx |
two |
xxx |
Tescom Technologies Pvt. Ltd. |
www.tinkedge.com |
Bengaluru |
x |
x |
xxx |
xx |
Revenue |
two |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 25, 2026 @ 2:39 PM |
Harsh Chauhan |
Legendharsh904@gmail.com |
https://www.linkedin.com/in/harsh-chauhan-a3531730a/ |
http://NA |
9012981324 |
Harsh Chauhan – Founder & CEO: Background in Electronics & Communication Engineering with experience in AI, embedded systems, and assistive technology development. Leads product vision, AI–hardware integration, and overall strategy. Lalita Chauhan – Co-Founder: Background in operations and coordination, focusing on partnerships, execution, and community engagement. Core Team: Includes members with expertise in AI/ML, hardware development, UI/UX, and product design, contributing to development and scaling of Signify. |
The team was formed through academic collaborations, shared interest in innovation, and startup activities. We have been working together for approximately 1 year, collaborating on projects, prototype development, and startup execution. |
All |
Our biggest strength is our ability to combine strong technical expertise with real-world problem understanding. We work closely with users, allowing us to build practical, user-centric solutions rather than just theoretical products. Our team is highly adaptable, collaborative, and execution-focused, enabling us to rapidly prototype, iterate, and deliver impactful innovations like Signify. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 25, 2026 @ 2:31 PM |
Swarasya V L, Sujay P, Prithiv Palani |
swarasya.vl@schijnenn.com |
https://www.linkedin.com/in/swarasya-v-l-b52730207/ |
https://schijnenn.com/ |
9480258896 |
|
|
All |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 25, 2026 @ 12:34 PM |
Swapna Dasoju |
swapna.dasoju@aventiqinnovations.com |
https://www.linkedin.com/company/aventiq-innovations-pvt-ltd/ |
https://aventiqinnovations.com/ |
+91 81785 37398 |
Mrs. Swapna Dasoju Co-Founder, Director, and Principal Investigator Mrs. Swapna Dasoju is a Co-Founder and Director at Aventiq Innovations Pvt. Ltd., specializing in leading translational healthcare technology projects. With over a decade of cross-disciplinary expertise spanning microbiology,business, and healthcare operations, she bridges the gap between scientific innovation and market application. Her groundbreaking work in the industry includes scientific training at CSIR-IICT , Hyderabad and extensive industry exposure at NATCO, covering drug manufacturing, Quality Control (QC), Quality Assurance (QA), and regulatory processes. Her ground exposure to systemic gaps in medical waste handling directly led to the conceptualization of the company’s novel AI-assisted waste management platform. Recognized as the Best Women Entrepreneur (2022-23) under the PMBJP Scheme, she leverages her academic background - holding an M.Sc. in Microbiology and MBA to strengthen Aventiq’s R&D efforts, scientific validation, and product development in sustainability-focused technologies. |
,,,,,,,,,,,,,, |
one |
mmm |
Aventiq Innovations Pvt Ltd |
https://aventiqinnovations.com/ |
Bengaluru |
ABIOT-E-SAFE is a decentralized, AI-enabled biomedical waste treatment platform that disinfects and secures infectious waste at the point of generation |
Biomedical waste management today is highly centralized and delayed, with 24–72 hour gaps between waste generation and treatment. During this time, infectious waste can amplify microbial loads and contribute to antimicrobial resistance (AMR). Additionally, ~60% of healthcare facilities in India fail compliance audits due to manual errors, poor segregation, and lack of real-time monitoring. |
ABIOT-E-SAFE™ is a ward-level, standalone system that performs on-site disinfection of biomedical waste using a combination of UV-C and oxidizing micro-fog technology.
It integrates AI-based waste classification, sealed containment, and automated compliance logging, converting infectious waste into safe, transport-ready units while ensuring regulatory compliance. |
ABIOT-E-SAFE™ is not a replacement for existing disposal systems but a pre-treatment and compliance layer at the point of waste generation. Key differentiators include: At-source treatment (before storage/transport) AI-assisted waste validation Sequential multi-modal disinfection (UV-C + micro-fog) SmartTrace™ digital compliance system Patent-backed design and deep-tech integration This combination of hardware, AI, and compliance infrastructure creates strong defensibility. |
two |
two, Pilots |
Hospitals (private & government), diagnostic labs, clinics, and healthcare facilities facing regulatory pressure for biomedical waste compliance. |
₹4,500–6,500 Cr |
₹1,500–2,000 Cr |
₹500–800 Cr |
Hybrid model: One-time hardware sales Recurring SaaS (compliance dashboard – SmartTrace™) Consumables (disinfection cartridges, cassettes) AMC/service contracts |
Existing players include: Centralized biomedical waste treatment facilities (CBWTFs) Autoclave and incineration systems Compliance software platforms However, none operate at the point of generation, which is our product's key differentiation. |
Initial traction through: Hospital partnerships and pilot programs Infection control committees Regulatory-driven adoption (CPCB/NABH compliance) Strategic partnerships with hospital networks and incubators |
Start with pilot deployments in NABH-accredited hospitals and high-load wards. Use validation data and compliance benefits to expand within hospital networks, followed by scale through partnerships with healthcare chains and government programs. |
To become a global platform for decentralized infection control and biomedical waste management, enabling safer healthcare systems and reducing AMR risks at scale. |
Incorporated as Aventiq Innovations Pvt. Ltd. |
Bootstrapped; no external funding raised yet. Currently applying for grants and early-stage support. |
one |
Avniya provides a strong ecosystem for deep-tech and impact-driven healthcare innovations. The program’s focus on validation, mentorship, and ecosystem access aligns well with our current stage of prototype development and pilot readiness. |
Build and validate TRL-5 prototype , Pilot deployment in hospitals ,Strengthen regulatory and compliance readiness, Refine GTM and partnerships |
Yes, open to hybrid/virtual participation. |
one |
two |
Edge-based computer vision model for waste classification, combined with rule-based decision systems aligned with biomedical waste management regulations. |
Supporting layer (not core, but critical for automation and compliance) |
Data generated from real-world hospital workflows, waste patterns, and compliance logs collected during pilots. |
Accuracy of classification, false positives/negatives, compliance validation rates, and operational reliability in real-world conditions. |
No patient data is collected.
All data is operational (waste/compliance), encrypted, and aligned with regulatory requirements. |
Biomedical waste regulations (CPCB), hospital compliance standards (NABH), and safety certifications. |
Potential challenges in:
1- Model generalization across hospitals
2- Hardware deployment consistency
3- Data synchronization at scale |
Yes – supported through internal team and ecosystem partners (IKP-EDEN, etc.) |
Internal operational datasets (pilot + testing), not dependent on external sensitive datasets. |
Continuous learning from deployment data, feedback loops, and edge-model updates. |
Starting with India, with long-term plans for global expansion. |
|
|
Yes. Focused on reducing AMR, improving healthcare safety, and enabling sustainable waste management aligned with public health and environmental goals. |
Women-led startup. |
N/A |
checked |
| Mar 25, 2026 @ 10:02 AM |
sagar mainkar |
sagar@theprojectskin.com |
https://www.linkedin.com/in/sagar-dilip-mainkar-7160252b0/ |
https://theprojectskin.com/ |
9071144458 |
CEO |
We met through networking meets and some through y combinator |
2 |
Our core strength is **execution driven by deep user insight**.
We combine strong product thinking with rapid iteration—having already completed 250+ AI skin scans and validated our solution with 100+ active beta users. This allows us to continuously refine both our technology and user experience based on real-world data, not assumptions.
As a team, we move fast, stay close to the customer, and are highly outcome-focused—enabling us to translate insights into a scalable, personalized skincare platform. |
nureya beauty and wellness pvt ltd |
https://theprojectskin.com/ |
Bangalore |
An AI-powered platform delivering personalized, data-driven skincare tailored to each individual. |
Skincare today is generic and trial-and-error, lacking personalized, data-driven solutions that deliver consistent results. |
We are building an AI-powered, end-to-end skincare platform that starts with a simple mobile-based skin scan to assess key parameters like acne, pigmentation, texture, hydration, and sensitivity. This data is processed through our proprietary models to generate a highly personalized skincare routine, including product recommendations tailored to the user’s skin type, concerns, and environment.
Beyond a one-time recommendation, the platform continuously tracks skin progress through periodic scans, creating a feedback loop that allows the system to adapt routines dynamically based on what is actually working for the user. This shifts skincare from a static, trial-and-error process to a data-driven, outcome-oriented journey.
Additionally, we layer in user behavior, lifestyle inputs, and contextual factors (like climate and pollution) to further refine recommendations, while building a growing dataset that strengthens our models over time—enabling scalable, consistent, and measurable skincare results without requiring frequent dermatologist visits. |
An AI-powered platform that analyzes, tracks, and personalizes skincare routines dynamically based on real user data and outcomes. |
MVP |
Testimonials |
Digitally native consumers seeking personalized, effective skincare without the cost and friction of traditional dermatology. |
The global personalized skincare market represents a **$20B+ TAM**, driven by rising demand for AI-led, data-driven beauty solutions and increasing consumer spend on skincare. |
Our SAM is the $3–5B digital skincare market across urban, digitally active consumers in India and similar emerging markets. |
Our SOM is a$100–200M opportunity, targeting digitally native skincare users in Tier 1 Indian cities through a scalable, AI-first platform. |
Our revenue model is **product-led and recurring**—we monetize through personalized skincare product sales, subscription-based routines, and premium features like advanced skin analysis and progress tracking. |
Our main competitors include **CureSkin**, **SkinKraft**, and **Proven Skincare**, along with traditional dermatology clinics and generic skincare brands that lack personalization. |
We acquire customers through a **digital-first, performance-driven strategy**—leveraging social media marketing, influencer-led content, and targeted ads to drive users to try our free skin scan. This is complemented by **content-led education**, referral loops, and partnerships with dermatologists and beauty creators, converting users into paying customers through personalized results and visible outcomes. |
Digital-first GTM using influencer and performance marketing to drive users to a free skin scan, then converting and retaining them through personalized results and subscriptions. |
Our long-term vision is to build the **category-defining, AI-powered skin health platform**—the default interface through which consumers discover, manage, and improve their skin.
We are scaling from a point solution into a **full-stack, data-driven ecosystem** combining diagnostics, personalization, and commerce. By continuously capturing longitudinal skin data, we create a **compounding data moat** that drives better outcomes, higher retention, and increasing LTV.
Over time, we expand into **predictive and preventive skincare**, proprietary formulations, and deeper integrations with dermatology—unlocking multiple high-margin revenue streams.
At scale, this positions us to become a **global platform business in a large, fragmented market**, with strong network effects, recurring revenue, and defensible differentiation driven by data and personalization. |
incorporated |
0 |
Yes |
We’re applying to Avinya Ventures to accelerate our transition from a validated product to a scalable, category-defining business. Avinya’s strong focus on founder-first support, strategic mentorship, and access to high-quality networks aligns with our need to refine our GTM, strengthen unit economics, and scale distribution efficiently. We believe their ecosystem can help us compress learning cycles, avoid execution pitfalls, and unlock faster growth. Additionally, access to capital, operators, and potential partnerships through Avinya will be critical as we build a defensible, data-driven skincare platform at scale. |
During the program, we aim to **achieve product–market fit at scale and build a repeatable growth engine**. Specifically, we want to: * Scale from early traction to **strong user growth with optimized CAC–LTV economics** * Improve **conversion and retention** through sharper personalization and outcomes * Build a **predictable acquisition funnel** across paid, organic, and referral channels * Strengthen our **data and AI capabilities** to improve accuracy and defensibility * Lay the foundation for **recurring revenue at scale** through subscriptions and product sales The goal is to emerge with a **validated, scalable business model ready for rapid expansion and institutional capital**. |
yes |
Yes |
Core engine |
Our AI architecture is a multi-layered, feedback-driven system designed for continuous learning and personalization at scale.
1. Data Ingestion Layer
We capture user skin images via mobile scans along with metadata such as age, skin type, lifestyle inputs, and environmental factors like climate and pollution.
2. Computer Vision Layer
Deep learning models analyze images to detect and quantify concerns like acne, pigmentation, texture, pores, and sensitivity, converting visual data into structured parameters.
3. Personalization Engine
A hybrid of rules and machine learning maps skin data to personalized routines, combining dermatology logic with learned user behavior.
4. Feedback Loop and Continuous Learning
Users rescan over time, enabling outcome tracking and creating a closed loop where the system improves recommendations based on what actually works.
5. Data Layer and Moat
We build longitudinal skin profiles, creating proprietary data that strengthens accuracy, retention, and defensibility.
6. Scalable Infrastructure
Deployed on a cloud-native, modular architecture that supports real-time personalization and easy scaling across markets. |
.Our key advantage is a growing longitudinal skin dataset built from real user scans, outcomes, and behavior over time.
Unlike static datasets, we capture before–after progression linked to routines, products, and environmental factors, allowing us to understand not just skin conditions, but what actually works for each user. This creates a closed-loop feedback system that continuously improves model accuracy.
As we scale, this dataset compounds into a defensible data moat, enabling superior personalization, higher retention, and better outcomes that are difficult for competitors to replicate. |
Our AI is defensible because it is built on a closed-loop, outcome-driven system powered by proprietary longitudinal data.
We don’t just analyze skin—we continuously track what works over time, linking inputs like skin condition, routines, and environment to outcomes. This creates a compounding data advantage that improves model accuracy with scale.
Additionally, our tightly integrated stack—combining computer vision, personalization, and feedback loops—is embedded into the user journey, driving high retention and making it difficult to replicate without both data depth and distribution.
Over time, this leads to stronger personalization, higher switching costs, and a scalable data moat that competitors cannot easily match. |
We evaluate model performance and reliability across three layers:
1. Computer Vision Accuracy
Metrics: precision, recall, F1-score, AUC, and agreement with dermatologist-labeled ground truth.
2. Personalization Effectiveness
Metrics: improvement in skin scores over time, reduction in specific concerns, and routine adherence rates.
3. Business & Outcome Metrics
Metrics: conversion rate, retention, repeat purchase rate, and LTV/CAC to ensure real-world impact.
We also track model drift, performance across different skin types and conditions, and incorporate continuous user feedback to maintain reliability at scale. |
We follow a privacy-by-design approach across our system:
1. Data Minimization and Consent
We collect only essential data (skin images and relevant metadata) with explicit user consent and clear transparency on usage.
2. Encryption and Secure Storage
All data is encrypted in transit (TLS) and at rest (AES-256), with strict access controls and role-based permissions.
3. Anonymization
Personally identifiable information is separated from skin data, and training datasets are anonymized or pseudonymized wherever possible.
4. Compliance
We align with GDPR principles and India’s DPDP Act, ensuring user rights such as data access, correction, and deletion.
5. Infrastructure Security
Our cloud stack follows best practices with continuous monitoring, logging, and regular security audits.
6. User Control
Users can access, manage, or delete their data at any time, ensuring transparency and trust. |
Yes—there are a few key regulatory risks to consider:
1. Medical Classification Risk
If our product is interpreted as providing medical diagnosis or treatment, it could be classified as a medical device, triggering stricter approvals and compliance requirements.
2. Data Privacy Regulations
Evolving laws like India’s DPDP Act and global frameworks may impact how we collect, store, and process user data, especially sensitive image data.
3. Advertising & Claims
Restrictions on skincare and health-related claims could limit how we market outcomes, requiring careful positioning to avoid misleading or non-compliant claims.
4. Cross-border Expansion
Different countries have varying regulations around AI, health tech, and cosmetics, which could slow international scaling.
We mitigate these by positioning as a skincare and wellness platform (not a diagnostic tool), maintaining strong compliance practices, and building flexibility into our product and messaging. |
At 10x scale, the main risks are:
1. Model Generalization
Performance may drop across diverse skin tones, conditions, and geographies if the training data isn’t sufficiently representative.
2. Data Quality & Noise
Higher volume of user-generated images can introduce variability (lighting, angles, device quality), affecting prediction accuracy.
3. Model Drift
As user behavior and environmental conditions change, models may degrade without continuous retraining and monitoring.
4. Infrastructure Bottlenecks
Real-time image processing and personalization at scale can strain compute, latency, and cost efficiency.
5. Personalization Complexity
As the dataset grows, maintaining fast, accurate, and explainable recommendations becomes more complex.
6. Feedback Loop Integrity
Noisy or inconsistent user feedback can weaken the learning loop and impact recommendation quality.
We address this through continuous retraining pipelines, strong data validation, scalable cloud infrastructure, and robust monitoring systems. |
Yes, we have an in-house AI/ML team with expertise across computer vision, applied machine learning, and data engineering.
The team has experience building and deploying CNN-based image analysis models, working with noisy, real-world datasets, and optimizing models for mobile-first environments. On the ML side, we focus on personalization systems, recommendation engines, and feedback-loop learning, combining rule-based logic with data-driven approaches.
We also have strong capabilities in data pipeline design, model deployment, and cloud infrastructure, enabling continuous training, monitoring, and scaling of models in production.
Overall, the team is execution-focused with hands-on experience in taking AI systems from prototype to real-world deployment and iteration. |
We use a combination of **proprietary and publicly available datasets**.
Our core dataset is **owned and proprietary**, built from user skin scans, metadata, and longitudinal outcome data collected through our platform with user consent. This is our primary source of model training and continuous improvement.
In the early stages, we supplemented training with **publicly available and licensed dermatology image datasets** to bootstrap model performance and improve baseline accuracy.
As we scale, the system increasingly relies on our **first-party data**, which is continuously expanding and becoming our key competitive and defensible asset. |
|
We improve model performance through a **closed-loop, data-driven pipeline**:
1. Continuous Data Collection
Ongoing user scans, metadata, and outcomes expand our longitudinal dataset.
2. Feedback Loop Learning
We link recommendations to real outcomes (improvement, adherence), allowing models to learn what actually works.
3. Regular Retraining
Models are periodically retrained on fresh data to reduce drift and improve generalization across skin types and conditions.
4. Data Quality & Annotation
We refine labeling pipelines using automated checks and expert validation to maintain high-quality training data.
5. A/B Testing
We run controlled experiments on recommendations and models to optimize for both accuracy and user outcomes.
6. Monitoring & Drift Detection
We track performance metrics in production and trigger updates when accuracy or consistency drops.
This creates a compounding system where more usage directly improves model accuracy and personalization over time. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/iL66UxKG2DNR-Project-Skin-Pitch-Deck-120326-India-2-compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/iL66UxKG2DNR-Project-Skin-Pitch-Deck-120326-India-2-compressed.pdf |
http://NA |
Yes, we are building a mission-driven company focused on making effective, personalized skincare accessible to everyone.
Today, skincare is often expensive, generic, or confusing, leaving most people underserved. Our goal is to democratize skin health by replacing trial-and-error with data-driven, outcome-based care that is affordable and scalable.
We aim to improve user confidence, reduce long-term skin issues, and enable informed decisions without reliance on costly consultations.
At scale, we will build an inclusive, global skin dataset across diverse skin types and conditions, ensuring better representation and outcomes. |
NA |
LINKEDIN LINK |
checked |
| Mar 25, 2026 @ 9:42 AM |
SIDDHARTH BORDOLOI |
admin@qinocaisolutions.com |
https://www.linkedin.com/in/dhrubabordoloi/ |
https://qinocai.com/ |
9945658761 |
Product Owner - Graduate in Computer Science with honours in AI. Having 4 years of Software Development experiences |
We are family members and working together since last 5 years. |
2 |
- Business domain knowledge
- Technical skillsets
- Development experience |
QINOC AI SOLUTIONS LLP |
https://qinocai.com/ |
Bangalore |
Developing AI based workforce management solution in HR Tech domain |
Hiring & management of temporary Blue-collar workers |
The Indian temporary workforce market remains operationally fragmented. Employers often rely on local contractors, spreadsheets, phone calls, and manual reconciliation. Workers lack digital identity, structured work history, and easy access to verified opportunities. The result is poor visibility, inconsistent quality, attendance disputes, payroll errors, delayed wage processing, and compliance risk.
Key pain points
• Fragmented hiring channels and slow time-to-fill for urgent shift-based roles.
• Low transparency between employers, workers, and local intermediaries.
• Manual attendance, timesheet, and payable-time reconciliation.
• Compliance burden for contracts, KYC, and tax/statutory documentation.
• Limited digital access for regional-language and low-digital-literacy workers.
• Lack of workforce analytics for planning, forecasting, and cost control |
Our platform KAARYA competes against four broad alternatives: informal brokers and contractors, traditional staffing operations that remain operationally manual, horizontal job boards that stop short of workforce operations, and generic HR/payroll systems that are not designed for blue-collar temporary staffing workflows |
MVP |
Pilots |
Our platform targets high-frequency, shift-based, temporary and contract workforce categories across manufacturing, warehousing, logistics, retail, hospitality/events, construction support, field operations, and seasonal campaign work. |
₹1.296 lakh crore |
₹15,552 crore |
: ₹32.4 Cr Year 3 transaction revenue / ₹135 Cr Year 5 transaction revenue, with ₹35.64 Cr Year 3 total revenue / ₹148.5 Cr Year 5 total revenue |
Subscription Fee & 10% Commission on each Assignment |
Staffing Agencies, Recruitment Companies |
By collaborating with Staffing Agencies |
KAAYRA’s GTM strategy is to start with high-volume sectors such as manufacturing, warehousing, logistics, retail, and events, where temporary workforce demand is frequent and operational pain points are highest. It will acquire anchor employers through direct enterprise sales and channel partnerships, while using local job boards and multilingual voice-based onboarding to rapidly build worker supply and scale city by city. |
KAARYA’s long-term vision is to become the leading digital infrastructure platform for temporary and contract workforce management in India. It aims to connect employers and blue-collar workers through an AI-powered ecosystem that makes hiring, onboarding, attendance, compliance, and billing seamless and transparent. Over time, KAARYA seeks to build a large trusted worker network with multilingual, voice-based access so that even informal workers can participate in the formal economy. The platform also envisions expanding beyond India into Southeast Asia and other global markets with similar workforce challenges. In the long run, KAARYA wants to evolve from a staffing operations platform into a full workforce intelligence and productivity platform for businesses. |
Incorporated in 22 July 2025 |
NA |
Yes |
To raise fund, complete MVP and do pilot. |
Raise fund and complete the Build of MVP |
Hybrid. Virtual both |
Yes |
Automation layer |
KAAYRA’s AI architecture combines multilingual voice onboarding, worker-profile intelligence, AI-based candidate matching, and analytics-driven workforce operations on a single digital platform. The architecture is designed to reduce onboarding friction for blue-collar workers, improve job-fit accuracy for employers, and automate downstream processes such as contracts, attendance, payable-time calculation, invoicing, and enterprise reporting. |
KAAYRA’s proprietary data advantage comes from capturing multilingual worker onboarding data plus downstream operational data across attendance, payables, invoicing, and workforce analytics. This gives us a richer India-specific workforce dataset than platforms that only manage job discovery or staffing transactions. |
Usage of Speech to Text Generative AI with Indian Languages. KAAYRA’s AI is defensible because it is trained around proprietary multilingual worker data and embedded into the full workforce operating cycle, creating strong data loops, workflow lock-in, and India-specific execution advantages that are difficult to replicate quickly |
KAARYA evaluates AI performance using a mix of model accuracy metrics (Precision@K, Recall@K, F1, WER), business outcome metrics (fill rate, join rate, time-to-fill, no-show reduction), and production reliability metrics (uptime, latency, drift, failure rate) to ensure the platform is both accurate and operationally dependable. |
KAARYA handles data privacy, compliance, and security through a privacy-by-design architecture aligned to India’s DPDP Act and Rules, supported by ISO 27001-style controls, role-based access, encryption, auditability, consent management, incident response, and AI governance for fair and secure decision-making. |
Yes, KAARYA has regulatory risk, mainly in labour compliance, worker classification, data privacy, cybersecurity reporting, identity verification, and any future payment or lending workflows. The risk is not unusual for this category, but it must be handled through a configurable compliance engine, state-wise rules management, strong consent and audit trails, and a clear operating model that defines whether KAARYA is a platform, intermediary, or employer-facing staffing system. |
If KAARYA scales 10x without redesign, the AI will become slower, less accurate, harder to trust, and more expensive to run |
Yes. Our in-house AI/ML team brings hands-on expertise across multilingual speech AI, NLP, recommendation systems, predictive analytics, and MLOps, enabling KAARYA to build domain-specific AI for temporary workforce hiring and operations in India. The team’s capability spans the full lifecycle from data engineering and model development to production deployment, governance, explainability, and continuous improvement. |
KAARYA uses four classes of datasets: (1) proprietary internally labeled datasets and synthetic datasets created by our AI team, (2) customer-permissioned platform data such as job, candidate, onboarding, and workforce workflow data, (3) licensed third-party AI/data services for speech, OCR, verification, and foundation models where needed, and (4) selected public/open datasets for benchmarking and model bootstrapping. We own the derived models, labels, evaluation sets, and workflow intelligence generated within the platform, while third-party and customer-originated data is governed by contract, consent, and licensing terms. |
We continuously improve KAARYA’s AI through a closed-loop MLOps process: capture live outcomes and corrections, monitor drift/quality/latency in production, retrain on fresh labeled data using both scheduled and threshold-based triggers, and deploy updated models through shadow testing and progressive rollout. |
Our current MVP is developed for India but our AI Powered WFM Platform "KAARYA" can go global by expanding through India-linked labor corridors, starting with UAE and Singapore, and by packaging the platform as a configurable B2B workforce operating system with local compliance, partner-led delivery, multilingual AI, and country-specific workflow packs. Europe should be a later-stage market once GDPR and AI Act readiness is fully in place. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/MRo0LySctk5U-KAARYA-Digital-Workforce-Platform-for-Temp-Jobs-25032026.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/MRo0LySctk5U-KAARYA-Digital-Workforce-Platform-for-Temp-Jobs-25032026.pdf |
https://www.youtube.com/watch?v=S3DWbda5dkY |
KAARYA is a mission-driven startup focused on bringing transparency, inclusion, and formalization to India’s temporary workforce economy. We use AI and digital workflows to help blue-collar and entry-level workers access jobs more easily, while helping employers hire faster, stay compliant, and manage temporary labour with greater trust and efficiency. |
NA |
LinkedIn |
checked |
| Mar 25, 2026 @ 12:45 AM |
Akash Gupta |
farmspherica.innovations@gmail.com |
https://www.linkedin.com/in/akash-gupta-96bb48261/ |
http://www.aavrt.com |
7459805391 |
CEO and Founder |
We met through professional/academic connections and have been working together for the past few months as a team. |
All |
Our biggest strength as a team is strong collaboration, clear communication, and a shared vision towards building sustainable and innovative farming solutions. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 25, 2026 @ 12:25 AM |
Akash Gupta |
framspherica.innovations@gmail.com |
https://www.linkedin.com/in/akash-gupta-96bb48261/ |
http://www.aavrt.com |
7459805391 |
CEO and Founder |
We met through professional and academic connections and have been working together for the past few months as a team. |
All |
Our biggest strength as a team is strong coordination, effective communication, and a shared commitment towards innovation and sustainable farming solutions. |
Farmspherica Innovations Pvt. Ltd. |
www.aavrt.com |
Pilani |
A startup that grows plants without soil using hydroponics. |
We propose a next generation hydroponic system utilizing nanotechnologies- specifically, Plasma Activated Water(PAW) & nanobubbles to increase the growth rate of plants, suppress pathogens and providing a boost in agricultural productivity by reduced dependency on external agri-inputs such as fertilizers and herbicides. |
We are currently working to solve the challenges of traditional agriculture and low agricultural productivity in semi-arid and arid regions of India, currently Rajasthan. This is a deep and complex problem which has several causes, geographical, economic and social. To tackle the technical aspects, we are developing a hydroponics system. Hydroponics is a viable technology which is suitable for arid agriculture, however it remains a semi-mature industry due to difficulty of breakthrough and high initial cost to establish farms. However, the core of our innovation lies in the usage of specific nanotechnologies to increase the growth rate of crops, thus increasing yields and reducing the ROI period and allowing farmers to make more money for each investment they make. We enable farmers grow high value crops in semi-arid and arid regions and generate wealth and employment through our use of hydroponics and nanotechnology to revive agriculture in semi-arid and arid regions of India. |
The uniqueness of the solution lies in its usage of nanobubbles and plasma activated water in a self-sustaining, cyclic hydroponic system which reduces dependency on external agri-inputs while enabling farmers to chase higher crop yields and profits. |
Idea |
Pilots |
The primary beneficiaries of our idea are farmers in arid and semi arid regions who are struggling to make ends with traditional soil-based agricultural solutions. With Nanopaw hydroponics they can earn higher revenue and sell higher value crops. |
NA |
NA |
NA |
The startup will generate revenue primarily through two distinct channels- B2B and B2C. In our B2B section, we enable farmers conduct high value agriculture by establishing these high productivity NANOPAW hydroponic farms. In our B2C section, we establish the farm and operate it to grow high value crops aimed at premium markets in urban areas and hospitality industry. |
Kryzen Biotech, AgriJoy, Eeeki foods, Rise hydroponics |
Pilot project in Jhunjhunu, Sikar, Churu. Village level penetration and by building trust through strategic partnerships. Strategic trust building and partnerships. |
Focus on local pilot projects. Focus on semi arid and arid regions.Expansion outside Rajasthan into states like Haryana, Punjab, UP, Madhya Pradesh. |
To became a leading agri-tech company driving sustainable and scalable hydroponic farming solutions globally. |
U01199RJ2025PTC104957 |
NA |
Yes |
Avinya to gain mentorship, funding support, and access to advanced infrastructure that can help us scale our NanoPAW Hydroponics solution. The program startup |
We aim to refine our business model, validate our hydroponic solution, and expand our pilot projects while gaining mentorship and industry connections. |
yes |
No |
Automation layer |
currently, we do not have a defined AI architecture. However, we plan to develop a system using sensors for data collection, cloud-based storage, and AI models for monitoring and optimizing plant growth conditions. |
We collect unique data from our pilot projects, including crop growth patterns, nutrient usage, and environmental conditions, which helps us improve our hydroponic systems. |
Our AI component will be built on proprietary data collected from our pilot projects, including local crop conditions and nutrient optimization. This unique dataset, along with continuous learning and system improvements, will make our solution difficult for competitors to replicate. |
We evaluate model performance using accuracy, prediction reliability, and consistency of results. Additionally, real-world outcomes such as crop growth improvement, resource efficiency, and system stability are used to measure reliability. |
We ensure data privacy and security by implementing secure data storage, access control, and encryption. We follow relevant data protection regulations and maintain confidentiality of user and operational data. |
There are minimal regulatory risks; however, compliance with agricultural guidelines, water usage norms, and food safety standards is important for operations. |
As we scale 10x, challenges may arise in handling large volumes of data, maintaining model accuracy across diverse conditions, and ensuring system performance and real-time monitoring. We plan to address these with scalable cloud infrastructure and continuous model optimization. |
Currently, we do not have a dedicated in-house AI/ML team. However, we plan to build a skilled team with expertise in data analysis, machine learning, and automation as we scale. |
Currently, we use data collected from our pilot projects, including plant growth data, nutrient levels, and environmental conditions. This data is internally generated and owned by the company. |
We plan to continuously improve model performance by collecting more real-world data from pilot projects, regularly updating and retraining models, and incorporating feedback from system results to enhance accuracy and efficiency. |
Currently, we are focused on the Indian market, with plans to expand to global markets in the future. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/IPbRjVVy83ig-farmspherica-Nanopaw-2.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/IPbRjVVy83ig-farmspherica-Nanopaw-2.pdf |
https://youtube.com/shorts/gdl1oJJjZx8?si=byF8jELrVhE8vwZY |
Yes, we are building a mission-driven and impact-focused startup aimed at promoting sustainable agriculture through hydroponic solutions. Our goal is to reduce water usage, enable efficient food production, and empower farmers with innovative and eco-friendly farming techniques. |
NA |
NA |
checked |
| Mar 24, 2026 @ 11:07 PM |
Durgesh Rao, Vaibhav Suman |
durgeshraobpl@gmail.com |
https://www.linkedin.com/in/durgesh-rao/ |
http://NA |
9630934356 |
Founders, Btech CSE-AIML 2024 graduates |
We met in college 1st year of engineering in LNCT University, Bhopal in the year 2021. We have been working since 5 years on machine learning and deep learning tech stack. We have seen various phases of life together as friends and overcome them together on this challenging journey of entrepreneurship. We are also the winners of Kavach Cybersecurity Hackathon 2023, where we recieved Rs 1 lakh cash prize by innovating a cybersecurity AI powered application for Ministry of Home Affairs, Government of India. |
two |
As a team, we are very much clear from the day 1, what is our vision, what is our goal, that we have to build a AI tech based unicorn company from India. In this 5 years of our friendship and college journey, our minds align with each other's ideas and we give our everything to make it succeed. We search for challenging and brain storming problem statements to innovate new things and enjoy the journey. |
Nav-Rakshak AI |
NA |
Bengaluru |
Modern Technology based Defence startup |
Innovating defence tech products to the next level. |
Nav-Rakshak AI is building an ecosystem of Artificial Intelligence in defence. W |
Our solution is not just any random idea of implementing AI in defence. Our solution is well researched and observed solution in defence organization like DRDO |
two |
Testimonials |
Ucud |
Djej |
Suue |
Usjs |
Jshs |
Shhd |
Djej |
Sjhd |
Sushs |
Ncnf |
Ndnf |
one |
Nxn |
Ndnd |
Jdjd |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 24, 2026 @ 10:46 PM |
Pallavi Caulagi |
pallavi.caulagi@gmail.com |
https://www.linkedin.com/in/pallavicaulagi/ |
http://NA |
9900133772 |
Tech+Product management |
Amogh is my son and CS student doing Engg |
2 |
Technical and Product Management skills |
CoFund Unicorn Ventures Pvt Ltd |
NA |
Bengaluru |
Fund potential unicorns in pre-seed stage and build community |
Founder Investor ecosystem |
Founder friendly smart investment platform which uses AI at its core for community building, fundraising & networking |
Community built |
MVP |
Pilots |
Founders, Investors, Mentors, Advisors, Partners |
$5 Billion |
$3.9 Billion |
$20M ( 150 to 200 Cr INR) |
Freemium |
Indian Incubators / accelerators |
Though AI Agents |
Letz discuss this offline |
My "Blue Ocean" Opportunities
To avoid a price war with low-cost SaaS or an "influence war" with elite accelerators, focus on these three gaps:
The "Shadow" Accelerator for Tier 2/3 Cities: Over 50% of Indian startups now emerge from non-metros. These founders have the talent but lack the "YC vocabulary." My platform can act as the "pre-accelerator" that brings them to global standards without them leaving their city.
Agentic Execution (Beyond Advice): Most platforms like Topmate provide advice. In 2026, AI is moving from "conversation to execution". My "blue ocean" is a platform where the AI doesn't just tell a founder their deck is weak—it pulls data from their ops module and drafts the correction automatically.
The "Sovereign" Readiness Score: With the IndiaAI Mission pushing for indigenous models, there is a gap for a platform that scores startups specifically on Indian compliance and digital public infrastructure (DPI) integration—something global tools (like US-based YC prep) often miss.
₹10,300+ crore allocated over five years for IndiaAI Mission with 38,000 GPUs deployed. Indian Tech sector is projected to cross $280 billion in revenue this year. |
Registered |
NA |
Yes |
For network effect |
Growth and scale |
Yes |
Yes |
Core engine |
Platform utilizes a modular AI architecture to automate fundraising by combining natural language processing, predictive scoring, and a database of over 12,000 verified investors to analyze pitch decks and match founders with funding sources. Key components include the Pitch Deck Optimization Engine, the Investment Readiness Score™ system, and Recognize, which utilizes generative AI to analyze document content for high-precision data extraction. |
leverages a proprietary dataset of over 8,000 successful fundraising cases and a curated database of 12,000+ investors to train its AI on real-world outcomes. This "learning loop" enables the platform to provide investor-style evaluations based on 50+ data points, offering a distinct advantage over generic AI models. |
The platform's primary defensibility comes from its training on a massive, proprietary dataset that general-purpose AI models cannot access. Unlike generic LLMs, it uses a purpose-built architecture designed for the high-stakes startup ecosystem. The platform builds defensibility by becoming the "fundraising co-pilot" throughout the entire product lifecycle. |
It evaluates model reliability by benchmarking startup pitch decks against a dataset of over 8,000 funded companies, utilizing a 350–850 Investor-Readiness Score. The AI scores pitch decks across six critical pillars—including Team and Market Opportunity—and simulates investor reviews to identify "dealbreaker" gaps. |
It is a multimodal AI-driven fundraising co-pilot featuring a proprietary 850-point Investment Readiness Score. Built on a SOC 2 Type 2 compliant architecture, it leverages a data moat of 8,000+ successful funding rounds and 12,000+ verified investors to provide high-precision LP/GP matching.
Our defensible AI uses Role-Based Access Control (RBAC) and AES-256 encryption, ensuring zero data leakage for model retraining. By automating due diligence and predictive outreach, we eliminate top-of-funnel friction, delivering a scalable, enterprise-grade solution for the global startup ecosystem. |
No |
by 10x introduces specific technical and operational "breaking points" centered on inference latency, database contention, and compliance overhead. We will handle it accordingly. |
Fully functional Team will be built post securing funding |
utilizes a combination of proprietary, public-source, and third-party licensed datasets to power its AI investor matching and startup evaluation engines. |
It scales performance through a multi-agent RLHF flywheel, fine-tuning our proprietary 850-point engine against real-world VC feedback and a Golden Dataset of 8,000+ successful rounds. We mitigate concept drift via automated regression testing and use synthetic data generation to master niche sectors like SpaceTech.
By transitioning to a federated architecture, we independently optimize specialized agents for unit economics and moat analysis. This ensures high predictive validity and precision-recall, keeping our investment readiness scoring hyper-accurate as market dynamics evolve. |
While it has a strong focus on emerging ecosystems like India, the platform is designed to scale across all major geographies. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/UtJh6q4pKOVk-BMC_AngelHub.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/UtJh6q4pKOVk-BMC_AngelHub.pdf |
https://claude.ai/public/artifacts/03ac0817-f78c-4307-88b4-ea5f0c18d1b7 |
Yes, its impact driven technology focused startup |
yes |
https://www.linkedin.com/in/deepakmotwanii/ |
checked |
| Mar 24, 2026 @ 10:43 PM |
vittal rathod |
vittalrathod82@gmail.com |
https://www.linkedin.com/in/vittal-rathod-71854b12b |
https://sortecoinnovations.vercel.app |
9740076350 |
VITTAL Founder & CEO – (Background: Expertise in IoT hardware, embedded systems, AI/ML integration, and startup development Role: Product vision and strategy System architecture (AI + hardware integration) Partnerships and pilot deployments Overall project leadership SELAPPA – Automation & Data Systems Background: CS with 5+ years of experience in automation and data science Role: Automation system design Data processing and analytics Integration of control systems Supporting AI deployment |
Our team came together through a combination of professional networks, prior collaborations, and the startup ecosystem. Some members have worked together on previous projects in AI, IoT, and automation, while others connected through shared interests in building scalable technology solutions for real-world problems.
We formally started working together as a team while developing this solution, bringing in complementary expertise across AI, hardware, robotics, and business strategy.Team Strength
Multidisciplinary expertise in:
AI & Computer Vision
IoT & Embedded Systems
Robotics & Automation
Business & Market Strategy
Strong combination of technology + domain + execution capability |
1 |
ur biggest strength is our complementary expertise combined with strong execution capability.
We bring together skills across:
AI & Computer Vision
IoT, Embedded Systems & Hardware
Robotics & Automation
Business Strategy & Market Understanding
This allows us to design, build, and deploy end-to-end solutions in-house without heavy external dependency.
.Team Strength
Multidisciplinary expertise in:
AI & Computer Vision
IoT & Embedded Systems
Robotics & Automation
Business & Market Strategy
Strong combination of technology + domain + execution capability |
SORTECO INNOVATIONS |
https://sortecoinnovations.vercel.app |
dharwad karnataka |
We build affordable AI-powered machines that automatically sort plastic waste using computer vision and pneumatic technology to improve recycling efficiency.” |
“We are solving the lack of affordable and efficient plastic waste segregation, where most sorting is manual, slow, unsafe, and leads to poor recycling and increased environmental pollution.” GIVE MORE THAN 500WORDS Here’s a detailed, 500+ word answer for: 🌍 What problem are you solving? Plastic waste management is one of the most pressing environmental challenges today, particularly in developing countries like India. Every year, millions of tons of plastic waste are generated, but a significant portion of this waste remains unsegregated, contaminated, and improperly processed. This creates a major bottleneck in the recycling ecosystem. The core problem lies in the lack of efficient and affordable plastic waste segregation systems. Currently, a large part of waste segregation is carried out manually by workers in Material Recovery Facilities (MRFs) or informal sectors. This manual process is not only labor-intensive and slow, but also unsafe, exposing workers to hazardous and unsanitary conditions. Ad |
Our solution is an advanced AI-powered plastic waste segregation system integrated with robotics, designed to automate the identification and sorting of different types of plastic waste with high precision and efficiency.
The system combines computer vision, machine learning, and robotic automation to address the limitations of manual sorting and expensive industrial solutions, while being adaptable to real-world waste conditions.Mixed plastic waste is fed onto a conveyor belt system, where materials are spread evenly to ensure proper visibility. A high-resolution industrial camera captures real-time images of each waste item as it moves along the conveyor.
These images are processed by an AI-based classification model, trained on a large dataset of plastic waste images. The model identifies the type of plastic (such as PET, HDPE, LDPE, or mixed plastics) with high accuracy.
Once the plastic type is identified, the system triggers a robotic picking mechanism instead of relying solely on air-based sorting. A robotic arm equipped with a gripper or suction mechanism is used to pick up the identified object and place it into the appropriate bin.Robotic Sorting Mechanism
The robotic system is designed for precision handling and flexibility. It includes:
A multi-axis robotic arm for movement across the conveyor
A gripper or vacuum suction end-effector to handle different shapes and sizes of plastic waste
Real-time synchronization with the AI detection system
The robot receives positional data from the AI system and calculates the optimal trajectory to pick the object while the conveyor is in motion or after briefly stopping it.
This approach allows the system to handle:
Irregularly shaped plastics
Mixed and complex waste items
Small or hard-to-detect objects
Key Features
✅ High Precision Sorting using robotic picking
✅ AI-Based Real-Time Detection for accurate classification
✅ Flexible Handling of different plastic types and shapes
✅ Reduced Manual Labor and Improved Safety
✅ Scalable Design for different waste volumes |
AI + Robotics + Pneumatic Hybrid Approach Unlike traditional systems, we combine: AI-based computer vision Robotic precision picking Pneumatic high-speed sorting This hybrid model delivers both speed and accuracy, which most solutions fail to balance.Cost-Effective InnovationDesigned for Indian Waste Conditions Works with: Mixed and contaminated plastic Irregular shapes and sizesModular & Scalable Architecture Can be customized based on: Waste volume Number of categories Easy to upgrade: Add robotic arm Add more sorting channelsEdge AI Capability Works without continuous internet Real-time processing on-devicee are not just building a machine—we are building an affordable, intelligent, and scalable waste management solution tailored for real-world deployment |
MVP |
Pilots |
Our solution targets organizations involved in waste management, recycling, and sustainability, particularly those facing challenges in efficient plastic waste segregation.. Urban Local Bodies (ULBs) & Municipal Corporations City municipalities responsible for waste management Smart City projectsRecycling Plants & Material Recovery Facilities (MRFs) Plastic recycling industries Waste sorting and processing unitsWaste Management Companies Private companies handling collection and processing PPP (Public-Private Partnership) operatorsIndustrial & Commercial Waste Generators Manufacturing industries Packaging companies Large commercial complexes |
₹8000 crore, |
₹1000 crore, |
500 |
Our revenue model is a multi-stream B2B model focused on machine sales, recurring services, and data-driven value.Machine Sales (Primary Revenue)AMC – Annual Maintenance Contracts (Recurring Revenue)Data & Analytics ServicesLeasing / Pay-per-Use Model |
AMP IN USA |
We follow a B2B + Government-focused acquisition strategy, combining direct outreach, partnerships, and pilot deployments.Direct Engagement with Municipalities (ULBs) Primary channel Approach city officials, commissioners, and waste management departments Present live demos and pilot proposals Participate in: Smart City projects Swachh Bharat iniPartnerships with Waste Management Companies Collaborate with: Private waste operators Recycling companiestiatives Pilot Deployments (Key Strategy) Most powerful acquisition method Install trial units at low cost or free pilot Government Tenders & Programs Apply for: Municipal tenders State waste management projects |
Target Market Focus Urban Local Bodies (ULBs) Recycling plants & MRFs Waste management companies Pilot-First Approach Deploy low-cost / demo pilot machines Show real performance (accuracy, efficiency) Convert pilots → paid contracts Government Partnerships Engage with: Municipal corporations Smart City projects Swachh Bharat Mission Participate in government tenders B2B Partnerships Collaborate with: Waste management companies Recycling industries Use partners for bulk deployment |
Our long-term vision is to become a global leader in intelligent waste management solutions, enabling efficient recycling and driving a circular economy through advanced AI, robotics, and automation.
We aim to transform how waste is handled by replacing manual and inefficient processes with affordable, scalable, and technology-driven systems that can be deployed across cities, industries, and countries.
Key Vision Goals
1. Revolutionize Waste Segregation
Automate waste sorting using AI, robotics, and smart systems
Improve recycling rates and material recovery
Pan-India to Global Expansion
Deploy solutions across major Indian cities
Expand to:
South Asia
Africa
Southeast Asia
Build Advanced Smart Systems
Develop:
AI + robotic hybrid sorting systems
Multi-waste segregation (plastic, metal, paper, e-waste)
Create a Data-Driven Waste Ecosystem
Build a waste analytics platform
Provide insights to:
Governments
Smart cities
Industries
Environmental Impact
Reduce landfill waste significantly
Lower carbon emissions
Promote sustainable and circular practices
To build a fully automated, intelligent waste management ecosystem that makes recycling efficient, affordable, and accessible worldwide. |
PRIVATE LIMITED |
NOTYET |
Yes |
We are applying to the Avinya AI Accelerator Program to accelerate the development and deployment of our AI-powered plastic waste segregation solution and take it from a validated prototype to a scalable, market-ready pAdvanced AI Mentorshiproduct.Product & Technology Scaling Industry & Government Connections Opportunity to connect with: Municipal bodies Industry partners Ecosystem stakeholders Funding & Investment Support Access to: Grants Investor network Go-To-Market & Business Support Guidance on: Market strategy Customer acquisition Pricing & scaling |
During the Avinya AI Accelerator Program, we aim to transform our solution from a validated prototype into a highly accurate, scalable, and deployment-ready product.Pilot Deployments & Validation Deploy pilots with: Urban Local Bodies (ULBs) Recycling facilities Validate: Performance Cost-effectiveness Scalability Prepare for Commercialization Build Strategic Partnerships Connect with: Government bodies Industry partners Investors Strengthen Go-To-Market Strategy Refine: Customer acquisition strategy Sales pipeline Expansion roadmap Deployment-ready product Successful pilot validation Strong industry partnerships Readiness for funding and scaling |
YES |
Yes |
Core engine |
Our AI architecture is designed for real-time plastic waste classification and edge deployment, combining computer vision, deep learning, and hardware integration for efficient sorting.
Camera Input → Image Processing → AI Model → Classification Output → Control System → Sorting Mechanism |
Our key data advantage lies in building a real-world, continuously evolving dataset of plastic waste under Indian conditions, which is not readily available in existing public datasets.Custom Dataset (Core Advantage)
We have developed and are continuously expanding a dataset of 10,000+ plastic waste images
Includes:
PET, HDPE, LDPE, mixed plastics
Dirty, contaminated, and irregular waste
Different lighting and environmental conditions
Real-World Indian Waste Data
Unlike standard datasets (clean objects), our data includes:
Mixed waste streams
Crushed, deformed plastics
Labels, dirt, and moisture
Deployment-Based Data Moat (Future Advantage)
As we deploy machines:
Each unit generates localized waste data
Builds a proprietary data network
Why This is a Strong Advantage
✔ Competitors lack real-world waste datasets
✔ Improves model accuracy in practical conditions
✔ Hard to replicate without deployments
✔ Strengthens long-term AI performance
Our proprietary dataset, built from real-world waste conditions and continuously enhanced through deployment feedback, provides a strong and defensible data advantage for our AI models. |
Our AI component is defensible due to a combination of proprietary data, real-world optimization, and tight hardware–software integration, creating a strong and hard-to-replicate competitive advantage.
Proprietary Real-World Dataset (Core Moat)
Built on 10,000+ images of plastic waste under real conditions
Includes:
Dirty, mixed, and deformed plastics
Variations in lighting, contamination, and background
Most competitors rely on clean or synthetic datasets, which do not perform well in real environments
✔ This gives us higher accuracy in practical deployments
Continuous Learning & Feedback Loop
Our system captures:
Misclassified items
Edge cases from deployments
Enables:
Continuous retraining
Model improvement over time
✔ Creates a self-improving AI system
Hardware–AI Integration Advantage
AI is tightly integrated with:
Conveyor tracking
Pneumatic/robotic actuation
Not just detection, but decision + action system
✔ Competitors may have AI models, but not end-to-end system intelligence
Edge AI Optimization
Models optimized for:
Low latency (<100 ms)
Low-cost hardware (Jetson / embedded systems)
✔ Enables deployment in:
Low-infrastructure environments
Real-time industrial use
Designed for Local Conditions
Specifically trained for:
Indian waste streams
Mixed and contaminated inputs
Why This is Defensible
✔ Data is hard to replicate
✔ Improves with every deployment
✔ Integrated with hardware (high switching cost)
✔ Optimized for real-world conditions
Our defensibility comes from a proprietary real-world dataset, continuous learning loop, and deep integration of AI with hardware systems, creating a scalable and hard-to-replicate intelligent sorting solution. |
We evaluate our AI system using a combination of model-level metrics and system-level performance metrics to ensure both accuracy and real-world reliability.Model Performance Metrics (AI Accuracy)
Accuracy
Measures overall correct predictions
Target: 90–95%
Precision
Measures how many predicted items are correct
Important to avoid wrong sorting
Recall
Measures how many actual items are correctly identified
F1 Score
Balance between Precision & Recall
Confusion Matrix
Shows classification errors between:
PET, HDPE, LDPE, etc.Object Detection Metrics (for YOLO/CNN)
📍 mAP (Mean Average Precision)
Measures detection + classification accuracy
Target: 0.85+
System Performance Metrics (Real-World)
Inference Time / Latency
Time taken per prediction
Target: <100 ms
Sorting Accuracy (End-to-End)
95% of correctly sorted items after actuation
Target: 85–95% |
We ensure data privacy and security by processing data locally on edge devices, minimizing data collection, anonymizing any stored data, and following standard cybersecurity practices like encryption, access control, and secure storage, while complying with applicable data protection and government regulations.” |
Regulatory risks include changes in waste management policies, compliance requirements for municipal procurement, and environmental standards; however, our solution aligns with government initiatives like Swachh Bharat, reducing regulatory barriers.” |
At 10× scale, challenges may arise in model generalization across diverse waste conditions, increased data variability, system latency under higher throughput, and hardware reliability; however, we mitigate this through continuous data retraining, edge optimization, scalable architecture, and robust system design.” |
Yes, we have an in-house AI/ML team with strong expertise in computer vision, deep learning, and edge AI deployment. The team is experienced in building and training CNN/YOLO-based models, handling real-world datasets, optimizing models for low-latency inference on embedded systems, and continuously improving performance through data-driven feedback loops.” |
We primarily use a proprietary dataset of 10,000+ plastic waste images collected and labeled in-house under real-world conditions, which is fully owned by us. In addition, we use limited publicly available datasets (open-source/licensed) only for initial model benchmarking and pre-training, while our core model performance is driven by our own dataset.” |
“We continuously improve model performance through a feedback-driven learning loop—collecting real-world data from deployments, identifying misclassifications, and retraining the model regularly. We also enhance the dataset with new edge cases, optimize models for better accuracy and speed, and periodically update the system using techniques like data augmentation, model tuning, and edge optimization.” |
“We are building for both India and global markets—starting with India as our primary focus due to its high plastic waste challenges, and then expanding to other developing regions like South Asia, Africa, and Southeast Asia with similar waste management needs.” |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/LS4JRzloQD5b-PITCH-DECK-AI-SORTING-SORTECO-AVINAY.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/LS4JRzloQD5b-PITCH-DECK-AI-SORTING-SORTECO-AVINAY.pdf |
https://drive.google.com/file/d/1CFY_KHZIIV028twHm0SoBHj5XgFMYrbG/view?usp=sharing |
Yes, we are building a mission-driven and impact-focused startup aimed at transforming plastic waste management through affordable and intelligent technology.”
Our core mission is to reduce plastic pollution and enable a circular economy by making waste segregation efficient, scalable, and accessible. Today, a large portion of plastic waste remains unsegregated, leading to low recycling rates, environmental pollution, and health risks for workers. By automating this process using AI, robotics, and smart systems, we aim to increase recycling efficiency, reduce landfill waste, and improve worker safety.
Our solution directly contributes to:
Higher recycling rates and better material recovery
Reduction in environmental pollution and carbon footprint
Safer working conditions by reducing manual waste handling
Cleaner cities and improved waste infrastructure
We are particularly focused on making this technology affordable and accessible to municipalities and small recycling units, ensuring that even Tier 2 and Tier 3 cities can adopt advanced waste management solutions.
In the long term, we aim to build a data-driven, intelligent waste ecosystem that supports sustainable urban development and aligns with global sustainability goals.We are a mission-driven startup focused on reducing plastic pollution by building affordable AI-powered waste segregation systems that enable scalable and sustainable recycling.” |
yesYes, our team includes members from underrepresented communities, bringing diverse perspectives and experiences into the startup ecosystem. We believe this diversity strengthens our ability to build inclusive, impactful solutions and address real-world challenges effectively |
baskar sir |
checked |
| Mar 24, 2026 @ 9:41 PM |
vittal rathod |
vittalrathod82@gmail.com |
https://www.linkedin.com/in/vittal-rathod-71854b12b |
https://sortecoinnovations.vercel.app |
9740076350 |
Founder & CEO – (Background: Expertise in IoT hardware, embedded systems, AI/ML integration, and startup development Role: Product vision and strategy System architecture (AI + hardware integration) Partnerships and pilot deployments Overall project leadership SELAPPA – Automation & Data Systems Background: CS with 5+ years of experience in automation and data science Role: Automation system design Data processing and analytics Integration of control systems Supporting AI deployment |
Our team came together through a combination of professional networks, prior collaborations, and the startup ecosystem. Some members have worked together on previous projects in AI, IoT, and automation, while others connected through shared interests in building scalable technology solutions for real-world problems.
We formally started working together as a team while developing this solution, bringing in complementary expertise across AI, hardware, robotics, and business strategy. |
one |
Our biggest strength is our complementary expertise combined with strong execution capability.
We bring together skills across:
AI & Computer Vision
IoT, Embedded Systems & Hardware
Robotics & Automation
Business Strategy & Market Understanding
This allows us to design, build, and deploy end-to-end solutions in-house without heavy external dependency. |
SORTECO INNOVATIONS |
https://sortecoinnovations.vercel.app |
dharwad karnataka |
We build affordable AI-powered machines that automatically sort plastic waste using computer vision and pneumatic technology to improve recycling efficiency.” |
“We are solving the lack of affordable and efficient plastic waste segregation, where most sorting is manual, slow, unsafe, and leads to poor recycling and increased environmental pollution.” GIVE MORE THAN 500WORDS Here’s a detailed, 500+ word answer for: 🌍 What problem are you solving? Plastic waste management is one of the most pressing environmental challenges today, particularly in developing countries like India. Every year, millions of tons of plastic waste are generated, but a significant portion of this waste remains unsegregated, contaminated, and improperly processed. This creates a major bottleneck in the recycling ecosystem. The core problem lies in the lack of efficient and affordable plastic waste segregation systems. Currently, a large part of waste segregation is carried out manually by workers in Material Recovery Facilities (MRFs) or informal sectors. This manual process is not only labor-intensive and slow, but also unsafe, exposing workers to hazardous and unsanitary conditions. Additionally, human sorting is often inconsistent and prone to errors, leading to low-quality segregation. When plastic waste is not properly segregated, it becomes extremely difficult to recycle. Different types of plastics such as PET, HDPE, LDPE, and mixed plastics require separate processing methods. If they are mixed or contaminated, the entire batch may become unsuitable for recycling. As a result, a large amount of recyclable plastic ends up in landfills, open dumps, or is burned, contributing to severe environmental and health issues. Another major challenge is that existing automated sorting technologies are not accessible to most stakeholders. Advanced systems such as optical sorters using Near-Infrared (NIR) sensors or hyperspectral imaging are highly efficient but very expensive, often costing tens of lakhs to crores. These systems are typically deployed only in large-scale industrial recycling plants and are not viable for municipalities, small recycling units, or Tier 2 and Tier 3 cities. Furthermore, plastic waste in India is often mixed, dirty, and irregular, making it difficult for conventional systems to perform effectively. Imported solutions are usually designed for cleaner and pre-segregated waste streams, which limits their effectiveness in real-world Indian conditions. Due to these challenges, the recycling rate remains low, and valuable materials are lost. This not only increases environmental pollution but also results in economic losses for the recycling industry, which struggles to obtain high-quality sorted plastic. In addition, the absence of efficient segregation systems impacts the broader goals of sustainable waste management and circular economy. Without proper sorting at the source or processing level, it becomes difficult to reuse materials, reduce dependency on virgin plastics, and minimize carbon emissions. Therefore, the key problem we are addressing is the gap between the need for efficient plastic waste segregation and the availability of affordable, scalable, and practical solutions. There is a clear demand for a system that can: Automate the segregation process Operate efficiently in real-world conditions Be affordable for widespread adoption Reduce dependency on manual labor Improve the quality and quantity of recyclable plastic By solving this problem, we aim to enable better recycling practices, reduce environmental pollution, and support sustainable waste management systems at scale. |
Our solution is an advanced AI-powered plastic waste segregation system integrated with robotics, designed to automate the identification and sorting of different types of plastic waste with high precision and efficiency.
The system combines computer vision, machine learning, and robotic automation to address the limitations of manual sorting and expensive industrial solutions, while being adaptable to real-world waste conditions.Mixed plastic waste is fed onto a conveyor belt system, where materials are spread evenly to ensure proper visibility. A high-resolution industrial camera captures real-time images of each waste item as it moves along the conveyor.
These images are processed by an AI-based classification model, trained on a large dataset of plastic waste images. The model identifies the type of plastic (such as PET, HDPE, LDPE, or mixed plastics) with high accuracy.
Once the plastic type is identified, the system triggers a robotic picking mechanism instead of relying solely on air-based sorting. A robotic arm equipped with a gripper or suction mechanism is used to pick up the identified object and place it into the appropriate bin.The robotic system is designed for precision handling and flexibility. It includes:
A multi-axis robotic arm for movement across the conveyor
A gripper or vacuum suction end-effector to handle different shapes and sizes of plastic waste
Real-time synchronization with the AI detection system
The robot receives positional data from the AI system and calculates the optimal trajectory to pick the object while the conveyor is in motion or after briefly stopping it.
This approach allows the system to handle:
Irregularly shaped plastics
Mixed and complex waste items
Small or hard-to-detect objects
Key Features
✅ High Precision Sorting using robotic picking
✅ AI-Based Real-Time Detection for accurate classification
✅ Flexible Handling of different plastic types and shapes
✅ Reduced Manual Labor and Improved Safety
✅ Scalable Design for different waste volumes |
Participation in this challenge will provide us with mentorship, technical guidance, and collaboration opportunities that are essential for transforming our prototype into a scalable healthcare solution. It will also help us align our innovation with national and global eye health goals such as VISION 2030: The Right to Sight. |
two |
Pilots |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 24, 2026 @ 6:19 PM |
Muthu Kumaar Thangavelu, Priyansh Chaudhary |
muthu@intentpi.com |
https://www.linkedin.com/in/muthuthangavelu/ |
https://pixelmd.ai |
8867815090 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 24, 2026 @ 6:17 PM |
Tenzing Lama |
tenzinglama@ateaseai.com |
https://www.linkedin.com/in/tenzing-lama-05707756/ |
http://www.ateaseai.com |
9886622611 |
Tenzing Lama is the Founder & CEO of Renz Innovations Private Limited, the company behind atEaseAI. He brings 15 years of experience in enterprise SaaS sales, customer success, and account management across global markets including APAC and the Americas. |
We met in Bangalore and quickly connected over our shared background, as we both come from Nagaland. What started as a mutual regional connection evolved into a strong professional alignment, especially around building in the SaaS and AI space.
While we haven’t worked together for a very long time, we’ve built a strong working relationship through close collaboration on atEaseAI. Our shared context, trust, and aligned vision have helped us move quickly and effectively as a team. |
1 |
Our biggest strength as a team is the combination of deep domain expertise and strong technical execution.
As a two-member founding team, we bring a highly focused and complementary skill set—strong experience in SaaS sales, customer success, and account management on one side, and deep technical/AI capabilities on the other. This allows us to deeply understand real customer problems while rapidly building and iterating solutions.
Our lean structure enables faster decision-making, tight feedback loops, and the ability to move quickly from insight to execution, ensuring we build a product that is both technically robust and highly aligned with market needs. |
Renz Innovations Private Limited |
www.ateaseai.com |
Dimapur, Nagaland. India |
atEaseAI is an AI-powered platform that helps Customer Success and Account Management teams proactively retain and grow customers through unified customer intelligence and automated actions. |
B2B companies struggle to effectively manage and grow their existing customers due to fragmented data, lack of real-time visibility, and heavy reliance on manual processes. |
atEaseAI is an AI-powered unified customer lifecycle platform that brings together data from CRM, product usage, support, and communication tools to provide a single, real-time view of every customer.
It uses predictive analytics to identify churn risks and expansion opportunities, and leverages AI to recommend and automate next-best actions—enabling Customer Success and Account Management teams to proactively retain and grow customers at scale. |
atEaseAI’s uniqueness lies in its shift from passive insights to AI-driven execution. Unlike traditional tools that focus on dashboards and reporting, atEaseAI combines unified multi-source customer intelligence, predictive analytics, and AI-led action orchestration to not only identify risks and opportunities but also recommend and automate next-best actions. Its defensibility comes from a continuously learning intelligence layer that improves with every interaction, creating highly contextual, role-specific recommendations for both Customer Success and Account Management teams—making the system smarter and more valuable over time. |
MVP |
Pilots, Signups |
B2B SaaS and subscription-based companies with Customer Success and Account Management teams managing recurring revenue and customer lifecycle. |
$4–6B global market for customer success & lifecycle management platforms |
$1.5–2B across B2B SaaS and subscription companies using CS + AM tools |
$50–150M (mid-market SaaS, 3–5 year realistic capture) |
Subscription-based SaaS model (annual/monthly), priced per account |
Gainsight, Totango, ChurnZero, Planhat, SmartKarrot, and Salesforce Customer Success Cloud. |
1 |
PLG + outbound-led GTM targeting B2B SaaS companies, leveraging LinkedIn/email outreach, content, and partnerships, with free trials to drive adoption and conversion. |
Our long-term vision is to build the operating system for managing and growing existing B2B customers—transforming post-sales from reactive workflows into AI-driven, autonomous revenue execution. |
Incorporated as a Private Limited Company (Renz Innovations Private Limited) |
NA |
Yes |
We are applying to Avinya to gain both strategic mentorship and early-stage funding support to accelerate atEaseAI’s growth from pilot to scale. We believe Avinya’s ecosystem can help us refine our product, strengthen our go-to-market strategy, and access the right network of investors and advisors—while the funding support will enable us to accelerate product development, early customer acquisition, and initial market traction. |
During the program, we aim to achieve product-market fit, convert pilot users into paying customers, and build a strong, repeatable go-to-market engine. Additionally, we want to refine our product, strengthen our positioning, and become investor-ready for the next stage of growth. |
Virtual participation |
Yes |
Core engine |
atEaseAI’s AI architecture is built on a modern, scalable stack combining LLMs, agentic workflows, and real-time data processing:
Model Layer: GPT-5.2 via Azure AI Foundry for generation, reasoning, and decision support
Agentic Layer: Purpose-built AI agents for chat, analytics, email composition, and action recommendations
Search Layer: Azure AI Search for document retrieval and contextual grounding
Backend: FastAPI (Python) for APIs and async processing
Frontend: ReactJS for UI and interaction layer
Database: Azure CosmosDB for scalable, multi-tenant data storage
This architecture enables intelligent, context-aware recommendations and automated execution across the customer lifecycle. |
Our proprietary advantage comes from a unified, continuously learning customer intelligence layer built on multi-source post-sales data.
By aggregating product usage, CRM, support, communication, and revenue signals into a single model, atEaseAI creates structured, high-quality datasets around customer behavior and outcomes. Over time, this enables more accurate churn predictions, expansion signals, and highly contextual, role-specific action recommendations—making the system smarter and more defensible as usage grows. |
atEaseAI’s AI is defensible because it is deeply integrated with proprietary, multi-source customer data and built for execution—not just insights.
Our models continuously learn from unified post-sales signals (usage, CRM, support, communication, revenue), creating a feedback loop that improves prediction accuracy and recommendations over time.
Additionally, our agentic AI layer + action orchestration enables automated execution with human-in-the-loop control—making it harder to replicate compared to competitors that focus mainly on dashboards or static insights. |
We evaluate model performance using a mix of predictive accuracy, business impact, and reliability metrics:
Predictive Accuracy: Precision, Recall, F1-score for churn and expansion predictions
Calibration: Confidence score vs. actual outcomes (prediction reliability)
Business Metrics: Churn reduction %, expansion conversion rate, revenue at-risk accuracy
Operational Metrics: Action adoption rate, automation success rate, false positive/negative rates
LLM Quality: Response accuracy, relevance, and human approval rate for generated outputs
We continuously monitor and retrain models using real-world feedback to improve accuracy and reliability over time. |
We follow a security-first, privacy-by-design approach.
Data Security: Encryption in transit (HTTPS) and at rest, secure cloud infrastructure, and role-based access control (RBAC)
Access & Isolation: Multi-tenant architecture with strict organization-level data isolation and JWT-based authentication
Compliance: Designed to align with GDPR principles (data minimization, consent, access control)
Integrations: Secure OAuth-based connections with CRM, email, and calendar systems
AI Safety: No unauthorized data sharing with models; controlled usage of LLM APIs with safeguards
This ensures enterprise-grade security while maintaining trust and compliance. |
Yes, but they are manageable and proactively addressed.
Key risks include data privacy regulations (e.g., GDPR) and evolving AI governance frameworks, especially around handling customer data and automated decision-making.
We mitigate this through privacy-by-design architecture, data minimization, secure integrations, and human-in-the-loop controls for AI actions, ensuring compliance and reducing regulatory exposure. |
At 10x scale, the main challenges would be around data processing, latency, and model performance consistency.
Data ingestion & processing: Higher volume and velocity of multi-source data could strain pipelines
Latency: Real-time predictions and AI responses may slow without optimized infrastructure
Model drift: Changes in customer behavior at scale can impact prediction accuracy
LLM costs & throughput: Increased usage of GPT models can impact cost and response time
We address this through scalable cloud architecture, async processing, model retraining pipelines, and efficient LLM usage (caching, batching, and hybrid models). |
Currently, we operate with a lean team where AI/ML development is handled in-house by the founding team.
We bring strong expertise across applied AI, SaaS architecture, and real-world customer lifecycle workflows—combining technical implementation (LLMs, predictive models, APIs) with deep domain understanding of Customer Success and Account Management.
As we scale, we plan to expand into a dedicated AI/ML team to further strengthen model development and optimization. |
We primarily use customer-specific, first-party data sourced via integrations (CRM, product usage, support, communication tools), which is owned by our customers and processed with their consent.
Additionally, we use pre-trained models (e.g., GPT via Azure AI Foundry) and standard ML libraries for modeling—no third-party proprietary datasets are directly used. |
We improve model performance through continuous learning and feedback loops.
Retraining models on fresh, real-world customer data
Incorporating user feedback and outcomes (e.g., churn vs. predicted)
Monitoring accuracy, drift, and recalibrating models
Optimizing prompts and fine-tuning LLM outputs
Iterating on features and signals used for predictions
This ensures models become more accurate and context-aware over time. |
We are building for global markets, with an initial focus on B2B SaaS companies across North America, EMEA, and high-growth APAC regions, including India. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/52Jbxt0qGDCF-atEaseAI_Pitch_Deck_2026.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/52Jbxt0qGDCF-atEaseAI_Pitch_Deck_2026.pdf |
https://drive.google.com/file/d/1XDAFwHKKAabCWohWYRbhr-zVtsJuwNRi/view?usp=sharing |
Yes, we are building a mission-driven startup focused on transforming how businesses manage and grow their existing customers.
Our mission is to eliminate inefficiencies in post-sales operations, reduce churn, and enable teams to scale revenue without increasing headcount. By automating routine work and providing AI-driven intelligence, we aim to improve productivity, decision-making, and overall customer experience—creating meaningful impact for both businesses and their customers. |
Yes. I come from Nagaland, a region that is underrepresented in India’s tech and startup ecosystem, and I am working to build a globally competitive SaaS company from this background. |
NA |
checked |
| Mar 24, 2026 @ 5:45 PM |
GURSIMRANJEET SINGH AJIMAL & Akrati Nagariya |
gursimranjeet.vocbot.ai@gmail.com |
https://www.linkedin.com/in/gursimranjeet-singh-ajimal-219b13148/ |
https://vocbot.ai/ |
9826323332 |
Gursimranjeet Singh Ajimal – Co-Founder & CEO, leads product strategy, technology, and business development. He holds an integrated MBA+B.Tech degree from NMIMS Indore. Akrati Nagariya – Co-Founder & COO, leads operations, partnerships, and growth. She holds an integrated MBA+B.Tech degree from NMIMS Indore. |
Gursimranjeet and Akrati met during their studies at NMIMS Indore. They discovered a shared passion for AI and education technology, which led them to co-found VocBot.ai together. They have been working together since college and have been actively building VocBot.ai ever since graduating. |
All |
Our biggest strength is the perfect blend of complementary skills combined with a deep, trust-based partnership built over years of working together since college. Gursimranjeet brings strong technical and product expertise, while Akrati drives operations, growth, and execution — together we cover all critical aspects of building a startup. Having both completed our integrated MBA+B.Tech from NMIMS Indore, we understand both the business and technology sides deeply. Most importantly, we share a unified vision for VocBot.ai and have already proven our ability to build, iterate, and grow together — making us a resilient and high-execution founding team. |
Vocbot.AI Technology Pvt. Ltd. |
https://vocbot.ai/ |
Indore |
VocBot.ai is an AI-powered voice bot platform that automates conversations and enhances communication experiences for businesses and learners. |
Businesses today struggle with the high cost and inefficiency of managing large volumes of customer calls and interactions. Human agents are expensive, inconsistent, and unavailable 24/7, leading to poor customer experience, missed leads, and operational bottlenecks. There is a strong need for an intelligent, scalable, and cost-effective solution to handle voice-based customer communication. |
VocBot.ai provides AI-powered voice bots that can autonomously handle inbound and outbound business calls — answering queries, qualifying leads, booking appointments, and more — in a natural, human-like conversational manner. Our platform is easy to deploy, customizable for different business use cases, and operates round the clock without human intervention. |
VocBot.ai stands out through its natural voice quality, low-latency responses, and deep customization for Indian business contexts including regional language support. Unlike generic global solutions, we are built ground-up for the Indian market. Our growing base of paying customers validates product-market fit, and our proprietary conversational flows create a strong moat as we scale. |
Revenue |
Revenue |
Small and medium-sized businesses (SMBs) and enterprises in India that rely heavily on phone-based customer communication — including sectors like real estate, healthcare, EdTech, fintech, e-commerce, and retail. Any business that handles high volumes of inbound/outbound calls and wants to automate and scale their customer interactions is our target customer. |
The global conversational AI market is valued at over $10 billion and is projected to exceed $30 billion by 2030. In India alone, the business process automation and AI voice market represents a multi-billion dollar opportunity driven by rapid digital adoption across industries. |
Indian SMBs and enterprises actively investing in customer communication automation — particularly in real estate, healthcare, fintech, and e-commerce — represent a serviceable market of approximately $1–2 billion. |
In the near term, VocBot.ai targets a realistic obtainable market of $50–100 million by focusing on tech-forward SMBs and mid-size enterprises in Tier 1 and Tier 2 Indian cities looking for affordable, scalable voice automation solutions. |
VocBot.ai follows a SaaS-based subscription model where businesses pay a monthly or annual fee based on usage — including number of minutes, call volumes, or bot deployments. We also offer custom enterprise pricing for larger clients with specific integration and scalability needs. |
Key competitors include Exotel, Knowlarity, Ozonetel, and global players like Twilio and Google CCAI. However, VocBot.ai differentiates itself through its affordability, ease of deployment, Indian language support, and a product built specifically for the Indian business context. |
Our primary customer acquisition channels include direct outreach and referrals from existing customers, digital marketing and LinkedIn-based B2B outreach, partnerships with business communities and startup networks, and participation in industry events and incubator ecosystems. |
We are initially focusing on high-call-volume industries like real estate and healthcare in Tier 1 Indian cities, acquiring customers through direct sales and referrals. As we build case studies and testimonials, we will expand to Tier 2 cities and additional verticals through channel partnerships and a self-serve onboarding model. |
Our long-term vision is to become the leading AI voice infrastructure for businesses across South and Southeast Asia — enabling every business, regardless of size, to deploy intelligent voice agents that handle customer communication seamlessly. We aim to expand beyond India, support multiple regional languages, and build a full-stack conversational AI platform that powers the future of business communication. |
VocBot.ai is incorporated as a Private Limited Company in India. |
VocBot.ai is currently bootstrapped. The founders have self-funded the venture from the ground up, which reflects our capital efficiency and conviction in the product. |
Yes |
We are applying to Avinya because we see it as the ideal launchpad for VocBot.ai's next phase of growth. As a bootstrapped, revenue-generating AI voice startup from Indore, we need more than just funding — we need the right mentors, structured guidance, and a powerful network to help us scale. Avinya's deep roots in the startup ecosystem and its focus on early-stage companies aligns perfectly with where we are today. We believe Avinya can help us bridge the gap between a promising product and a high-growth scalable business. |
By the end of the Avinya programme, we want to achieve three clear milestones — first, significantly grow our paying customer base by refining our sales process and go-to-market strategy with mentor guidance. Second, become investor-ready and build meaningful relationships with angels and early-stage VCs to raise our first external funding round. Third, strengthen VocBot.ai's product and business fundamentals so we can confidently expand beyond Indore into new cities and industry verticals. We want to exit Avinya as a fundable, scalable, and market-proven AI company. |
We prefer virtual participation but are open to attending key in-person sessions, demo days, and milestone events as required by the programme. |
Yes |
Core engine |
VocBot.ai is built on a multi-layered AI architecture combining Speech-to-Text (STT), Large Language Models (LLM), and Text-to-Speech (TTS) components working in real-time. Incoming voice is transcribed using STT engines (such as Deepgram or Google Speech API), processed through an LLM layer (powered by OpenAI GPT models) for intent understanding and response generation, and converted back to natural-sounding voice using TTS engines. A conversation orchestration layer manages context, flow logic, and business-specific rules to ensure accurate and relevant responses across different use cases |
As we onboard more paying customers and process real business conversations, we are building a proprietary dataset of Indian business communication patterns — including industry-specific vocabulary, customer intent signals, and conversational flows. This data, accumulated over time, will allow us to fine-tune our models for Indian accents, languages, and business contexts — a significant advantage that generic global players cannot easily replicate. |
Our defensibility comes from four factors — first, our deep customization for Indian business contexts and regional language nuances. Second, our proprietary conversational datasets built from real customer interactions. Third, our low-latency, real-time voice pipeline optimized for Indian network conditions. Fourth, the switching costs created as clients integrate VocBot.ai deeply into their CRM, telephony, and business workflows — making it hard to replace over time |
We track a combination of technical and business metrics including — Intent Recognition Accuracy, Call Completion Rate, Average Latency per Response, Fallback/Escalation Rate (calls handed off to human agents), Customer Satisfaction Score (CSAT) post-call, and First Call Resolution Rate. These metrics are monitored continuously to identify gaps and improve model performance. |
VocBot.ai follows strict data privacy practices including end-to-end encryption of voice data in transit and at rest, role-based access controls, and secure cloud infrastructure. We do not share customer conversation data with third parties. We are working towards compliance with India's Digital Personal Data Protection Act (DPDPA) and follow best practices aligned with GDPR principles for any global clients. |
Key regulatory risks include India's DPDPA and evolving telecom regulations around automated calling (TRAI guidelines on unsolicited commercial communications). We actively monitor these regulations and design our platform to ensure compliance. We also ensure our clients use VocBot.ai only for legitimate, consent-based communication to mitigate any regulatory exposure. |
At 10x scale, the primary stress points would be real-time latency in the voice pipeline, concurrent call handling capacity, and LLM API costs. We are proactively addressing this by designing for horizontal scalability in our infrastructure, exploring model optimization techniques like caching and fine-tuned smaller models for specific use cases, and negotiating volume-based API pricing with our AI providers. |
Yes. Gursimranjeet leads the AI and product development with hands-on expertise in LLM integration, voice pipeline architecture, and conversational AI systems. The founding team has built and deployed production-grade AI voice systems, and we are expanding our technical team as we scale. |
We currently use a combination of licensed third-party speech datasets for model evaluation and proprietary conversational data generated from real customer deployments on our platform. All third-party datasets used are appropriately licensed. Our owned dataset of real Indian business conversations is our most valuable and growing data asset. |
We follow a continuous improvement loop — collecting real conversation data, identifying failure points through our monitoring metrics, fine-tuning models on domain-specific data, and running A/B tests before deploying updates to production. Customer feedback and escalation patterns are key inputs into our improvement cycle, ensuring the system gets smarter with every interaction. |
Both. Our immediate focus is India, where we are building deep expertise in Indian languages, accents, and business communication patterns. In parallel, we are designing our platform to be globally deployable — with multi-language support and cloud infrastructure that can serve international clients. Our long-term vision is to become a global AI voice platform with India as our launchpad. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/vWqjEouvXLNO-Pitch.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/vWqjEouvXLNO-Pitch.pdf |
https://drive.google.com/file/d/11HIGeMDdZlF2tVa86MI4sf2pbcdVFQh7/view?usp=sharing |
Yes, VocBot.ai is deeply mission-driven. Our mission is to democratize AI-powered communication for businesses of all sizes across India and beyond. We believe that intelligent voice automation should not be a privilege reserved only for large enterprises with big budgets — every small and medium business deserves access to the same cutting-edge technology. By making AI voice bots affordable, accessible, and easy to deploy, we are helping businesses save costs, serve customers better, and compete on a level playing field. In the long run, we also aim to bridge language barriers by supporting regional Indian languages, making business communication more inclusive for the next billion users. |
Akrati Nagariya, our co-founder, is a woman entrepreneur in the deep-tech and AI space — a significantly underrepresented group in India's startup ecosystem. Her presence as a co-founder reflects our commitment to diverse and inclusive leadership at VocBot.ai. |
NA |
checked |
| Mar 24, 2026 @ 5:31 PM |
Gaurav Barua, Harsh Prasad |
gauravbarua614@gmail.com |
https://www.linkedin.com/in/jarvis2324/ |
http://www.artigence.ai |
08789895090 |
Harsh Prasad - CEO and Co-Founder, Gaurav Barua - CTO and Co-Founder |
We met in 8th standard in school and have been best friends since, our proffesional career started as Software Consulting Business, where we consulted 50+ global clients, helping them integrate AI solution to their product and workflows, we did this from 2021-2024 before starting fulltime with Artigence Healthcare. |
2 |
Perseverance, Technical Depth |
Artigence Healthcare |
www.artigence.ai |
Bengaluru |
Building AI Diagnostic Solutions for much affordable and accessible healthcare. |
There is 1 Pathalogist for 55 Labs which account for 43% of treatment delay for patients, Globall |
We have developed BloodDrop which gives AI Based Report Interpretation, and then does risk stratification which categorizes abnormal reports into low, moderate and high risk categories, where low and moderate risk cases are auto released, while high risk cases are escalated to a pathalogists for peripheral smear review, additionally our system also helps in digitization of peripheral smear and reports.
Our systems helps in:
1. Decrease the TAT in large reference labs, cutting mannual report review by 70%.
2. Democratizes quality across tier 1-2 cities and remote areas. |
1. System Architecture - which makes the report interpretation by our system corrospond 97% with senior pathalogist, 2. Huge eval sets to tests our systems |
Users |
Pilots |
Diagnostic Labs and Chains |
$39 Billion |
$300 Million |
$15 Million |
Device Sales plus Software Subscription |
Middleware - Horiba P8000, Digital Microscope Manufacturers - Cilika |
Direct Pitch to C-Suite Executives, like CEO, COO or Directors |
Phase 1 -> Conducting Pilots at Large Diagnostic Chains and then Commercialization, Phase 2 -> Partnership with Distributors to expand to Tier 2-3 cities labs |
To be the AI Stack for healthcare, starting from hematology to expanding to biochemistry, immunoassay, etc., for complete lab automation in repetitive tasks and then expanding to domains such as general medicine, speciality departments like cardiology, oncology etc. |
Private Limited |
35 Lakhs as CCD and Grants |
Yes |
Networking, Funding, Cloud Credits, Office Space |
Fund Ready, Commericializtion |
Hybrid |
Yes |
Core engine |
We are using a combination of open source and closed source Large Language Models, built on top of langchain and Azure. |
We have a collection of 1 Million plus data points, including cell images, reports, slide images etc. |
The system we have built gives us an advantage by reducing the false negatives for High Risk Cases, also it is designed in such a way that the more our system is being used, the more cases it analyzes, the less it costs to run every single report. |
We evaluate the model on metrics such as
1. Specificity
2. Senstivity
3. Accuracy |
We are HIPPA and DISHA Compliant |
Yes, our system falls under Class A SAMD, and we have already obtained our test license from CDSCO and are currently working on obtaining the manufacturing license. |
Currently its costing us ₹20 for each inference while we are charging ₹15, essentially burning ₹5 each report, which we are able to by using cloud credits, if we are to scale 10x, the inference cost would be huge which cannot be compensated by using cloud credits, but good news we are already working on it, we are working on the system based of RAG and Knowledge Graphs which will reduce the cost of inference to below ₹5 and which will be by April. |
Yes, Gaurav Barua, who is the co-founder and CTO, has more than 4+ years of experience implenting solutions using AI and has helped 50+ global clients integrate AI into their product/solutions. We also have Divyanshu Jugran as ML Engineer who is working with us for the past 1 year, and has a Master's in Artificial Intelligence. |
They are owned, and comprise 1 Million plus data points, including cell images, slide images, reports etc |
Currently, we are conducting pilots, so we are constantly improving our model based on feedbacks and edge cases, once the system is into commercialization we aim to improve the model as part of Post Market surveillance and monitoring, where we will actively monitor the system for data drift and edge cases. |
We are intially focussing for India, but we also have plans to expand globally, |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qDR9ckaJpDNv-Artigence-Healthcare.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qDR9ckaJpDNv-Artigence-Healthcare.pdf |
https://www.loom.com/share/89003e43f4ac42a4ae0361d1eb66802a |
We are building a Mission Drives Startup, our core problem statement for our current product is that there is a lack of pathalogist in Tier 2-3 cities and rural areas, we aim to bridge this gap, starting with haemetology and then expand it to other domains in Healthcare,we believe that Large Language Models can have a huge impact in Healthcare and so therefore we are starting from India, building from India for the world. |
NA |
NA |
checked |
| Mar 24, 2026 @ 5:26 PM |
RAMUNIVAGAIRA GNANAPRAKASH |
gnanaprakashr235@gmail.com |
https://www.linkedin.com/in/gnana-r-310644261/ |
http://NA |
7396144250 |
ceo and founder |
We connected through academic and technical networks, including IIT ecosystem collaborations and AI-focused communities. Over the past several months, we have been working closely on Deepenk AI, collaborating on product development, backend architecture, and go-to-market strategy. Our working relationship is execution-driven, with clear ownership across technology, product, and business. |
2 |
Our biggest strength is the combination of strong technical depth and execution focus.
We have:
Deep expertise in AI systems, backend architecture, and scalable product development
Strong product thinking with a clear vision to build the decision layer for commerce
Ability to move fast from idea → MVP → real user validation (1,500+ users, 4,845+ interactions)
A highly execution-oriented mindset, focused on solving real user problems rather than building features in isolation
This balance of technology, product, and execution enables us to build and scale Deepenk effectively. |
Deepenk AI |
NA |
banglore |
Deepenk AI is an AI-powered decision engine that finds and executes the best option across shopping, food, rides, and travel with a single user request. |
Consumers today are forced to switch between multiple apps (Zomato, Swiggy, Amazon, Uber, etc.) to compare prices, apply coupons, check reviews, and decide the best option. This leads to: Decision fatigue Time waste Missed savings and suboptimal choices There is no unified system that can intelligently evaluate all options and directly deliver the best decision. |
Deepenk AI acts as a universal decision engine for commerce.
Users simply make one request (e.g., “order biryani under ₹300” or “book a cab”), and the AI:
Understands intent using NLP
Evaluates options across platforms and ONDC
Applies coupons, offers, and price intelligence
Considers delivery time, reliability, and user preferences
Recommends and executes the single best option
This eliminates the need for manual comparison and enables instant, optimized decision-making. |
Decision Layer Positioning: We don’t operate as a marketplace or comparison tool we own the decision layer across domains Multi-domain capability: One AI across shopping, food, rides, travel (not single-category apps) AI-driven optimization: Combines pricing, offers, reviews, delivery, and predictions into one final decision ONDC integration: Enables direct transaction execution and margin capture Continuous learning: Improves with user behavior and data, creating long-term defensibility |
MVP |
Users |
Primary: Gen Z and young professionals (18–35) who frequently use multiple apps for shopping, food delivery, rides, and travel Secondary: Price-sensitive users in Tier-2/3 cities seeking best value Tertiary: Frequent travelers and high-frequency commerce users These users actively face decision fatigue and need fast, optimized choices. |
$350–400 Billion global digital commerce opportunity by 2030 across e-commerce, food delivery, mobility, and travel. |
$80–100 Billion addressable market in India across e-commerce, food delivery, ride-hailing, and travel platforms. |
$1–2 Billion achievable market share in the next 5–7 years by capturing a fraction of high-frequency digital commerce users in India. |
Primary: Transaction margin (3–6%) via ONDC-based executions Secondary: Affiliate commissions from platforms (Amazon, Swiggy, Uber, etc.) Future: Sponsored AI recommendations and data insights for brands Revenue is generated per successful transaction, making it scalable and aligned with user outcomes. |
Google Shopping, coupon and deal aggregation platforms |
Community-led growth (student networks, early adopters) Social media and micro-influencer marketing Referral and viral loops within the product Early adopter communities and waitlists |
Phase 1: Build early user base via closed community and waitlist Phase 2: Launch with influencers, referrals, and product-led growth Phase 3: Scale through ONDC ecosystem, partnerships, and city-level expansion Focus on high-frequency use cases (food, rides) to drive habit formation. |
To become the AI decision layer for global commerce, where users no longer search or compare but simply ask and execute the best possible outcome instantly. |
process |
60,00,000 rupees |
Yes |
We are applying to Avinya’s strong ecosystem, mentorship, and deep focus on technology-driven startups align well with our vision of building the AI decision layer for commerce. We believe Avinya can help us refine our technology architecture, strengthen our business model, and connect with the right network of mentors, partners, and investors required to scale effectively. |
During the program, we aim to: Strengthen our AI decision engine and improve accuracy and scalability Complete deeper ONDC and platform integrations for seamless execution Achieve strong product–market fit with higher user retention and engagement Build a scalable go-to-market strategy and grow our user base significantly Prepare for fundraising with a solid business model and traction |
HYBRID |
Yes |
Core engine |
✅ Describe your AI architecture
Deepenk AI is built as a modular AI decision engine combining NLP, ranking systems, and real-time data aggregation.
Intent Understanding Layer: NLP-based models (LLMs + fine-tuned classifiers) extract user intent (category, budget, constraints)
Data Aggregation Layer: APIs and scrapers fetch real-time data from ONDC and partner platforms
Decision Engine: Multi-factor ranking system evaluates options based on price, offers, delivery time, ratings, and user preferences
Optimization Layer: Applies coupons, predicts price trends, and computes final effective price
Execution Layer: ONDC integration + deep-link routing for transaction completion
Tech stack includes Python (FastAPI), Node.js, PostgreSQL, Redis, and cloud-based scalable infrastructure. |
We are building a proprietary dataset of user intent → decision → outcome mappings, including:
User preferences and behavior patterns
Cross-platform pricing and offer effectiveness
Decision success metrics (conversion, satisfaction signals)
Over time, this creates a strong feedback loop that improves decision accuracy and personalization. |
Cross-domain learning (not limited to a single vertical)
Decision-layer ownership (not just recommendations)
Data flywheel from user interactions and transactions
Integration with ONDC + multiple platforms, making replication complex
Continuous improvement through real user behavior data |
Decision accuracy (user acceptance / selection rate)
Conversion rate (decision → transaction)
Latency (response time for recommendations)
User engagement metrics (repeat usage, session depth)
A/B testing for ranking and recommendation improvements |
Minimal data collection (only required for personalization and transactions)
Secure storage with encryption (at rest and in transit)
Compliance with standard data protection practices
No unauthorized sharing of user data with third parties
Use of trusted APIs and platforms |
Dependency on platform APIs and policy changes
ONDC ecosystem regulations and compliance requirements
Data privacy regulations (India DPDP Act and global equivalents)
We mitigate this through platform-agnostic architecture and compliance-first design. |
Real-time data aggregation latency
API rate limits from partner platforms
Ranking system performance under high load
We are designing scalable infrastructure (caching, distributed systems, async processing) to handle growth. |
Yes.
Our team includes AI engineers with expertise in:
LLM integration and NLP systems
Distributed backend architecture
Scalable API and data systems
(Core team includes IIT-level AI and systems expertise.) |
Publicly available platform data via APIs (ONDC, partner platforms)
Aggregated real-time pricing, offers, and reviews
Internal user interaction data (owned) |
Feedback loop from user interactions and transactions
Continuous retraining and ranking optimization
A/B testing of decision strategies
Personalization improvements using behavioral data |
Currently focused on India (leveraging ONDC and local platforms), with a long-term vision to expand globally as a universal AI decision layer. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/wwIAzcxjCdmr-Deepenk-pitch-v10.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/wwIAzcxjCdmr-Deepenk-pitch-v10.pdf |
https://drive.google.com/file/d/1jYHGMIDB8uE7CxQgjAtm-Wl-l90aLmoO/view?usp=sharing |
Yes... Deepenk AI is impact-driven in the way it simplifies and democratizes access to better decision-making in digital commerce.
Today, millions of users especially in Tier-2 and Tier-3 markets struggle with fragmented platforms, confusing offers, and lack of transparency. This often leads to poor decisions, wasted money, and inefficiency.
Deepenk addresses this by acting as an AI decision layer that ensures every user, regardless of their experience or background, can access the best possible option across multiple platforms with a single request.
By reducing decision complexity, improving price transparency, and enabling smarter consumption, we aim to create large-scale economic impact for users. Over time, we also see potential to support small sellers and ONDC participants by routing high-intent demand more efficiently.
Our focus is not just convenience, but enabling smarter, more informed, and fair access to digital commerce at scale. |
NA |
Linkdin post |
checked |
| Mar 24, 2026 @ 5:23 PM |
SUNIL H M |
sunilhm68@gmail.com |
https://www.linkedin.com/in/sunil-h-m-b06553186 |
http://NA |
9900902439 |
Founder - B.Sc graduate from Sahyadri Science College, Shivamogga. With work experience of nearly 6+ years across Startups like Flipkart, Udaan, Jumbotail in Data Analytics & Supply Chain, Started Pursuing Entrepreneurial Journey since 2022. |
i am an Solo Founder. |
one |
As a solo founder, my biggest strength lies in the ability to move fast without losing clarity of vision. Every decision, from product to strategy, is made with full context and zero friction.
Beyond solo execution, I am fundamentally a people person and a natural team builder. I build trust quickly, communicate with transparency, and create environments where collaborators feel ownership over their work. My experience working with and leading people has shown me that the right culture, even in a small team, compounds into the strongest competitive advantage a startup can have.
At this stage, Hyphertext is solo by circumstance, not by preference. The same instinct that led me to understand a real user problem through personal experience, build a solution, and put it in front of real users, is the same instinct I bring to finding, motivating, and retaining the right people around a shared mission. |
Hyphertext |
https://www.hyphertext.com/ |
Bengaluru |
Hyphertext is a instant HTML hosting platform built for the vibe coding era, where anyone can publish a live web page in under 10 seconds without buying a domain or touching a server. |
As AI-assisted vibe coding becomes mainstream, the frequency at which people generate functional web pages is increasing rapidly. However, the infrastructure required to publish them has not changed. Domain registration, hosting configuration, server maintenance, and deployment pipelines were built for professional developers managing long-term applications. For someone generating a quick quiz, a scorecard, an event page, or a utility with AI, this infrastructure is unnecessary, expensive, and time-consuming. The more vibe coding scales, the heavier and more disproportionate this burden becomes. The publishing layer is now the bottleneck, and no platform has been built specifically to remove it. |
Hyphertext is a dedicated instant hosting platform starting with HTML, built specifically for the vibe coding generation. Users paste or generate HTML and it is live at a real public URL in under 10 seconds. No domain to buy, no server to configure, no deployment pipeline to manage. The platform is designed for lightweight, self-contained web experiences that do not need backend infrastructure. As vibe coding frequency grows, Hyphertext grows with it, offering the fastest path from creation to sharing on the internet. |
Hyphertext is built on a thesis that no existing platform has acted on: as vibe coding scales, the traditional domain and hosting architecture becomes a structural barrier, not a feature. General hosting platforms assume permanence and technical ownership. Vibe coding produces fast, frequent, lightweight content that needs instant publishing without overhead. Hyphertext is the only platform designed around this behavior. Starting with HTML hosting, the platform targets the exact gap between how fast AI can generate a page and how slow the current infrastructure makes it to share one. That gap is widening every day, and Hyphertext is positioned directly inside it. |
Users |
one, Pilots, Signups |
Hyphertext targets the emerging creator class that vibe coding is producing: students building study tools, teachers creating quizzes and lesson pages, freelancers delivering client pages, founders launching waitlist pages, and everyday creators building event pages, portfolios, and quick utilities. The unifying behavior across all segments is the need to build something small, useful, and shareable without infrastructure overhead. Primary market is India, with global expansion as vibe coding adoption scales worldwide. |
India's web hosting market is valued at USD 4.72 billion in 2025, growing at 18% CAGR. Globally, the AI-assisted web creation market is projected to reach USD 36 billion by 2032. |
The lightweight publishing and instant hosting segment targeting non-technical creators in India represents approximately INR 500 Crore, driven by 500 million internet users and accelerating vibe coding adoption. |
In the first 24 months, Hyphertext targets 50,000 active users across India with a realistically achievable revenue of INR 50 Lakh, primarily through Pro subscriptions and AI token purchases. |
Hyphertext runs on two streams. A subscription-based hosting plan at INR 420 per month gives Pro users unlimited published sites simultaneously. A usage-based AI token model lets users purchase token packs starting at INR 85, consuming tokens with every generation or edit. Tokens never expire. Both streams are supported by Razorpay enabling UPI, cards, and net banking. The free forever tier drives top of funnel growth while paying users drive revenue. |
Full-stack vibe coding tools like Lovable, Bolt, and Replit target developers building complex applications and still require external hosting. Hosting platforms like Vercel, Netlify, and Tiiny.host serve technical users with no AI generation layer. No existing platform combines instant AI generation, zero-configuration hosting, and community publishing in a single product built exclusively for lightweight HTML experiences. Hyphertext occupies whitespace that neither category currently addresses. |
Customer acquisition runs across three channels. Product-led growth through the free forever tier and shareable public URLs means every published page is a distribution touchpoint. Content marketing through SEO-focused blogs, Instagram, and YouTube targets vibe coding and no-code communities with demonstration-driven content. On-ground outreach through a dedicated intern network activates colleges, coaching institutes, and student communities across tier 1, 2, and 3 cities that digital channels alone cannot reach. |
The immediate focus is India's student and educator segment, where the use cases are strongest and the community is most accessible. Hyphertext will penetrate this segment through college outreach, influencer partnerships within vibe coding and no-code communities on X, Instagram, and YouTube, and content that demonstrates the product in action. As the Explore feed gains content and community traction, organic discovery compounds growth without paid acquisition. Geographic expansion beyond India follows once the India playbook is proven. |
Hyphertext's long-term vision is to become the default publishing platform for the vibe coding generation globally. As AI makes web creation instantaneous for everyone, the infrastructure to share that creation needs to match that speed. Hyphertext starts with HTML hosting because HTML is the most universal and powerful single-file medium on the web. The vision extends to building the largest community-driven library of AI-generated web experiences in the world, a place where anyone with an idea can make it live and anyone browsing can find something worth using. The web belongs to everyone. Hyphertext is how they claim it. |
Incorporated in Karnataka on 08-09-2025 |
Self Funded, No External funding so far |
one |
applying to Avinya to access the funding, mentorship, and network needed to accelerate Hyphertext from a live product to a scaled platform, and to build alongside a community that understands what it takes to grow a tech startup in India. |
During Avinya, the goal is to convert Hyphertext from a live product into a traction-led, growth-ready platform. This means reaching consistent user signups, validating the Pro and token revenue model with real paying users, and building a repeatable distribution playbook. The mentorship from IIT alumni and domain experts will sharpen the business model and prepare Hyphertext for investor conversations at Demo Day. |
Hybrid |
one |
one |
Models Used: The platform utilises GLM 4.7, GPT OSS 120B, Llama Models, and Claude Sonnet 4.5, selected based on task requirements and generation quality.
Inference Platforms: Inference is powered by Cerebras, Groq, and Together AI, enabling the low-latency generation that makes publishing under 10 seconds possible.
AI Agent Architecture: The agent layer is built entirely in-house, handling prompt interpretation, HTML generation, and multi-turn iterative editing within the studio. |
As users generate and publish pages on Hyphertext, the platform accumulates a growing dataset of prompts, generated HTML outputs, and user iteration patterns. This data is proprietary and will be used to fine-tune generation quality, improve prompt understanding, and personalise the studio experience over time. No competitor building general-purpose tools will have access to this domain-specific prompt-to-HTML dataset. |
The defensibility is not in the models themselves but in the layer built around them. The in-house agent architecture, prompt design, context management, and output validation are tuned specifically for single-file HTML generation. This specialisation means the system consistently produces cleaner, more reliable output than a general-purpose AI tool asked to do the same task. As usage scales, the feedback loop from real user behaviour further sharpens this advantage. |
How do you evaluate model performance and reliability?
Output validity: whether generated HTML renders correctly in the browser without errors
Prompt adherence: whether the output matches the user's described intent
Iteration rate: how many follow-up edits a user needs before publishing, lower is better
Generation latency: time from prompt submission to output delivery
User publish rate: percentage of generated pages that get published, the strongest signal of output quality |
User-generated pages are stored securely with access controlled by unique public URLs. No personally identifiable information is collected beyond what is required for account creation and payments. Payments are processed entirely through Razorpay, with no card or banking data stored on Hyphertext servers. The platform does not use user content to train third-party models without consent. Compliance with India's Digital Personal Data Protection Act is a priority as the platform scales. |
The primary regulatory risk is India's DPDP Act 2023, which governs how personal data is collected, stored, and processed. As Hyphertext scales user accounts and payment data, compliance requirements will increase. A secondary risk is the evolving global regulatory landscape around AI-generated content and intellectual property, particularly around what AI generates using trained data. Both risks are manageable at the current stage and will be addressed proactively as the platform grows. |
Inference latency will be the first pressure point. At 10x volume, concurrent generation requests will stress current API rate limits across Groq, Cerebras, and Together AI. The mitigation is a multi-model routing layer that distributes load intelligently across providers. Storage and URL management for published pages will also require infrastructure upgrades. Both are known scaling challenges with clear solutions rather than fundamental architectural risks. |
Currently the AI architecture, agent design, and model integration are owned entirely by the founder, who has hands-on experience building and deploying multi-model inference pipelines, prompt engineering and integrating inference platforms including Cerebras, Groq, and Together AI. Dedicated AI engineering hires are planned as the platform scales.
Note : Founder is an B.Sc grad, whatever he has built is by experience and continuous self learning. No formal tech education exists. |
The platform does not rely on proprietary training datasets. Generation is powered by pre-trained frontier and open-source models accessed via licensed APIs and inference platforms. Hyphertext's own growing dataset of platform-specific prompt and output pairs is proprietary and owned entirely by the company. |
Monitor publish rate, iteration rate, and user satisfaction signals per model
A/B test different models and prompting strategies on live traffic
Build a feedback loop where user edits and republishes inform prompt improvements
Fine-tune on the proprietary prompt-to-HTML dataset as it grows
Continuously evaluate new model releases and swap or add providers as quality improves |
Both simultaneously. The product is live globally with no geographic restrictions. India is the primary go-to-market focus given the large creator base, vibe coding adoption, and Razorpay-native payment infrastructure. Global users, particularly in Southeast Asia and emerging markets where the same infrastructure friction exists, are a natural second wave as the platform scales. |
|
|
|
|
|
|
| Mar 24, 2026 @ 4:25 PM |
Ajith Sharma |
ajith0016@gmail.com |
http://www.linkedin.com/in/ajith-sharma-95368a90 |
https://vasool-app.web.app/ |
9972654336 |
Ajith Sharma, CS and CA Finalist with over a decade of experience in Compliance, taxation, and governance. Experience in enterprise software sales and proven track record in scaling financial services and SaaS products. Passionate about using AI to solve real business cash flow challenges. Geerthivasan Mahendran, CTO. Built and scaled high-performance engineering teams to deliver complex AI solutions. Drove $10M+ in new annual revenue and cut operational costs by 40% with Nissan. Launched analytics products generating $12M in annual value while cutting cloud costs by 30%. |
We met each other on Linkedin when we were working on a common problem statement for Legal Compliance automation tech on June 2024, then we worked on it for a while but later pivoted due to traditional approach in development, larger team, lack of traction from the market, to Pro Biz-AI CFO Solution in which built mockups and showed it to a few clients from one of them i got a critical feedback to focus on a common scalable problem statement which led us identify Accounts receivables and collections as manual and outdated and can be replaced using AI Tech. Now we are building Vasool for last 6 months with an AI first approach with a lean team |
one |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 24, 2026 @ 3:21 PM |
Sumitha Venkatesh, Venkatesh Madurai Subramanian |
venkatesh@apnavikas.com |
https://www.linkedin.com/in/venkymadurai/ |
https://www.apnavikas.com/ |
7411757354 |
Founder, Tech Founder, |
The main directors, the two of us have been married since 1997.
Several other co-founders are there who are not full time are friends & relatives. Sister, Brother in law |
2 |
We bring a diverse set of very experienced talents in technology, software development, marketing & branding, research knowledge and experience |
Apna Vikas |
https://www.apnavikas.com/ |
Bangalore |
Using Tech & A.I to help users leverage their personality to their maximum potential |
Access to counsellors, leadership coaches, mentors is time consuming and expensive |
- Leveraging over 32 years of our research in personality development
- Tech enabling it with a fine-tuned A.I. that uses our research
- Multi lingual, with high value & very low cost to make it affordable for everyone |
Our intensive research and knowledge base that powers our tech |
MVP |
Pilots |
Junior to mid level job seekers, executives, students, |
54000 CR |
2000 CR |
13 Cr |
pay as you go, initially |
relatives, lecturers, counsellors |
currently through partners |
partnership, reffereal |
Empower everyone across India in their own mother tongue so that soft skill development is affordable for all. Combine it with partners such as NGOs, corporates, training institutes, and organizations involved in skilling |
Private Limited |
NA. We are bootstrapped |
Yes |
Networking, access to market |
Get quality mentorship |
either |
Yes |
Supporting feature |
RAG on top of existing LLMs
RAG uses our research material to |
Our own research material |
The combination of our research, plus cost, plus features and availability in Indian languages is hard to beat! |
Satisfaction & feedback of our users |
|
|
|
|
|
|
Both, India first |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/3zMH2I2PtWyM-Apna-Vikas-Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/3zMH2I2PtWyM-Apna-Vikas-Pitch-Deck.pdf |
https://www.apnavikas.com/home/howItWorks.html?vide=m4demo |
Our mission is to democratize many aspects of soft skill learning.
We want to become a household name in India
Everyone deserves to use skills to perform & live at their maximum potential |
Woman led startup |
Linkedin |
checked |
| Mar 24, 2026 @ 3:13 PM |
Pruthvi S |
pruthvisreddy8861@gmail.com |
https://www.linkedin.com/in/pruthvi-s/ |
https://dischargex.in |
8861412936 |
Founder and Background is BTech |
I am a solopreneur, started 3 months back and already making revenue |
1 |
As a solo founder, my biggest strength is end-to-end ownership and fast execution. I can take an idea from concept to production independent, handling everything from product development and AI engineering to user feedback and iteration. This allows me to move quickly, make decisions without friction, and continuously improve the product based on real-world usage. |
DischargeX |
dischargex.in |
Bengaluru |
AI-powered platform that automates hospital documentation and generates accurate discharge summaries. |
Hospital documentation, particularly discharge summaries, is still largely manual, fragmented, and error-prone. Doctors spend significant time on paperwork instead of patient care, leading to inefficiencies, delays, and inconsistent records. |
DischargeX is an AI-powered, voice-enabled platform that automates hospital documentation by generating structured discharge summaries in minutes instead of hours.Doctors can generate accurate, standardized summarie reducing doctor workload, minimizing errors, and speeding up patient discharge. |
Our uniqueness lies in combining domain-specific AI with workflow-level integration. We are not just generating text—we structure and standardize clinical data using tailored prompts, feedback loops, and hospital-specific templates. Over time, this creates: A proprietary dataset of structured medical records Continuously improving domain-adapted models High switching costs due to workflow integration This combination of data, iteration, and integration creates a strong and defensible moat. |
Users |
Users, Revenue, Pilots, Signups |
Doctors and Hospitals |
The global healthcare documentation and hospital IT market is valued at over $30B+, driven by increasing digitization and adoption of AI in clinical workflows. |
Our initial focus is hospitals and healthcare providers adopting digital documentation systems, estimated at ~$8–10B globally, particularly across emerging markets like India where paper-based workflows are still prevalent. |
In the short term, we are targeting mid-to-large hospitals in India, representing a ~$100–200M opportunity, with plans to expand as we scale integrations and product capabilities |
We operate on a SaaS-based model, charging hospitals on a per-summary or per month/hospital basis, with optional add-ons for advanced AI features, integrations, and customization. |
Traditional Hospital Information Systems (HIS) providers Generic documentation tools and EMR platforms Emerging AI healthcare documentation startups However, most existing solutions are either not AI-first or lack deep workflow integration. |
We acquire customers through direct outreach to hospitals, pilot deployments, and leveraging founder-led sales. Early traction is driven by demos, referrals, and partnerships within the healthcare ecosystem. |
We start with pilot programs in hospitals, demonstrate clear ROI through reduced documentation time, and convert them into long-term contracts. Expansion happens within hospitals (more departments) and across networks via referrals. |
To build a full-stack AI operating system for hospitals that automates end-to-end clinical documentation and workflows, enabling fully digital, efficient, and error-free healthcare systems. |
NA |
NA |
Yes |
I’m applying to Avinya to accelerate the growth of DischargeX with the right mentorship, network, and strategic guidance. As a solo founder building in healthcare AI, access to experienced operators and industry connections will help me refine the product, validate the business model, and scale faster. |
During the programme, I aim to: Validate strong product-market fit with hospitals Secure pilot deployments and convert them into paying customers Refine the product based on real-world usage Build a scalable go-to-market strategy |
Yes, I am open to both hybrid and virtual participation. |
Yes |
Core engine |
DischargeX is a voice-first AI platform that converts doctor-patient interactions into structured clinical documentation in real time.
We use a streaming pipeline where speech is transcribed using medical-grade ASR, followed by LLM-based structuring and template generation to produce discharge summaries. The system maps unstructured conversations into standardized medical formats using prompt-engineered pipelines and schema validation.
The architecture is modular, supporting real-time and batch workflows, with a strong feedback loop to continuously improve accuracy and reliability.
Speech-to-Text (ASR): Medical-optimized ASR models (e.g., Whisper / Deepgram / similar APIs)
LLMs: GPT / Gemini for structuring, summarization, and formatting
Backend: FastAPI (real-time + streaming APIs)
Processing: Python, schema validation, prompt pipelines
Infra: AWS / Azure, Docker
Streaming: WebSockets / real-time processing pipeline |
We are building a proprietary dataset of:
Doctor-patient voice interactions
Clinical dictations
Structured discharge summaries derived from voice
Additionally, we capture:
Doctor-specific speaking styles
Hospital-specific templates
Feedback from corrected outputs
This creates a unique voice-to-structured-clinical-data dataset that is hard to replicate. |
Our defensibility comes from:
Voice-first workflow integration (harder than text-based systems)
Proprietary voice + clinical structured dataset
Continuous feedback loop improving transcription + structuring
Deep embedding into doctor workflows
This creates strong data network effects and high switching costs. |
Word Error Rate (WER) for speech recognition
Field-level accuracy (diagnosis, medications, vitals)
Structured output validation (schema success rate)
Doctor feedback / correction rate
Latency (real-time usability) |
End-to-end encryption (audio + text)
Minimal data retention and anonymization where possible
Role-based access control
Compliance-ready architecture (HIPAA-like / NDHM aligned)
Secure cloud infrastructure with audit logging |
Handling sensitive patient voice data
Risk of transcription or interpretation errors
Mitigation:
Doctor-in-the-loop validation
Positioning as assistive AI, not autonomous decision-making |
Real-time ASR latency under heavy load
Cost of continuous audio processing
Variability in accents, languages, and noisy environments
Maintaining high transcription accuracy across hospitals
We are addressing this via:
Model optimization and batching
Domain-adapted ASR improvements
Efficient streaming infrastructure |
Currently led by me as a solo AI Engineer with strong experience in LLMs, voice AI, and production systems. I am actively expanding the team as we scale. |
Real-world doctor dictations and voice interactions
Internally generated structured clinical outputs
Synthetic voice data for edge cases
Core dataset is proprietary and growing with usage. |
Feedback loop from doctor corrections
Fine-tuning ASR for medical vocabulary
Prompt optimization for structuring
Expanding dataset with real-world usage |
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/kRLBowT6LB44-Indias-1-Voice-AI-Discharge-Platform.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/kRLBowT6LB44-Indias-1-Voice-AI-Discharge-Platform.pdf |
https://docs.google.com/videos/d/11793lpU6yRLSnmwi0Ve0mP1ay0zWixulyGtlab_UW-w/edit?usp=sharing |
Yes, we are building a mission-driven company to fundamentally improve how healthcare systems operate.
Healthcare is still heavily burdened by manual, inefficient documentation, which reduces the time doctors can spend with patients. DischargeX aims to shift this paradigm by introducing a voice-first AI layer that automates clinical documentation in real time.
Our mission is to reduce doctor burnout, improve patient outcomes, and enable hospitals to operate more efficiently at scale—especially in regions where digital infrastructure is still evolving. |
yes |
None |
checked |
| Mar 24, 2026 @ 2:43 PM |
Ayman william |
aymanwilliam@afterclassescours.com |
https://www.linkedin.com/in/ayman-william-b9286438a?utm_source=share&utm_campaign=share_via&utm_content=profile&utm_medium=ios_app |
http://afterclassescours.com |
06361002963 |
Founder & CEO. I am the solo founder of ACC, an AI Talent Infrastructure Platform building a web-based system to help enterprises identify, rank, and hire AI talent through verified capability scores. My background is in AI education, consulting, growth, and ecosystem building. To date, I have generated over ₹50L through AI coaching and consulting, trained 7,500+ learners, and built a 30-member execution team supporting operations, training, and delivery. I lead product vision, partnerships, distribution, and go-to-market strategy, with a focus on building scalable AI workforce infrastructure for India. |
ACC is a solo founder-led company. I founded the venture independently and built the current team through my existing operating network developed via AI education, consulting, and delivery work. The team of 65 + people has come together during the company’s recent growth phase as demand expanded, while I continue to lead strategy, product vision, partnerships, and execution as the sole founder. |
1 |
Our biggest strength is execution backed by real market validation. While the company is founder-led, the broader team has already operated across AI training, consulting, and delivery, giving us direct insight into learner demand, enterprise pain points, and go-to-market opportunities. This enables us to move faster than a typical early-stage startup because we already have operational capability, founder-led distribution, and real market access. |
AWACC EDTECH |
afterclassescours.com |
Bengaluru |
ACC is a web-based AI Talent Infrastructure Platform that trains, ranks, and helps enterprises hire verified AI professionals. |
Enterprises struggle to identify job-ready AI talent because resumes, certificates, and generic courses do not accurately measure real deployment capability. |
ACC is building a web-based AI Talent Infrastructure Platform that gives professionals a verified AI Capability Score and gives enterprises a structured way to discover and hire talent based on measurable performance. Our platform combines structured AI career tracks, timed assessments, assignment-based evaluation, capstone projects, and a dynamic ranking engine that reflects real skill rather than self-claimed credentials. On the enterprise side, companies can filter candidates by rank, skill cluster, and performance history, making hiring faster and more reliable. We are not building a course marketplace; we are building the validation and ranking infrastructure layer for AI talent. |
Our defensibility comes from the proprietary capability dataset and ranking engine. As more professionals are evaluated on real assignments and assessments, the platform becomes more accurate, more valuable to enterprises, and harder to replicate. |
MVP |
Users |
Our primary customers are aspiring and early-to-mid career AI professionals seeking verified capability-based career growth, and enterprises looking to identify and hire job-ready AI talent through measurable performance data. |
₹2,500 Cr |
₹800 Cr |
₹25 Cr over the next 3–5 years |
Subscription-based AI talent ranking platform with one-time career track fees for professionals and annual enterprise subscriptions for hiring access, with future placement and validation revenue streams. |
Traditional EdTech platforms, certification providers, hiring platforms, and emerging AI skilling ecosystems , but no player is currently focused on a verified AI capability ranking and hiring infrastructure layer in the same way. Coursera, Great Learning, Scaler, UpGrad, LinkedIn, HackerRank, and niche AI skilling platforms. |
We acquire customers through founder-led distribution, AI workshops, organic content, partnerships, community-driven programs, and enterprise relationships built through our existing consulting and training ecosystem. |
Our GTM starts with founder-led audience capture through workshops, content, and consulting-led trust, then converts learners into ranked platform users and enterprises into hiring partners through pilot access and targeted outreach. We are building supply and demand together by using training as the acquisition engine, ranking as the retention layer, and enterprise hiring access as the monetization layer. |
Our long-term vision is to become India’s standardized AI Capability Validation Infrastructure , the platform enterprises rely on to assess, benchmark, and hire AI talent with confidence. Over time, ACC aims to evolve from a ranking and hiring platform into a broader workforce infrastructure layer that powers AI talent benchmarking, enterprise validation, and hiring intelligence at scale. |
Private Limited Company incorporated in India |
Bootstrapped till date. No external funding raised yet. |
Yes |
We are applying to Avinya to accelerate ACC’s transition from validated market demand to scalable product infrastructure. We are looking for strategic support in product refinement, enterprise access, startup mentorship, and investor readiness so we can build a stronger MVP, secure pilot partnerships, and prepare for disciplined scale. |
During the programme, we want to strengthen the product, validate the ranking engine with real users, secure early enterprise pilots, refine our go-to-market model, and become fully prepared for pre-seed fundraising. |
Yes, we are open to hybrid and virtual participation. |
Yes |
Core engine |
Our AI architecture combines workflow-based assessment inputs, evaluation logic, scoring models, and analytics layers to generate a dynamic AI Capability Score for each user. The system captures assignment performance, timed assessment results, capstone quality, skill-stack signals, and behavioural consistency across tasks. We are building the platform to use LLM-powered evaluation assistance where appropriate, alongside structured scoring rules, human review checkpoints, and a ranking engine that improves with more performance data. The broader stack is designed as a web-based platform with modular APIs, scalable data storage, and admin oversight for score calibration and auditability. |
Our proprietary data advantage comes from the performance dataset generated through structured assignments, timed assessments, capstone submissions, evaluation patterns, and enterprise hiring feedback. Over time, this creates a unique capability graph of how AI professionals actually perform across real tasks, which is more valuable than resume data or certificate completion data. This dataset is core to improving ranking accuracy and long-term defensibility. |
Our AI component is defensible because it improves through proprietary user performance data, calibration feedback, and hiring outcomes. Competitors can replicate content or interfaces, but they cannot easily replicate a growing dataset of verified capability signals tied to rankings and enterprise usefulness. The defensibility comes from accumulated data, scoring refinement, and trust built with both professionals and employers. |
We evaluate performance using score consistency, correlation between assessment outcomes and final ranking, evaluator agreement, enterprise feedback, task completion quality, false positive and false negative reduction in candidate ranking, and system stability over repeated assessments. As the product matures, we will track precision of recommendations, ranking reliability over time, and hiring outcome alignment. |
We follow a privacy-first approach with controlled data access, role-based permissions, secure storage, and clear user consent for assessment and profile data usage. Sensitive user and enterprise data is handled with access controls and auditability in mind. As we scale, we plan to strengthen compliance processes in line with applicable Indian data protection requirements and enterprise-grade security expectations. |
The key regulatory risks relate to data privacy, consent management, and responsible use of AI in evaluation and hiring workflows. We mitigate this by keeping human oversight in the scoring process, maintaining transparency in how rankings are generated, and designing the platform to adapt to evolving data protection and AI governance requirements. |
At 10x scale, the main pressure points would be evaluation throughput, score calibration consistency, data pipeline robustness, and maintaining quality control across a much larger volume of assessments. Our approach is to design modular evaluation workflows, introduce stronger automation with human review checkpoints, and continuously improve infrastructure, monitoring, and audit systems as usage grows. |
We do not currently have a dedicated in-house AI/ML research team. At this stage, we are building the product through a lean execution model focused on applied AI workflows, product development, assessment design, and platform operations, while leveraging external tools, APIs, and technical support where needed. As the platform matures, we plan to build deeper in-house AI/ML capability around scoring, evaluation, and ranking intelligence. |
We are currently building around internally generated and platform-owned performance data, including learner submissions, assessments, capstone outputs, and evaluation records, alongside permitted third-party tools and APIs where required for product functionality. Our long-term data asset is the proprietary capability dataset generated within the platform. |
We will improve model performance through continuous data collection, score calibration, evaluator feedback loops, enterprise hiring feedback, error analysis, and iterative refinement of the ranking logic. As more users and hiring outcomes flow through the platform, the system becomes more accurate, more benchmarked, and more useful for both professionals and enterprises. |
We are starting with India as our primary market, where the need for AI talent validation is immediate and large, while designing the platform architecture and category positioning to expand into global markets over time. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/LiCZTNWcV9NB-ACC-2_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/LiCZTNWcV9NB-ACC-2_compressed.pdf |
https://drive.google.com/file/d/1RDtpDcAi_IkwvAqeJwDqBmWoablH453Y/view?usp=drive_link |
Yes. ACC is mission-driven because we are addressing a structural gap in how AI talent is identified and trusted in India. Today, many capable people are filtered out by resume-led hiring, weak credential signals, and lack of measurable validation, while enterprises struggle to find deployment-ready talent. Our goal is to build a more merit-based and transparent AI talent infrastructure where professionals are evaluated on real capability and enterprises can hire with greater confidence. Over time, we want to improve access to opportunity for high-potential talent beyond traditional pedigree filters and help strengthen India’s AI workforce readiness at scale. |
no |
Reva Nest - Reva university |
checked |
| Mar 24, 2026 @ 2:36 PM |
Geethapriyan E |
geethapriyan@shoptainment.in |
http://www.linkedin.com/in/geethapriyan-e-294a69a9 |
http://N/A |
9738232730 |
Geethapriyan E is the Founder of Shoptainment, building a plug-and-play layer that turns content into instant shopping experiences. With 12 years in video production and prior experience building a marketplace for unorganized food vendors, he brings strong insight into user behavior and content-driven commerce. |
I'm a solo founder I've been researching on this idea since November 2023. And now with a small team started building this. A full time backend developer who joined us a week back, an AI engineer who joined us in the month of December & a part time android developer who started working since January. |
1 |
Our biggest strength is domain insight. We’ve spent years in content creation and have firsthand understanding of user behavior, which helps us build a seamless bridge between content and commerce. |
Shoptainment pvt ltd |
NA (in progress) |
Bangalore |
Plug & play commerce layer. |
Removing the friction that exists between marketing to sales. |
Shoptainment, a plug & play commerce layer that allows brands to sell their products at the point of attention/interaction. |
Our product is unique because it layers on top of existing platforms without requiring heavy infrastructure, turning existing user attention and interactions into instant commerce opportunities. |
MVP |
Pilots |
Our target customers are content platforms, publishers, and creators who want to monetize their content beyond ads, as well as brands looking to convert viewer attention into direct sales. |
TAM $5.8T |
SAM $500B |
SOM $1.76M |
Commission on every sales. |
We compete with affiliate platforms, link aggregation tools, and platform-native shopping features. Our differentiation is that we act as a plug-and-play commerce layer, enabling instant transactions within content rather than redirecting users elsewhere. |
We acquire customers by collaborating with OTT platforms and creators through pilot integrations, demonstrating immediate monetization value, and expanding through partnerships and network effects. |
We begin with targeted pilot integrations with OTT platforms and creators to prove monetization value. Post-validation, we scale through platform-level partnerships and creator networks, using demonstrated revenue uplift to drive adoption. |
Our long-term vision is to build a universal commerce layer where everything users see across content can be instantly shopped, making discovery and purchase seamless. |
Incorporated on 14th January 2025 |
NA |
Yes |
We’re applying to Avinya to accelerate our growth by refining our go-to-market, validating our product with the right partners, and leveraging their network to unlock key platform collaborations. |
During the programme, we aim to run pilots with OTT platforms and creators, generate measurable revenue outcomes, and convert those into scalable partnerships to establish early market traction. |
Yes |
Yes |
Core engine |
Our AI architecture uses computer vision models to analyze video frames and detect products, followed by a matching layer that maps these detections to a product catalog. A backend pipeline handles processing, indexing, and retrieval, enabling real-time or near real-time commerce interactions. The system improves over time through data feedback and optimization. |
Our proprietary data advantage lies in capturing granular, in-content user behavior linking what users watch to what they engage with and buy. As we scale, this builds a unique dataset around content-driven commerce that is difficult to replicate. |
Our defensibility comes from proprietary data and tight integration. As we capture real user interactions within content, our models continuously improve, creating a feedback loop that competitors without similar access cannot easily replicate. |
We evaluate our models using precision, recall, and F1-score for detection and mapping accuracy. In production, we monitor latency, error rates, and system reliability. Most importantly, we measure real-world impact through CTR, conversion rates, and engagement to ensure the models drive actual commerce outcomes. |
We handle data privacy through minimal and consent-based data collection. Our system uses encryption in transit and at rest, along with role-based access control and secure cloud infrastructure. We follow industry best practices and align with relevant data protection regulations, ensuring user data is handled securely and responsibly. |
Potential regulatory risks include data privacy laws, consent management, and platform policy changes. We mitigate these through a privacy-first architecture, minimal data handling, and by aligning closely with platform guidelines, ensuring flexibility as regulations evolve. |
At 10x scale, bottlenecks may arise in real-time processing, model performance across varied content, and catalog matching at scale. We’re addressing this through scalable backend architecture, model optimization, and robust data pipelines to maintain accuracy and reliability as we grow. |
Yes, we have a lean in-house AI/ML capability with expertise in computer vision for product detection in video content and mapping to structured commerce catalogs. We focus on building scalable, real-time pipelines and integrating AI seamlessly into content workflows to enable reliable, production-ready solutions. |
Our current datasets include publicly available data for model training and structured product catalogs for mapping. As we onboard partners, we are generating proprietary datasets from content and user interactions. We ensure all data is either licensed, publicly available, or used with proper authorization. |
We improve model performance through closed feedback loops from real user behavior, continuous retraining with expanding datasets, and iterative optimization of accuracy and system performance. As more content and interactions flow through the system, the models become increasingly accurate and robust. |
We are building for both India and global markets starting with India and scaling globally, given the universal nature of content consumption and commerce. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/aw6xOMpU3bsK-Pitchdeck_shoptainment_In-2.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/aw6xOMpU3bsK-Pitchdeck_shoptainment_In-2.pdf |
https://drive.google.com/file/d/14UqmNXhPmO3k5wMeoF3bZMQKO1cxs9tG/view?usp=share_link |
We are impact-focused in improving how value is created and captured in the digital ecosystem. By bridging the gap between content and commerce, we empower creators with new revenue streams and help businesses convert attention into meaningful outcomes. |
NA |
Sanjay |
checked |
| Mar 24, 2026 @ 2:22 PM |
SANJEEV MARUTI POL |
sanjeev.mtech@gmail.com |
https://www.linkedin.com/in/sanjeev-pol-23809218/ |
https://www.bridgemetechnologies.com/ |
9448437143 |
I am the Founder and [Your Designation – e.g., CEO/CTO] of Bridgeme Technologies Private Limited, leading product strategy, technology development, and business growth. My background includes experience in enterprise software, ERP systems, and digital transformation solutions, with a strong focus on building scalable platforms for MSMEs and startups. I oversee the development of our Agentic AI-powered ERP ecosystem, driving innovation across HR, CRM, manufacturing, and finance modules. I have been actively involved in product architecture, go-to-market strategy, and ecosystem partnerships, and our work has been recognized through the Government of Karnataka’s ELEVATE Program grant. By qualification, I am BE, MTech and MBA with 24yrs of experience in IT Industry. |
Directors are my blood relations and I know them since childhood. |
All |
Our biggest strength is our ability to combine deep technology expertise with strong domain understanding of MSME operations. We are a cross-functional team of engineers, product builders, and business strategists who have hands-on experience in ERP systems, AI, and enterprise workflows. This enables us to not only build scalable technology but also design solutions that are practical, affordable, and easy to adopt. Additionally, our team demonstrates high execution speed, adaptability, and customer-centric innovation, allowing us to rapidly develop, deploy, and refine our platform based on real-world feedback. |
BRIDGEME TECHNOLOGIES PRIVATE LIMITED |
https://www.bridgemetechnologies.com/ |
BELGAUM AND BANGALORE |
Bridgeme Technologies builds a unified, Agentic AI-powered ERP platform that helps MSMEs automate operations, gain real-time insights, and scale efficiently. |
MSMEs and startups struggle with fragmented software, manual processes, and lack of real-time insights, leading to inefficiencies, compliance risks, and poor decision-making. Existing ERP solutions are either too complex, expensive, or not tailored for their needs, limiting digital adoption and scalability. |
We provide a unified ERP platform integrating HR, CRM, manufacturing, and finance through our products—PulseHR, Klyro, Prodexa, and Finlyr. Powered by Agentic AI, our platform automates workflows, predicts risks, and delivers actionable insights. It is modular, affordable, and deployable on cloud or on-premise, enabling MSMEs to digitize and scale seamlessly. |
Our solution is unique due to its Agentic AI-driven automation, unified multi-ERP architecture, and MSME-first design. Unlike competitors, we offer end-to-end integration across business functions, reducing data silos and complexity. Our defensibility comes from proprietary AI workflows, domain-specific ERP integrations, scalable architecture, and strong alignment with MSME and government ecosystems, creating high switching costs and long-term customer retention. |
MVP |
Pilots |
Our target customers are MSMEs, startups, and mid-sized enterprises (10–500 employees) across manufacturing, services, logistics, retail, and B2B sectors. We focus on businesses in Tier-2 and Tier-3 cities seeking affordable, scalable ERP solutions for digital transformation, compliance, and operational efficiency. |
₹20,000+ crore (India enterprise software market including ERP, CRM, HR, finance) |
₹5,000–7,000 crore (MSME-focused ERP and SaaS adoption segment) |
₹500–1,500 crore (initial reachable market via MSMEs, startups, and clusters in India) |
We follow a subscription-based SaaS model with per-user pricing across modules and bundled ERP offerings. Additional revenue comes from on-premise licensing, implementation, customization, training, AMC, and AI-based add-ons, ensuring recurring and scalable revenue streams. |
Our competitors include global players like SAP Business One, Oracle NetSuite, Microsoft Dynamics 365, Odoo, Zoho One, and Indian solutions such as Tally, Marg ERP, Busy, Focus Softnet. However, they lack unified AI-driven automation tailored for MSMEs. |
We acquire customers through a mix of direct sales, digital marketing, partner networks, and referrals. We also leverage government programs, incubators, and MSME clusters for pilot deployments and bulk onboarding. |
Our GTM strategy is partner-led and cluster-focused, targeting MSMEs through channel partners, system integrators, and industry associations. We combine SaaS onboarding, bundled ERP offerings, and government collaborations to scale efficiently across regions and sectors. |
Our vision is to become a global leader in AI-powered ERP platforms for MSMEs, enabling millions of businesses to operate on a unified, intelligent system that drives automation, productivity, and data-driven growth. |
Incorporated on 17th August, 2023 as Private Limited. |
NA |
Yes |
We are applying to Avinya to accelerate the growth of our Agentic AI-powered ERP platform through access to a strong AI-focused ecosystem, mentorship, and enterprise networks. Avinya’s focus on real-world AI applications aligns closely with our mission to transform MSMEs through intelligent automation. We see this as an opportunity to refine our AI capabilities, validate use cases at scale, and strengthen our positioning as a deeptech platform. |
During the programme, we aim to enhance our Agentic AI capabilities, onboard strategic pilot customers, and strengthen our go-to-market strategy. We want to build industry-specific AI use cases, improve product scalability, and establish partnerships with enterprises and government bodies. Additionally, we aim to achieve strong customer traction, revenue growth, and readiness for larger funding rounds. |
Yes, we are fully open to hybrid and virtual participation. Our team is structured to collaborate effectively across remote and on-site environments, ensuring active engagement, timely execution, and maximum value from the programme. |
Yes |
Core engine |
AI is embedded across all ERP modules using an Agentic AI architecture. We use a combination of ML models, rule engines, and AI agents that automate workflows, generate insights, and recommend actions. The system integrates data pipelines, event-driven processing, and analytics layers to enable real-time predictions (demand, cash flow, workforce) and anomaly detection across HR, CRM, finance, and manufacturing. |
We build proprietary datasets from multi-module ERP interactions—including HR, sales, finance, and production workflows. This cross-functional business data creates a unique dataset for training domain-specific models, improving prediction accuracy and personalization over time. |
1. Integrated ERP data layer across domains
2. Agentic AI workflows embedded into operations
3. Domain-specific models for MSMEs
4. High switching costs due to deep system integration and data accumulation |
We use metrics such as:
1. Prediction accuracy (MAPE, RMSE)
2. Precision/recall for anomaly detection
3. Model drift and stability monitoring
4. System latency and response time
5. Business KPIs (cost reduction, efficiency gains) |
We follow GDPR-aligned practices, role-based access control, encryption (data at rest & in transit), audit logs, and on-premise deployment options for sensitive customers. We ensure compliance with Indian regulations (GST, payroll, etc.). |
Minimal. Risks relate to data protection laws (GDPR, DPDP Act India) and industry-specific compliance, which we mitigate through secure architecture and configurable data controls. |
Potential challenges include data pipeline scalability, model latency, and infrastructure load. We are addressing this through cloud-native architecture, microservices, and scalable AI pipelines. |
Yes. Our team includes AI engineers, data engineers, and full-stack developers with experience in machine learning, data pipelines, and enterprise systems, focused on building scalable AI-driven ERP solutions. |
We use a mix of:
1. Customer ERP data (owned, anonymized)
2. Synthetic datasets for training
3. Public datasets where applicable
4. All data usage is compliant and secured. |
Through feedback loops, continuous learning pipelines, model retraining, and usage-based data insights, ensuring models improve with scale and real-world usage. |
We are building for India first, focusing on MSMEs, with a roadmap to expand into global markets (EU, Middle East, Southeast Asia) with compliance-ready architecture. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/FY6PSY6Xq85H-Pitchdeck_Bridgeme-Technologies_V1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/FY6PSY6Xq85H-Pitchdeck_Bridgeme-Technologies_V1.pdf |
https://youtu.be/5z3gHVGIKWY |
Yes, we are building a mission-driven and impact-focused startup.
Our mission is to digitally empower MSMEs and startups by providing affordable, intelligent, and easy-to-use ERP solutions that enable them to operate efficiently and scale sustainably. In India, millions of MSMEs still rely on fragmented systems and manual processes, limiting their productivity and growth potential. Through our Agentic AI-powered unified ERP platform, we aim to bridge this gap by bringing automation, real-time insights, and decision intelligence to everyday business operations.
Our impact is centered on improving productivity, reducing operational inefficiencies, enhancing compliance, and enabling data-driven decision-making. By focusing on Tier-2 and Tier-3 markets, we are helping underserved businesses adopt digital technologies and participate in the formal economy. Over time, we aim to contribute to MSME growth, job creation, and economic development, aligning with national initiatives like Digital India and Make in India. |
No, however we strongly believe in building an inclusive and diverse organization. Our focus is on creating equal opportunities, fostering diverse talent, and delivering solutions that empower underserved MSMEs and startups, especially in Tier-2 and Tier-3 markets. |
KITS, KDEM and NASSCOM |
checked |
| Mar 24, 2026 @ 1:48 PM |
Swapna Dasoju |
swapna.dasoju@aventiqinnovations.com |
https://www.linkedin.com/company/aventiq-innovations-pvt-ltd/%20 |
https://aventiqinnovations.com/ |
+91 81785 37398 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 24, 2026 @ 12:23 PM |
Akshay Hegde |
akshay@anomapro.com |
http://linkedin.com/in/akshayshridhar |
http://www.anomapro.com |
09483280080 |
CEO with Product Management Background and CTO With Data Science Background |
I and arun were in same college in 11th standard and we were friends since then and currently working together on product snce last one year |
1 |
Akshay's biggest strength is founder-market fit combining 4+ years of hands-on fintech product experience (lending, payments, NBFC partnerships) with a practical understanding of regulation and compliance. He's also a proven 0 to 1 builder who has worked directly with gig economy operators (vehicle fleet operators), giving him ground-level insight into the exact users Finndot is targeting.
Arun's biggest strength is his deep technical expertise in credit risk - 6+ years building lender-grade credit decision systems with real, measurable impact on NPAs and collections. Crucially, he specializes in explainable and causal AI modeling, which is essential for underwriting thin-file users and getting regulatory approval - a critical differentiator for Finndot's core proposition. |
Anomapro Pvt Ltd |
www.anomapro.com |
Bangalore |
Empowering the next billion with smarter finance |
Personal |
Finndot is India's first AI-powered financial health assistant built for the underserved. It reads users' financial SMS and Account Aggregator data - with consent - to build a real-time profile capturing income, EMIs, spending patterns, and savings behavior. Unlike CIBIL, which relies solely on credit history, Finndot constructs a richer picture of financial health. It then recommends smarter loans, tailored investments, and sends savings nudges and portfolio risk alerts. The AI model curates personalized financial plans based on each user's actual context. In short, Finndot converts raw financial signals into actionable financial decisions - empowering gig workers, freelancers, and first-jobbers who the system currently ignores. |
The core innovation is being Bharat's first fin-health assistant that reads financial SMS and Account Aggregator (AA) data - with user consent - to build a real-time financial profile. Unlike CIBIL, which only looks at credit history, Finndot captures income flow, spending discipline, and digital trust markers, making it possible to serve the 160M+ Indians who are invisible to legacy institutions. What Makes It Defensible (The Moat) Proprietary Data Network Effects - Every new user adds training data to their AI models. More users means better predictions, which attracts more users. This flywheel is very hard for competitors to replicate once it gets going. Regulatory Compliance as a Barrier - RBI-compliant Account Aggregator integration, DPDP compliance, and explainable AI models for regulatory approval aren't easy to build. This creates a structural barrier that slows down copycats. Strategic Partnerships - Already secured channel partnerships with leading NBFCs (Abhi Loans, Branch, Suryoday, Unity, IndusInd) and planning direct gig platform integrations. These relationships take time and trust to build. |
Users |
Pilots |
gig workers, first-jobbers, and micro-entrepreneurs |
30000 Crore |
12000 Crore |
500 Crore |
Referral commission and Subscription |
Axio,MoneyView,ET Money |
Digital Marketing Along with offline partners who are already selling finance products |
Gig Platform Partnerships & Digital Marketing (Primary) |
To be the
Financial Health Layer for Bharat.
Start with SMS/AA
Evolve into a Trust Layer that
Powers Access to Money for
The Next Billion. |
Incorporated |
0 |
Yes |
Pre-revenue stage, needs structured support to move from MVP to production Regulatory complexity - RBI, DPDP, AA compliance requires expert mentorship Incubator credibility builds user and NBFC partner trust Network access to lenders, regulators, and co-investors Small two-founder team needs mentorship in GTM, hiring, and finance Capital efficiency - grants and cloud credits Fintech advisor guidance on underwriting and credit model validation Faster partnership closures with gig platforms and NBFCs Access to pilot programs and sandbox environments for regulatory testing |
Move MVP to production-grade app Build robust API/SDK infrastructure for B2B Improve AI model accuracy with more user data Scale from 1,000 to 40,000+ monthly active users Hit 15-20% paid conversion rate Achieve break-even at 18 months |
Yes |
Yes |
Core engine |
|
Yes |
Data Network Effects |
|
|
|
|
Yes |
|
|
Building for india |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/teQmEt1ROwip-Finndot_Investor_Deck_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/teQmEt1ROwip-Finndot_Investor_Deck_compressed.pdf |
https://www.youtube.com/watch?v=nZyeJ0zR7qA |
Yes |
Yes Financial Inclusion |
KDEM |
checked |
| Mar 24, 2026 @ 12:09 PM |
Harish Jain |
harish.jain@cropsecure.org |
https://linkedin.com/in/harish-jain-3220871ab |
http://NA |
9738108108 |
Harish Jain is the Founder and Managing Director of CropSecure Agri Tech Private Limited. He established the company in 2021 with the aim of using technology to improve agriculture and support rural farmers. |
We met through our shared interest in agricultural innovation and rural development. Since then, we have been working together for the past 2-3 years at CropSecure Agri Tech Pvt Ltd, collaborating on developing technology-driven solutions to support farmers and improve farm productivity. |
2 |
Our biggest strength as a team is our shared vision and complementary skills. We combine expertise in agriculture, technology, and field-level farmer engagement, which helps us design practical solutions that truly address farmers’ problems. Our strong coordination and commitment to improving farmers’ livelihoods enable us to effectively develop and implement innovations through CropSecure Agri Tech Pvt Ltd. |
M/s Cropsecure Agri Tech Pvt Ltd |
https://www.cropsecure.org/ |
Huligi, Koppal |
To connect the rural farmers into digital platform |
We are solving the problem of labour shortage and time-sensitive harvesting in jasmine farming. Jasmine flowers must be harvested early in the morning and very gently to maintain quality, but farmers face increasing difficulty in finding skilled labour. Manual harvesting is also time-consuming, expensive, and physically demanding. Our jasmine harvesting robot aims to automate the harvesting process, reduce labour dependency, ensure timely picking, and improve efficiency and income for jasmine farmers. |
Our solution is an AI-enabled Jasmine Harvesting Robot designed to automate the delicate process of picking jasmine flowers. The robot uses computer vision to identify mature jasmine buds and a gentle robotic gripper to harvest them without damaging the flowers. It operates efficiently during early morning hours when harvesting is required, reducing dependence on manual labour. The system helps farmers save time, lower harvesting costs, and ensure timely and consistent flower collection, ultimately improving productivity and profitability in jasmine cultivation. |
Our solution is unique because it combines AI-based flower detection with a gentle robotic harvesting mechanism designed specifically for delicate jasmine buds. The robot can identify mature flowers and harvest them carefully without damaging the plant, ensuring quality and efficiency. It also addresses the time-sensitive nature of jasmine harvesting and labour shortages, making it a practical and innovative solution for jasmine farmers. |
MVP |
Testimonials |
Small and Medium Size Farmers |
Karnataka and neighbour states |
Karnataka |
Koppal and neighbouring districts |
Our revenue model is based on robot-as-a-service and equipment sales. Farmers can either pay per acre/per harvesting service using the jasmine harvesting robot or purchase the robot directly. Additionally, we may generate revenue through maintenance services, technology support, and partnerships with farmer groups or FPOs. |
Our main competitors are manual labour-based harvesting and emerging agricultural robotics companies developing automated harvesting solutions. Currently, most jasmine harvesting is still done manually, but some agri-robotics and automation startups are working on robotic harvesting technologies for delicate crops. However, very few solutions are specifically designed for jasmine flower harvesting, which gives us a strong opportunity in this niche market. |
We acquire customers through direct engagement with farmers, farmer producer organizations (FPOs), and local agricultural networks. We also demonstrate the jasmine harvesting robot through field trials, farmer meetings, and agricultural events, allowing farmers to see the benefits of the technology and adopt the solution. |
Our go-to-market strategy is to first launch pilot projects with jasmine farmers in key cultivation regions and demonstrate the robot’s effectiveness through field trials. We will partner with farmer groups, FPOs, and agricultural organizations to build trust and awareness. Based on successful demonstrations, we will scale adoption through service-based harvesting models and direct sales of the robot. |
Our long-term vision is to develop advanced agricultural robotics that solve labour shortages and improve farm efficiency. Starting with the jasmine harvesting robot, we aim to expand our technology to support automated harvesting and smart farming solutions for multiple crops, helping farmers increase productivity, reduce costs, and adopt modern agriculture. |
CropSecure Agri Tech Pvt Ltd is a registered private limited company in India, currently operating and developing agri-technology solutions for farmers. |
NA |
No |
We are applying to Avinya to gain mentorship, financial support, technical guidance, and incubation support to develop and scale our jasmine harvesting robot. The program will help us refine our technology, validate the product in real farming conditions, and connect with industry experts, investors, and partners to accelerate the growth of our agri-tech solution. |
During the programme, we aim to develop and refine the prototype of our jasmine harvesting robot, conduct field trials with farmers, and validate the technology. We also want to receive mentorship on product development, business strategy, and market expansion to prepare the solution for large-scale deployment. |
Virtual |
Yes |
Supporting feature |
Our AI architecture is based on computer vision and machine learning. Cameras capture images of jasmine plants, and an AI model processes these images to detect and identify mature jasmine buds ready for harvesting. The detection results are sent to the robot’s control system, which guides a gentle robotic gripper to pick the flowers without damage. The system continuously improves through collected field data, making the detection and harvesting process more accurate over time. |
We are building a proprietary dataset of jasmine flower images and harvesting patterns collected directly from farms. This dataset helps train our AI model to accurately detect mature jasmine buds under different lighting and field conditions. Over time, the continuous collection of real farm data will improve the performance of our harvesting system and create a strong data advantage. |
Our AI component is defensible because it is trained on proprietary field data collected from jasmine farms, including images of buds at different growth stages and environmental conditions. This crop-specific dataset helps our AI accurately detect mature jasmine buds and guide precise harvesting. Over time, continuous data collection and model improvement will make the system more accurate and difficult for competitors to replicate. |
We evaluate our AI model using accuracy, precision, and recall metrics to measure how well it detects mature jasmine buds. We also test the system in real farm conditions to check harvesting success rate, detection accuracy, and flower damage rate. Continuous field testing and farmer feedback help us improve the reliability and performance of the model over time. |
We ensure data privacy and security by collecting only necessary farm and image data and storing it in secure systems with controlled access. Farmers’ data is used only for improving the AI model and service performance. We follow standard data protection practices and comply with applicable data privacy regulations to ensure that farmer information remains safe and confidential |
Yes, there may be some regulatory risks related to agricultural machinery standards, safety compliance, and AI data usage. However, we plan to ensure that our jasmine harvesting robot complies with relevant government guidelines, safety regulations, and agricultural equipment standards. We will also follow proper data protection practices to minimize regulatory risks. |
If we scale 10x, the main challenges could be handling large volumes of field data, maintaining model accuracy across different farm environments, and ensuring reliable hardware–software integration. Variations in lighting, plant density, and farm conditions may affect detection accuracy. To address this, we plan to continuously retrain the AI model with more diverse data and strengthen our cloud and edge computing infrastructure for better scalability. |
Yes, we are building an in-house AI/ML capability focused on computer vision and agricultural robotics. Our team works on developing AI models for jasmine bud detection, image data processing, and integrating the models with robotic harvesting systems. The expertise includes machine learning, computer vision, and field data collection for training crop-specific AI models. |
We are currently using a combination of publicly available image datasets and proprietary datasets collected from jasmine farms. The farm images are collected directly during field visits and are owned by us, which helps train our AI model for accurate jasmine bud detection under real farming conditions. |
We plan to continuously improve model performance by collecting more real-time farm data and retraining the AI model regularly. Feedback from field trials, new images of jasmine buds under different conditions, and performance monitoring will help refine the model. This iterative learning process will increase detection accuracy and reliability over time. |
We are initially building for the Indian market, where jasmine cultivation is widespread and labour shortages are a major challenge. In the long term, we aim to expand to global markets where jasmine and similar delicate flowers are cultivated, adapting our technology to different crops and farming conditions. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/kuphB6XgPzkK-Jasmine-PPT-.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/kuphB6XgPzkK-Jasmine-PPT-.pdf |
https://www.linkedin.com/feed/update/urn:li:activity:7424136944022106113 |
Yes, we are building a mission-driven and impact-focused startup. Our goal is to solve labour shortages in agriculture and support farmers by introducing affordable robotic solutions. Starting with the jasmine harvesting robot, we aim to improve farm efficiency, reduce harvesting costs, and increase farmers’ income while promoting the adoption of modern agricultural technology. |
NA |
Through social media / LinkedIn |
checked |
| Mar 24, 2026 @ 11:25 AM |
Rohit Tuteja, Sachin Kamkar |
rohit@finny.club |
https://www.linkedin.com/in/rohit-tuteja/ |
https://finny.club/ |
+918800601515 |
Rohit: CEO; Sachin: CTO |
We met at Antler in June 2024. We didn’t team up straight away.
When we ended up talking about money stuff...why no one knows when they can actually stop working for money and how much they need. I was already thinking about this idea, and he was super into personal finance, so he basically became my first “interview.”
Both of us spoke to and worked with a bunch of other folks in the cohort. It helped us know what we liked or not in a co-founder. Then Sachin and I did a small work trial to see if we’re compatible. And that one week pretty much settled it. We ideated fast, argued a lot but came out as friends, spoke to users… and none of it felt heavy. It just worked.
By the end of that trial we said, “Let’s do it” and that’s how the team came together. |
two |
As a team, we’ve built and scaled products used by millions — from **1M+ users at HBR** to **3M cards issued at Hyperface**, and helped grow businesses from **$25M to $100M in revenue at upGrad** and turned **Myntra from a warehousing to a marketplace model**.
What stands out is not just scale, but zero-to-one + scale across different environments - consumer, fintech, and platforms and carrying that learning into building Finny from the ground up. |
Finny |
https://finny.club/ |
Bengaluru |
AI-first wealth advisory with a human in the loop fast tracking FIRE for mass affluent Indians |
The problem we’re solving is simple, but rarely answered clearly: “When will I stop working for money?” In India today, most people who’ve reached retirement have solved this either through government pensions or their children supporting them. That path won’t exist for most of us. People know they need a corpus to live independently. They don’t expect to rely on kids. But they don’t know how much is enough, when they’ll get there, or what to do next. And the system isn’t built to help: There are only ~900 registered human advisors for millions who need guidance. So you’re left with: guesswork fragmented advice or products being sold instead of a plan That’s the gap we’re solving. |
An AI-first wealth advisory, with a human in the loop, focused on one question everyone quietly has: “When will I stop working for money?“.
Finny answers that with your FIRE Age using real data fetched in under 60 seconds via Account Aggregator. It’s not a calculator but a complex financial modelling on top of your actual banking, investments, loans, and PF data translating into a net worth and cashflow view.
Post FIRE Age, we go deeper with your Potential FIRE Age, where you could be, if you fixed the wrong steps or made better choices. That’s the bridge into advisory.
In advisory, we build a personalised path to FIRE, work backwards from your corpus, and stay with you as life changes with constant advisory in form of periodical tax efficient rebalancing, behavioural nudges using user data and an annual check-in.
We’re live with ~100 users (building in public) with ~55% D7 retention, launching public beta in early Jan. Net worth + cashflow drive everyday stickiness; FIRE Age is the growth lever that converts to paid advisory. Starting with 50M mass affluents in India, but the problem is global. |
Finny is AI-first. We pull a user’s complete financial picture - assets, liabilities, and cashflows - all in under 60 seconds using open banking account aggregation. Our models continuously compute FIRE age and lifetime outcomes, not one-time plans or product recommendations. AI does the heavy lifting: data ingestion, modelling, and personalised advisory plan. Humans stay in the loop for emotional support during life and market events. This delivers objective, conflict-free financial planning that is always current and scalable, something traditional advisory models cannot offer. |
Users |
one |
Mass affluent Indians earning between 20L to 1cr as household income |
$130bn |
$89bn |
$29bn |
Advisory fee + Referral fee on insurance + home loans |
Friends and Family, Pseudo advisors in MFD and Bank RM |
Through FIRE Age product, Partnerships, and UGC Content Marketing |
Founder Led Branding, Partnerships with brokers and large companies, and performance |
Our ambition is to build the new HDFC of India, and then take it global.
The financial system is built on three anchors: access, trust, and convenience. Banks scaled access, but failed on trust and convenience. We’re starting from the opposite end with trust first.
We could have taken the easier path and started as a distributor or MFD. That model scales quickly because users don’t pay upfront. But it’s also where trust breaks. We deliberately chose to start as a fee-only advisor, even though it’s harder, because trust is the foundation we want to build on.
Finny begins by advising people on their money in a conflict-free way. Once trust is established, we expand vertically: helping users manage their money, keep their money, lend responsibly, and secure their money through insurance. Over time, this means becoming asset managers and insurance companies ourselves and not intermediaries.
In 5–10 years, Finny becomes a full-stack, trust-first financial institution starting in India and expanding globally. |
Incorporated |
3cr |
one |
To get AI-first product thinkers, network for GTM, and capital |
Don't have a clear one line answer but figure things around the above mentioned 3 objectives |
Preferably virtual but open to hybrid. |
one |
one |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 24, 2026 @ 11:18 AM |
Ramalingeswara Rao K V |
kvramalingeswararao@gmail.com |
https://www.linkedin.com/in/RamalingeswaraRaoKV/ |
http://www.RamalingeswaraRaoKV.blogspot.com |
9449501439 |
Founder, NIRAM |
Single Founder |
1 |
Commitment, Dedication, Respect, etc. |
NIRAM |
www.LanKeys.blogspot.com |
Bengaluru |
NIRAM LanKeys (www.LanKeys.blogspot.com) is a patented Multi-purpose Natural Language Processing Tool. |
Cyber Crime issues, Communication in Mother-tongue, etc. |
✳️ Information Technology (IT):
⭐️ NIRAM LanKeys (www.LanKeys.blogspot.com) is a patented Multi-purpose Natural Language Processing Tool.
I have participated in the Launchpad programme at NSRCEL in IIMB.
⭐️ Network (Pitch & Honor): https://LanKeys.blogspot.com/p/network.html
⭐️ Videos:
Demo: https://youtu.be/7TAhbNFFd48
Product Life Cycle: https://youtu.be/qRImjCgOdYo |
Number of Patents: 2 (Published-1 and Granted-1). Patent #1 [Published]: METHOD AND SYSTEM FOR TRANSLITERATING TEXT FROM ONE LANGUAGE TO MULTIPLE LANGUAGE SCRIPTS. Intellectual Property BHARAT/India 🇮🇳 Patent Application No. 931/CHE/2012. World Intellectual Property Organization 🌐 Patent Application No. PCT/IN2013/000155. Patent #2 [Granted]: SYSTEM AND METHOD FOR AUTOMATIC NAMING CONVENTION OF AUDIO FILES BASED ON USER PREFERENCE. Intellectual Property BHARAT/India 🇮🇳 Patent No. 473442. Patent Application No. 2686/CHE/2014. |
MVP |
Users |
B2G, B2B, B2C: Global (Digital) People |
-NA- |
-NA- |
-NA- |
Patent licence; Products & Services. |
-NA- |
Network |
B2G, B2B, B2C |
I want to serve the entire World, with ethics of Life (= Freedom + Peace + Love). |
In process/progress |
Bootstrapped |
No |
Network |
Network |
Preferably In-person/hybrid |
Yes |
Automation layer |
Patent [Granted]: SYSTEM AND METHOD FOR AUTOMATIC NAMING CONVENTION OF AUDIO FILES BASED ON USER PREFERENCE.
Intellectual Property BHARAT/India 🇮🇳 Patent No. 473442. Patent Application No. 2686/CHE/2014. |
Number of Patents: 2 (Published-1 and Granted-1). |
Number of Patents: 2 (Published-1 and Granted-1).
Patent #1 [Published]: METHOD AND SYSTEM FOR TRANSLITERATING TEXT FROM ONE LANGUAGE TO MULTIPLE LANGUAGE SCRIPTS.
Intellectual Property BHARAT/India 🇮🇳 Patent Application No. 931/CHE/2012.
World Intellectual Property Organization 🌐 Patent Application No. PCT/IN2013/000155.
Patent #2 [Granted]: SYSTEM AND METHOD FOR AUTOMATIC NAMING CONVENTION OF AUDIO FILES BASED ON USER PREFERENCE.
Intellectual Property BHARAT/India 🇮🇳 Patent No. 473442. Patent Application No. 2686/CHE/2014. |
|
|
-NA- |
|
|
|
|
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/VF0BaGzIhn1Y-NIRAM_LanKeys_PitchDeckBM_22Mar2026.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/VF0BaGzIhn1Y-NIRAM_LanKeys_PitchDeckBM_22Mar2026.pdf |
https://youtu.be/7TAhbNFFd48 |
https://youtu.be/qRImjCgOdYo |
IIMB NSRCEL |
KDEM |
checked |
| Mar 22, 2026 @ 8:39 AM |
Ashish Sonkusare and Vidhi Gaur |
ashish@everse.ai |
https://www.linkedin.com/in/ashish-s-50a7631/ |
https://eVerse.AI |
+91 9730740252 |
Our founding trio bridges deep-tech enterprise architecture with grassroots agricultural execution. CEO Ashish Sonkusare brings over 25 years of AI/ML leadership, including a tenure at Amazon Web Services, and chairs the Carbon Market Livestock & Dairy Committee at CMAI. Vidhi Gaur (President - Carbon Projects) leads our GreenCow digital MRV frameworks, ensuring rigorous global carbon compliance and relationships with Verra, SustainCert, and Earthood. |
We have been working together since the last three years and we complement each other really well as Ashish comes from a hardcore technology background and Vidhi comes from hardcore climate background. We need both these skills and experiences to build cutting edge and world class solutions to reduce methane emissions from dairy and livestock sector. |
two |
Our biggest strength as a team is our unique ability to bridge elite, enterprise-grade artificial intelligence with rigorous climate science and grassroots agricultural execution.
While many tech teams struggle to move beyond software, our founding triad perfectly covers the three essential pillars needed to scale a climate-tech startup in the Global South:
Deep-Tech Architecture: Our CEO, Ashish Sonkusare, brings over 25 years of AI and machine learning leadership experience, including a tenure at AWS and a computer science degree from IIT Bombay. He ensures our AGICow, ConnectedCow, and GreenCow platforms are built to highly scalable, enterprise standards.
Climate Science & Global Compliance: Our Director of Carbon, Vidhi Gaur, is a gold-medalist environmental scientist. She ensures our digital MRV frameworks meet the rigorous, registry-grade standards of global verifiers and successfully align with complex methodologies like Verra's VM0041.
Grassroots Operational Scale: Our President of Operations, Shailendra Narwade, brings deep expertise in executing large-scale, on-the-ground deployments. He drives our hardware and feed interventions across thousands of remote, off-grid villages.
Backed by a dedicated 23-member engineering team in Nagpur, our collective strength is the proven ability to seamlessly translate complex deep-tech into massive physical rollouts—such as our 20-million cattle deployment in Maharashtra—while maintaining strict financial discipline as a completely bootstrapped company |
eVerse.AI |
https://eVerse.AI |
Nagpur, Maharashtra |
We are a biotech and technology company for the livestock methane reduction space |
We are reducing methane emissions from livestock and dairy animals and generating carbon credits and we pass that income back to small holder dairy farmers in India. |
eVerse.AI addresses livestock enteric fermentation through a dual biological strategy: scaling proven feed additives today while providing the foundational infrastructure for tomorrow's methane reduction vaccines.
Feed Additives (Current Deployment):
Our immediate expertise is demonstrated by our "stacked" nutritional intervention. We utilize Harith Dhara—an anti-methanogenic additive licensed from ICAR-NIANP—combined with the National Dairy Development Board's Ration Balance Program. Together, this protocol physically reduces enteric methane emissions by 34% and increases milk yields by 16-18%. We are actively deploying this at an unprecedented scale across 23 million animals through our partnerships with the Maharashtra State Government and Banas Dairy.
Methane Reduction Vaccine (Development & Validation):
While feed additives are highly effective, they require daily administration. A methane reduction vaccine represents a critical low-frequency, highly scalable alternative that stimulates salivary antibodies to bind to and neutralize methanogenic microbes in the rumen.
Our Unique Expertise and Positioning:
The primary bottleneck in developing both novel feed additives and vaccines is the lack of infrastructure for large-scale, scientifically rigorous field validation. eVerse.AI solves this critical gap. We possess proprietary, patent-published portable methane measurement devices calibrated against global standards, alongside the AI-driven "GreenCow" digital Measurement, Reporting, and Verification (MRV) platform. Because our technology accurately quantifies emissions in off-grid environments and aligns with the stringent Verra VM0041 carbon crediting methodology, we provide biotech developers and global academic partners with a ready-made, highly auditable 23-million-animal testbed to rapidly validate, deploy, and monetize new biological methane reduction technologies. |
Our solution is highly unique and defensible due to a robust, four-layered moat encompassing deep-tech intellectual property, exclusive biological licenses, massive distribution channels, and a self-sustaining economic model: 1. Technological IP Moat: We do not rely on off-the-shelf hardware. We have filed five patent applications and developed proprietary, patent-published handheld methane measurement sensors that are calibrated against global standards like GreenFeed and Respiration Chambers. This hardware feeds directly into our GreenCow digital MRV platform, creating a closed-loop data architecture specifically aligned with the rigorous Verra VM0041 carbon crediting methodology. 2. Biological Exclusivity: We hold the commercialization license from the apex government research body, ICAR-NIANP, for the "Harith Dhara" anti-methanogenic feed additive. By scientifically stacking this with the National Dairy Development Board's Ration Balance Program, we have a proven, highly defensible protocol that physically reduces enteric methane by 34% while increasing milk yields by 16-18%. 3. Unprecedented Distribution Moat: Western competitors struggle in the Global South because of fragmented, high-cost, direct-to-farmer acquisition. We bypassed this by utilizing a B2B2C model. We have locked in long-term MOUs with the Maharashtra State Government (20 million cattle) and Banas Dairy (3 million cattle). This grants us immediate, frictionless, and highly defensible access to a pre-aggregated pipeline of 23 million animals. 4. Farmer-First Economics: Ultimately, our greatest defense is our economic model. We equitably return 60% of the generated voluntary carbon credit revenue directly to the smallholder farmers. Combined with the increased milk yields, our ecosystem injects an additional ₹6,200 ($85) in net-new annual income per cow. By transforming environmental mitigation into a highly lucrative financial dividend for the farmer, we ensure rapid grassroots adoption and create a barrier to entry that competitors cannot replicate without sacrificing their own margins. |
two |
Pilots |
We operate on a highly scalable B2B2C (Business-to-Business-to-Consumer) model, deliberately bypassing the prohibitive customer acquisition costs associated with targeting fragmented rural populations directly. Our primary target customers are massive, apex agricultural institutions that provide frictionless access to millions of pre-aggregated dairy animals. Our core customer segments include: State Governments: We partner with state bodies to drive population-scale climate action, such as our 20-year MOU with the Maharashtra State Government (covering 20 million cattle) and our pipeline with the Madhya Pradesh State Government (targeting 24 million cattle). Apex Dairy Cooperatives: We target the world's largest cooperative networks, exemplified by our active partnership with Banas Dairy (Asia's largest cooperative), which grants us immediate deployment access to 3 million cattle. Large Private Dairies: We work with major private milk procurement companies, such as Country Delight, representing pipelines of over 1 million animals. Additionally, on the monetization side of our business, our secondary target customers are global corporate ESG buyers and climate funds seeking to purchase high-quality, Verra-certified (VM0041) voluntary carbon credits generated by our GreenCow platform. While the marginalized smallholder dairy farmer is our ultimate end-user and primary beneficiary, these enterprise and government clients are our target customers to achieve rapid, global scale. |
USD 20 billion |
$10 Billion |
$3 billlion |
Carbon Credits |
None |
Word of Mouth and attending events and conferences and partnerships with WorldBank and other climate foundations |
Execute livestock methane reduction projects which are registered with Carbon Registry called as Verra under VM0041 methodology. We earn carbon credits and keep a share of that revenue and transfer the rest to the small holder farmers in India. |
Our long-term vision is to make the global dairy and livestock sector net-zero by 2040 while fundamentally transforming the economic livelihood of smallholder farmers.
While our current "GreenCow" ecosystem successfully reduces methane by 34% using daily feed additives, our ultimate long-term scientific vision is to pioneer, validate, and globally distribute an anti-methanogenic recombinant bovine vaccine.
A methane reduction vaccine represents the ultimate scalable intervention for climate-smart agriculture. Because it requires only low-frequency administration (e.g., annual boosters) rather than daily feeding, it overcomes the logistical barriers of treating the millions of free-roaming and grazing cattle across India, Africa, and Latin America. The vaccine will stimulate the animal's immune system to produce antibodies in the saliva that bind to and neutralize methane-producing microbes in the rumen, drastically cutting emissions without altering the animal's natural digestion. |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 22, 2026 @ 1:28 AM |
Kiran Shankar Chawan |
kiranksc55@gmail.com |
https://www.linkedin.com/in/kiran-chawan-548539172/ |
https://datatukey.com/ |
08660682508 |
Founder and CEO |
we met in hackathon and started our journey. |
2 |
Our5 members management team is uniquely qualified due to deep hands-on expertise in AI, data engineering, and rapid
product execution. The founders have built and shipped complex AI systems through 100+ hackathons and winner
of multiple hackathon and finalist of several hackathons with industry roles, led Datatukey from idea to MVP, and
are supported by mentors with 15+ years of AI/ML experience, including a 5-time Kaggle competition winner. |
DATATUKEY PRIVATE LIMITED |
https://datatukey.com/ |
Bengaluru |
Datatukey A Unified AI Platform for Analytics, Insights, and Data Labeling. |
We solve the problem of complex and inaccessible data analysis for businesses and non-technical users. With Datatukey, we enable startups, SMEs, and enterprises to easily understand data through AI-driven insights, interactive visuals, and natural language queries, helping them make faster, smarter, data-driven decisions. |
Our technology is unique in combining AI analytics, natural language interaction, and automated insights into a unified platform. We also integrate AI-powered data labeling for images, videos, and files. Our defensibility comes from proprietary workflows, continuous learning from user interactions, and a scalable architecture designed for non-technical accessibility |
Our technology is unique in combining AI analytics, natural language interaction, and automated insights into a unified platform. We also integrate AI-powered data labeling for images, videos, and files. Our defensibility comes from proprietary workflows, continuous learning from user interactions, and a scalable architecture designed for non-technical accessibility |
Users |
Pilots |
B2B |
50 |
100 |
300 |
Our target customers are startups, SMEs, and enterprises handling large data. They want Datatukey to simplify complex analytics, gain instant AI-driven insights, and leverage data labeling for faster model training, enabling smarter decisions without needing deep technical expertise. |
Databrick, ScaleAI |
initially free later will be on Acquire customers through hackathons, partnerships, outbound sales, content marketing, referrals, and startup ecosystems |
Product-led growth, startup partnerships, outbound sales, content marketing, and targeted enterprise onboarding strategy. |
Short-term: secure a seed funding round and refine product-market fit.
Mid-term: scale Datatukey across global markets, onboarding 1,000+ businesses.
Long-term: build a global AI data intelligence ecosystem, empowering billions to make smarter decisions and becoming foundational infrastructure beyond just a billion-dollar valuation |
Yes |
NA |
Yes |
To get initial funding and GTM support |
funding and GTM support |
ope for both |
Yes |
Core engine |
Modular LLM-driven architecture with data pipelines, auto-labeling models, real-time analytics, scalable cloud infrastructure. |
Proprietary labeled datasets, user interaction feedback loops, domain-specific data, continuously improving AI model performance. |
Continuous learning loops, proprietary workflows, domain data, integrated labeling, unified platform, high switching costs. |
Continuous learning loops, proprietary workflows, domain data, integrated labeling, unified platform, and high switching costs make our AI component defensible against competitors. |
We handle data privacy, compliance, and security through end-to-end encryption, strict access controls, and secure cloud infrastructure. We follow global standards like GDPR principles, implement data anonymization, and ensure user data isolation. Regular audits, monitoring, and compliance practices help maintain trust and protect sensitive information. |
Data privacy laws, AI regulations, cross-border data restrictions, compliance requirements, and evolving government policies could impact operations. |
Data privacy laws, AI regulations, cross-border restrictions, compliance requirements, evolving policies impact operations. |
Yes
Our 5 members management team is uniquely qualified due to deep hands-on expertise in AI, data engineering, and rapid product execution. The founders have built and shipped complex AI systems through 100+ hackathons and winner of multiple hackathon and finalist of several hackathons with industry roles, led Datatukey from idea to MVP, and
are supported by mentors with 15+ years of AI/ML experience, including a 5-time Kaggle competition winner. |
publically available |
by taking feedback on feature working from clients |
for Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/H5yua4cG4jBK-Datatukey-2.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/H5yua4cG4jBK-Datatukey-2.pdf |
http://AI%20Analytics%20Demo:%20https://youtu.be/RX0YR9jwCWg?si=0URv5gIqhezpsVnv%20%20%20%20Data%20labeling%20Automation%20Demo:%20%20https://youtu.be/98EuJXWscxU?si=EdBEpL8WHXl1gpCC |
in future it can be impactful for social driven mission also. |
Yes |
Linkdedin |
checked |
| Mar 21, 2026 @ 10:17 PM |
Hemanth Kumar |
hemanthkumar75691@gmail.com |
https://www.linkedin.com/in/hemanth-kumar-bhimavarapu-a70131271/ |
https://skyflyaviations.in |
7569129168 |
Founder |
Hemanth Kumar – Founder of SkyFly Aviations. Focused on developing AI-enabled aerial emergency response systems. Currently working on early-stage prototype planning, system design, and market validation. Background includes engineering education and hands-on exploration of drone technologies, autonomous systems, and startup development. |
1 |
My key strength is execution-driven thinking and clear problem understanding. As a solo founder, I have taken full responsibility for validating the idea, defining the solution, and planning the prototype. I am proactive in learning, adapting, and connecting with experts to build the right team and execute effectively. |
SkyFly Aviations |
https://skyflyaviations.in |
Vijayawada, Andhra Pradesh |
AI driven eVTOL drone system for rapid emergency medical response in urban areas. |
Emergency response in urban areas is delayed due to traffic congestion and lack of rapid aerial mobility solutions, leading to critical loss of time in life-saving situations. |
SkyFly Aviations is developing an Ai driven eVTOL-based aerial emergency response system designed to bypass ground traffic and deliver rapid medical support. The system focuses on a drone-based prototype capable of autonomous takeoff, landing, navigation, obstacle detection, and return-to-home functionality within a controlled operational range. The long-term vision is to scale this into a full air ambulance platform, but the current focus is validating core flight control, stability, and autonomous navigation systems through a working prototype. |
The solution focuses on building Ai Driven aerial emergency response system optimized for urban environments, integrating AI-driven flight control, real-time navigation, and safety features. Unlike traditional drones or existing air ambulance services, this approach aims to create a scalable, cost-efficient, and rapid-response system specifically designed for emergency use cases, starting with a validated prototype and progressing toward a full aerial mobility platform. |
MVP |
Pilots |
Hospitals, emergency medical service providers, government health departments, and disaster response agencies requiring rapid emergency response solutions. |
$8B+ global emergency medical services and aerial mobility market. |
$1.5B+ urban emergency response and medical logistics market in India. |
.$50M+ initial target market across Tier-1 and Tier-2 cities in India. |
B2B service model through partnerships with hospitals and emergency service providers, including per-use charges, subscription-based emergency response services, and long-term service contracts. |
Traditional air ambulance services, drone logistics companies, and emerging eVTOL startups focusing on urban aerial mobility. |
Direct partnerships with hospitals and emergency service providers, collaboration with government health departments, and pilot programs to validate operational effectiveness. |
Initial deployment through pilot programs with select hospitals, followed by strategic partnerships, regulatory approvals, and gradual expansion across major urban centers. |
To build a scalable AI Driven aerial emergency mobility network enabling rapid, affordable, and efficient life-saving transportation across cities, ultimately evolving into a full-scale air ambulance ecosystem. |
Not yet incorporated (currently in early-stage prototype development) |
Not yet incorporated (currently in early-stage prototype development) |
Yes |
To gain structured mentorship, technical guidance, and ecosystem support required to develop and validate our VTOL prototype through defined development stages. Additionally, we seek funding support and access to industry networks to accelerate prototype testing, achieve technical validation, and move toward real-world deployment of AI - Driven emergency response system. |
Build and validate a working prototype, achieve stable autonomous flight capabilities, develop initial pilot partnerships, and establish a clear roadmap for scaling into real-world emergency response deployments. |
Yes, I am fully committed to participating in all sessions and can attend both virtual and in-person activities as required. |
Yes |
Core engine |
The system integrates AI-driven autonomous flight control using sensor fusion from IMU, GPS, and airspeed data for real-time navigation and stability. Computer vision models are used for obstacle detection and environmental awareness. The architecture includes onboard embedded processing for real-time decision-making, along with system-level self-monitoring that continuously evaluates flight performance, detects anomalies, and enables corrective actions or safe fallback operations such as return-to-home. |
The system will generate proprietary real-world flight data during prototype testing, including navigation behavior, stability metrics, environmental interactions, and system performance logs. This data will be used to continuously improve AI models, enhance reliability, and create a domain-specific dataset that becomes a long-term competitive advantage. |
The AI system is tightly integrated with real-world flight operations, where performance improves through continuous testing, system tuning, and environment-specific learning. The combination of autonomous control, real-time decision-making, and proprietary operational data creates a high barrier to replication without similar hardware-software integration and field data. |
Performance is evaluated using real-world metrics such as flight stability accuracy, navigation precision, obstacle detection reliability, mission success rate, and system response time. Reliability is validated through repeated test flights under controlled conditions, including fail-safe scenarios such as signal loss and system anomaly detection. |
The system follows secure communication protocols and controlled data access practices. Operational data is protected through encryption and secure storage. Compliance with aviation guidelines and data handling standards is considered during system design, especially for mission-critical and safety-related operations. |
The primary risks involve UAV flight regulations, airspace restrictions, and safety compliance requirements defined by DGCA. These risks will be managed by adhering to regulatory guidelines, obtaining necessary permissions for testing, and progressively aligning the system with certification requirements for real-world deployment. |
The primary challenges at 10x scale would be real-time processing load, system latency, and consistency across different operating environments. As deployment increases, maintaining reliable performance across varied conditions and ensuring robust communication and coordination between multiple systems becomes critical. These challenges will be addressed through optimization of onboard processing, efficient data handling, and progressive system testing at scale. |
Currently, the development is founder-led with a focus on system design and prototype development. AI/ML capabilities are being developed through learning, external resources, and planned collaboration with technical experts as the system progresses. |
Currently, the system is in early development and does not rely on large proprietary datasets. Initial development uses simulated environments and publicly available datasets for basic model validation. Proprietary datasets will be generated during real-world prototype testing to improve system performance and reliability. |
Model performance will be improved through iterative testing, real-world data collection, and continuous system tuning. Feedback from flight tests will be used to refine navigation accuracy, stability control, and obstacle detection performance, creating a feedback loop for progressive system improvement. |
The initial focus is on the Indian market, particularly urban and semi-urban environments, where emergency response challenges are significant. In the long term, the solution is designed to be scalable and adaptable for global deployment. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/vaEGPESU63da-SkyFly-Aviations-3.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/vaEGPESU63da-SkyFly-Aviations-3.pdf |
http://NA |
Yes, SkyFly Aviations is a mission-driven startup focused on reducing emergency response time and improving access to life-saving services. By enabling rapid aerial response, the goal is to minimize delays caused by urban traffic and infrastructure limitations, ultimately improving survival outcomes in critical situations. The solution is designed to create scalable, affordable, and efficient emergency response systems that can benefit both urban and underserved regions. |
NA |
NA |
checked |
| Mar 21, 2026 @ 9:42 PM |
SOUMYADEEP SHAW |
shawsoumyashaw@gmail.com |
http://linkedin.com/in/soumyadeep-shaw-226703178 |
http://linkedin.com/in/soumyadeep-shaw |
6284066009 |
FOUNDER |
WE ARE WORKING AS A TEAM OF 10 MEMBERS |
1 |
CYBER SECURITY AND DEFENCE INNOVATION AND R & D TEAM |
OPERATION SINDOOR.AI (MSME) |
linkedin.com/in/soumyadeep-shaw |
DELHI |
CYBER SECURITY AND DEFENCE STARTUP |
IMPERSONIFICATION IN EXAMS AND CONTACTLESS BIOMETRICS |
IMPERSONIFICATION IN EXAMS AND CONTACTLESS BIOMETRICS SYSTEM CAN BE USED NOW A DAYS TO ENBLE A CYBER SAFE BHARAT ID TECH |
IMPERSONIFICATION IN EXAMS AND CONTACTLESS BIOMETRICS SYSTEM CAN BE USED NOW A DAYS TO ENBLE A CYBER SAFE BHARAT ID TECH |
Idea |
Users |
ZOMATO RBI UIDAI AIRPORTS UPSC BLINKIT |
100 |
80 |
78 |
800000 |
IIT BOM IIT MADRAS |
BY LINKEDIN |
PREVENT ZOMATO INSECT_FOOD_FAKE IMAGES SCAM TO CLAIM REFUND |
CONVERT MY STARTUP INTO MSME |
NA |
NA |
Yes |
SEED FUND FOR DELL ALIENWARE , PATENT FILINGS CHARGES |
SEED FUND FOR DELL ALIENWARE , PATENT FILINGS CHARGES , MSME UDYAM REGISTRATIONS |
YES |
Yes |
Core engine |
KERAS CLASSIFIER , MOBILENET FEATURE EXTRACTOR , INFERENCE FUNCTION AND GRADIO |
FILED PATENT |
|
PRECISION RECALL F1 SCORE |
|
|
|
|
|
|
BOTH |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/yd6MZoVaXLsk-OPERATION-SINDOOR-PITCH-DECK-PPT.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/yd6MZoVaXLsk-OPERATION-SINDOOR-PITCH-DECK-PPT.pdf |
https://drive.google.com/file/d/1ELXrBBXIdY4KbaBNK5Xf1iTewNcyifdI/view?usp=sharing |
IT IS MISSION DRIVEN FOR INDIAS MOST AGGRESIVE DEFENCE AND CYBER SECURITY WING |
NA |
DELHI VC BHARAT MANDAPAM |
checked |
| Mar 21, 2026 @ 7:37 PM |
George Johnson and Radha Baran Mohanty |
george.johnson@datamatrixai.co.in |
http://www.linkedin.com/in/georgejohnsonin%20 |
http://www.datamatrixai.co.in |
7829440444 |
Co - Founders |
We met while we were in the GenAI and Prompt Engineering course by Indian Institute of Science(IISc) Bangalore. |
All |
One founder is an IIT Kharagpur alumnus and was with Infosys for 27 years handing North American customers of Communication, Media and Entertainment managing a team of over 200 developers and responsible for delivery.
Other founder has 3 decades of sales, business management and firm management experience as MD and CEO of consulting firm.
Team members are a mix of Ex Infocians with solid development background and post graduates from Indian Mathematical Institute (IMA) with AI/ML.
All the AI solutions we developed came from industry who had real problems in their workflow. |
Datamatrix.AI Pvt Ltd |
https://datamatrixai.co.in |
Bengaluru and Bhubaneshwar |
At DatamatrixAI, we're building next-generation AI infrastructure to help businesses unlock the full potential of AI/ML and GenAI. Our platform leverages cutting-edge technologies to automate complex workflows, enhance decision-making, and reduce operational overhead. We design and develop scalable, and production-ready GenAI solutions with a big eye on security and privacy. |
Organizations today are drowning in documents and dashboards. Key business insights are locked inside fragmented reports, presentations, and siloed data platforms. CXOs and senior leaders face delays, blind spots, and inefficiencies due to: ·Time-consuming manual analysis ·Poor integration across formats and platforms ·Limited access to insights at the point of decision-making |
The AI Platform called EDIS(Enterprise Data Orchestration and Insight System) is an AI-powered Business Intelligence (BI) chatbot platform that extracts strategic and operational insights from unstructured, multimodal business content—including PDFs, PowerPoint presentations, SharePoint files, Confluence pages, and even Tableau/PowerBI dashboards. It empowers CXOs and business leaders with real-time, context-aware answers to complex business questions—without waiting on data teams. |
Our long-term defensibility is built on a combination of proprietary technology, domain-adapted models, and continuously improving enterprise data context. As organizations use the platform, it learns from multiple structured and unstructured data sources, building a unique knowledge layer that becomes increasingly valuable and difficult to replicate. The platform’s architecture integrates multimodal data understanding, contextual reasoning, and conversational analytics in a single system. Over time, this creates strong switching costs as the system becomes deeply embedded in enterprise workflows and knowledge ecosystems. In addition, our focus on solving complex enterprise insight generation, rather than generic analytics, positions the platform as a specialized decision intelligence layer. Continuous model refinement, domain customization, and integration into enterprise systems will further strengthen our competitive moat over time. |
MVP |
Pilots |
Our AI platform is sector and functional agnostic. Wherever there is data and need generation of insights, the platform is a must. Our target segment is B2B customers and ideally Fortune 1000 companies globally. |
USD 15 Billion |
USD 6 Billion |
USD 1.2 Billion |
Licence mode, User Mode and Subscription Mode |
HILA, Bloomberg, Alpha Sense, Seeking Alpha, Factset |
Focused event based engagement like CXO conferences, LinkedIn outreach, References, Social media Vlogs |
Our go-to-market strategy in Europe focuses on a targeted B2B approach through strategic partnerships and industry-specific deployments. We plan to collaborate with consulting firms, system integrators, and data platform providers that already serve enterprise clients and can embed our solution as an insight generation layer within their offerings. Initial focus will be on sectors such as financial services, consulting, and enterprise technology where organizations manage large volumes of structured and unstructured data and require faster decision intelligence. We will complement partnerships with direct outreach to data and strategy leaders, pilot deployments with early adopters, and participation in European AI and data innovation ecosystems. This partner-led and use-case-driven strategy enables faster market entry, lower customer acquisition costs, and scalable growth. |
Be the leader in insight generation market for major industries where the model learns from proprietary data which makes us the Bloomberg of insight generation and enable to company to scale to global repute with ethical values. |
Pvt Ltd |
0 |
Yes |
Primarily for business connect, and funding support to scale. |
Take the product to market. |
Yes |
Yes |
Core engine |
• Layer 1 — Input Layer: Voice/text/image input from MSE via mobile app or web portal → ASR (Whisper) → multilingual NLP (MuRIL) → structured data extraction.
• Layer 2 — Processing Layer: OCR document parsing → AI product categorisation (IndicBERT + XGBoost) → compliance validation engine → SNP matching engine (Sentence-Transformers + scoring matrix).
• Layer 3 — Output Layer: Ranked SNP recommendations → automated form submission → NSIC verification dashboard → ONDC onboarding confirmation.
Data flows through encrypted APIs. The system is cloud-deployable on NIC Cloud / MeitY-approved environments. |
Proprietary data would be from individual companies who would use the platform once its rolled out into the market. |
Our long-term defensibility is built on a combination of proprietary technology, domain-adapted models, and continuously improving enterprise data context. As organizations use the platform, it learns from multiple structured and unstructured data sources, building a unique knowledge layer that becomes increasingly valuable and difficult to replicate.
The platform’s architecture integrates multimodal data understanding, contextual reasoning, and conversational analytics in a single system. Over time, this creates strong switching costs as the system becomes deeply embedded in enterprise workflows and knowledge ecosystems.
In addition, our focus on solving complex enterprise insight generation, rather than generic analytics, positions the platform as a specialized decision intelligence layer. Continuous model refinement, domain customization, and integration into enterprise systems will further strengthen our competitive moat over time. |
Evaluate EDIS using a 4-layer approach: data extraction, retrieval, reasoning, and output quality.
Focus on 5 key metrics—accuracy, business value, grounding (no hallucinations), consistency, and coverage—using a gold dataset of real business cases. Combine human + LLM evaluation, stress-test for reliability, and track decision impact (time saved, better decisions). |
Security Framework
Data Encryption: AES-256 encryption for data at rest, TLS 1.3 for data in transit
Access Control: Multi-factor authentication, role-based permissions
Network Security: VPC, firewall rules, DDoS protection
API Security: Rate limiting, authentication tokens, input validation
Regular Security Audits: Penetration testing, vulnerability assessments
Compliance Standards
SOC 2 Type II: Annual compliance certification
GDPR: European data protection regulation compliance
HIPAA: Healthcare data protection (if applicable)
ISO 27001: Information security management certification
Industry-Specific: Financial (SOX), Government (FedRAMP) as needed
Data Governance
Data retention policies and automated deletion
Audit trails for all data access and modifications
Data anonymization and pseudonymization capabilities
Consent management for personal data processing |
EDIS risks come from data misuse, opaque AI decisions, and unreliable insights impacting business outcomes, so compliance, explainability, and auditability must be built into the core product. EDIS will be regulated primarily across three axes:
- Data (privacy & residency)
- AI (transparency & accountability)
- Decisions (liability & auditability) |
The system is robust to handle huge data for insight generation both at cloud and on premise level. |
One of the Co-founder had done his AI/ML from IIT Kharagpur and has built solutions while in Infosys. And with GenAI from IISc, he is proficient in addressing challenging AI problems which can be scaled seamlessly with privacy, governance and security.
A good number of our full time AI Engineers are from Indian Mathematical Institute(IMA) with PG in AI/ML. |
Datasets are from the company which use the tool as its a GenAI platform. Any customisation is done on the proprietary data of the specific company. |
Continuous model refinement, domain customization, and integration into enterprise systems will further strengthen our competitive moat over time. |
India and Global markets |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/OJwJg0q9ZcpC-EDI-Pitch-Deck-v_1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/OJwJg0q9ZcpC-EDI-Pitch-Deck-v_1.pdf |
https://drive.google.com/file/d/1G4CM4VAt3J78PHW7extdFRJe4ny8PaA4/view?usp=sharing |
We are building a mission driven startup and our mission is to Transform how organizations extract actionable insights from their multi -
modal data ecosystem through AI -powered conversational intelligence. |
Yes |
Did not understand this question hence unable to answer. |
checked |
| Mar 20, 2026 @ 9:26 PM |
Sateesh Ambesange |
sateesh.ambesange@pragyanai.com |
https://www.linkedin.com/in/sateesh-ambesange-3020185/ |
http://www.pragyanai.com |
9741007422 |
CEO, NITK with 25 Years of Industry Experience. Worked at MNCs like Intel, Samsung etc. |
ateesh Ambesange and Rashmi Uppin are husband and wife, united by a shared vision to build impactful AI-driven education solutions through PragyanAI. Both are currently pursuing their PhD in Artificial Intelligence and have over 5+ years of teaching experience. Sateesh brings 25+ years of overall professional experience, combining deep industry exposure with academic expertise.
Dr. Shailesh Harshita Patil has been working with the team for the past 2 years and brings 15+ years of teaching experience. He leads college partnerships and outreach, playing a key role in building and scaling institutional collaborations for AI skill programs.
Harshita Patil, a distant relative with 8+ years of industry experience, is part of PragyanAI as a Product Manager, leading product development and tracking end-to-end solution execution.
Together, the core team has been collaborating for around 2 years, supported by a growing team of employees and external partners who assist in strengthening college connections and ecosystem expansion. |
2 |
Our biggest strength is our deep understanding of both education systems and industry needs, combined with strong execution capability.
Both founders are pursuing PhDs in AI and bring extensive experience in education, along with significant industry and product development experience. This unique combination allows us to build solutions that are not just technically strong, but also aligned with how colleges actually function and how students learn.
We are not building in isolation—we have developed PragyanAI after closely working with colleges and gaining deep insights into the learning ecosystem, including curriculum gaps, student engagement challenges, and employability issues.
Additionally, our upskilling ecosystem acts as a strong advantage:
We have access to trained interns contributing to product development
This enables faster execution at significantly lower cost
It also creates a continuous talent pipeline aligned with our platform
Overall, we combine:
Academic depth (PhD in AI)
Real-world education insights
Industry + product execution experience
Cost-efficient scaling through our own talent ecosystem |
Pragyan SmartAI Technology LLP ( PragyanAI) |
www.pragyanai.com |
Bangalore. Founders from North Karnataka - Bidar |
An Agentic AI platform that grooms, verifies, and that convert education into employability. |
PragyanAI is building an AI-native learning ecosystem that transforms college education into industry-ready skills. |
PragyanAI is an AI-powered platform + 4-year program that transforms college education into industry-ready skills through projects, agentic AI, and real-world experience.
4-Year AI Program (skill-first learning). 10 Skills, 150+ Projects, 2+ Industry Projects + 60+ Tools/Softwares. At Year 4 (Tracks):
Job: placement + interviews
Startup: MVP + demo day
Research: papers + labs
Campus Intelligence Network (clubs, events, competitions)
Agentic AI personalized Learning System
Skill Passport (AI-verified Proof of Competence)
Outcome: Students don’t just learn → they build, prove, and get hired
Platform Modules
Skill Passport: projects, GitHub, certifications → PoC (replaces resume)
Clubs & Events: inter-college ecosystem, guest lectures, leaderboards
AI Olympiad: competitions + talent discovery
Resume + JD Prep: analysis, matching, skill gaps, AI Mock Interviews
Feedback Intelligence: event/class insights + marketing
AI LMS: courses, syllabus mapping, gamified + RAG learning
Evaluation: assignments, exams, AI viva
Code/Project AI: explanation, debugging, guidance
Syllabus + Teaching Planner: AI-generated, industry-aligned
Career Advisor: roadmap + job alignment
Industry Projects: real consulting work
Skill Programs + ISA: Select or Bundle skill Programs and ISA outcome-based education
Admissions: college selection guidance
Education Loans: financial support
AI Core (Differentiator)
Agentic AI: Mentor, Evaluator (AI viva), Matchmaker, Hype, Career agents
RAG Learning: PPT/video/code → AI tutor
Skill DNA: GitHub tracking + engagement + AI validation
Knowledge Graph: Skill → Project → Job mapping
Outcomes
Students: job-ready, strong portfolio, verified skills
Colleges: better placements, modern curriculum
Industry: hire skilled talent, reduce training cost
“AI Skill Development OS that converts education into verified, job-ready talent.” |
PragyanAI is not just a learning platform—it is a full-stack AI Skill Development OS that combines learning, validation, and career outcomes into one system. 1. Skill Passport (Proof of Competence) AI-verified skills via projects, GitHub activity, and AI viva Tracks real engagement, not just completion Replaces resumes with trusted, verifiable skill identity. Hard to replicate without deep data + AI integration 2. Knowledge Graph (Skill → Job Mapping) Connects: Events → Skills → Projects → Job roles Enables: Smart hiring signals Real-time career alignment Creates a proprietary data network effect over time 3. Agentic AI System (Execution, not just content) Mentor, Evaluator, Matchmaker, Career agents Automates: Learning Assessment Career guidance Moves from “AI tools” → “AI workforce” (strong moat) 4. Campus Intelligence Network (CaaS Model) Every college becomes a talent node Clubs, events, competitions → structured via AI Distribution moat + high switching cost 5. End-to-End Platform (Closed Loop) We cover: Learning → Practice → Evaluation → Proof → Hiring Competitors solve parts; We solve the entire lifecycle 6. Continuous Skill Data & Feedback Loop Tracks: Learning behavior Project output Performance Improves: Recommendations Career matching Data flywheel → stronger over time 7. Outcome-Based Model (ISA + Industry Projects) Students work on real problems Pay linked to outcomes Aligns incentives across: Students Colleges Industry |
Revenue |
Revenue |
Engg Colleges - Tier2/3/4 Colleges. Colleges where Faculty Strength is low and Placement is low. Management keen to Impove college quality |
~$3+ Trillion |
~$20–60 Billion |
~$25M – $50M |
Enter as Skilling - offline, BootCamp, Then Start pushing Long Term program and AI Agentic Platforms. |
Skill: NxtWave, Masai School, and Platforms - iDreamCareer,Internshala. PragyanAI combines: Learning (Coursera) Career (LinkedIn) Portfolio (GitHub) Events (Unstop) AI Tutor (ChatGPT) Into one unified, agentic AI-powered platform |
BootCamps, FDP, Then us Tools, Agentic AI platform for conducting Skill programs, then push for use by College |
Start with high-value skill programs → introduce AI tools → expand into full platform adoption 1. Phase 1: Entry via Bootcamps & FDPs (Beachhead) Target: Colleges (students + faculty) Strategy: Conduct: AI Bootcamps (students) FDPs – Faculty Development Programs Goal: Deliver immediate value + visible outcomes Output: Student projects Faculty upskilling Initial trust with institutions 2. Phase 2: Tool Adoption (Land & Expand) Introduce: Skill Passport AI LMS Resume + Career tools Event & Club platform Colleges start using: For courses For events For student tracking Goal: Become part of daily academic workflow 3. Phase 3: Agentic AI Platform (Deep Integration) Deploy: AI Mentor Evaluation system (AI viva) Career advisor Analytics dashboard Integrated into: Classrooms Assignments Internal assessments Goal: Make PragyanAI the core learning infrastructure 4. Phase 4: Campus-Wide Adoption (Scale) Expand to: Entire college All departments Enable: Clubs (CaaS model) Events ecosystem AI Olympiad Outcome: College becomes a PragyanAI Campus Node 5. Phase 5: Network Effects (Expansion) Build: Inter-college competitions National leaderboard Talent discovery network Result: Organic growth via: Student demand College competition (FOMO) Revenue Strategy (Embedded in GTM) Bootcamps / FDPs → Entry revenue SaaS (college license) → Recurring revenue Students (programs / ISA) → Outcome-based revenue Recruiters → Talent access |
Our vision is to build the global infrastructure for skill-based education, where every student’s abilities are continuously developed, verified, and directly connected to real-world opportunities through AI. We aim to transform education from a degree-based system into a skill-first, outcome-driven ecosystem powered by agentic AI. In this future, every student will have a Skill Passport instead of a resume, every college will function as an AI-powered talent hub (Campus Node), and every learning experience will be project-based, personalized, and continuously evaluated. Skills will no longer be assumed—they will be verified through real work and directly mapped to jobs, startups, and research opportunities. PragyanAI will power a global network of colleges connected through a unified platform, supported by a dynamic skill graph that maps students to skills and opportunities, and an intelligent AI system that guides learning, assessment, and career pathways. This will create a trusted talent layer for industry, enabling companies to hire verified, high-quality talent with confidence. Ultimately, we envision a world where students become globally competitive, colleges evolve into innovation hubs, and industry gains access to truly skilled talent—where degrees no longer define ability, but verified skills do.” |
19-02-2021 |
bootstraped |
Yes |
We have build AI Upskilling and Growing Fast, AI MVPs developed and started marketing, Now we need Mentorship to further align our business, scale faster, raise funds |
During the programme, we aim to validate and scale PragyanAI into a category-defining platform for industry-ready education. We plan to onboard 50–100 colleges through bootcamps and faculty development programs, and convert them into active campus nodes using our platform (Skill Passport, AI LMS, and event ecosystem). We will refine our agentic AI system—especially the Skill Passport (Proof of Competence), AI Mentor, and evaluation engine—by deploying them in real academic environments and iterating based on feedback. We also aim to establish strong industry partnerships for real-world projects and hiring pathways, while launching inter-college competitions and leaderboards to drive network effects. Additionally, we intend to raise seed funding to accelerate product development, expand our campus network, and scale operations. By the end of the programme, our goal is to demonstrate measurable impact in student outcomes (projects built, skills verified, placements improved) and position PragyanAI as a scalable AI-powered infrastructure for skill-based education.” |
Anything fine, Hybrid or virtual |
Yes |
Supporting feature |
1. Core AI Layers
LLM Layer (Reasoning Engine)
Models: GPT / Claude / LLaMA
Handles:
Content generation
Q&A
reasoning & explanations
evaluation (AI viva)
RAG Layer (Knowledge Intelligence)
Inputs:
PPTs, PDFs, videos, code
Pipeline:
Ingestion → chunking → embeddings → retrieval → LLM
Enables:
Context-aware AI mentor
Event-specific assistants
Personalized learning
Knowledge Graph (Skill Intelligence Layer)
Built using Neo4j
Maps:
Student → Skills → Projects → Events → Jobs
Enables:
Skill gap analysis
Career recommendations
Recruiter matching
2. Multi-Agent System (Agentic AI)
Core Agents:
Mentor Agent → Teaching, doubt solving
Evaluator Agent → AI viva, assignment validation
Matchmaker Agent → Student ↔ mentor / job matching
Hype Agent → Engagement & recommendations
Career Agent → Skill roadmap & job alignment
These agents collaborate via:
Orchestration frameworks (LangChain / CrewAI / LangGraph)
3. Skill DNA Engine (Validation Layer)
Tracks:
GitHub commits
Project quality
Quiz + viva performance
Engagement behavior
Outputs:
Proof of Competence (PoC score)
4. Content Intelligence Pipeline
Video:
Whisper → transcript → LLM → summaries
Documents:
PDF/PPT → embeddings → RAG
Code:
Static analysis + LLM explanation
5. Learning Analytics Engine
Tracks:
Engagement
performance
skill progression
Feeds:
personalization
recommendations
6. Infrastructure Layer
Backend:
FastAPI (microservices)
Data:
Vector DB → Pinecone / FAISS
Graph DB → Neo4j
SQL → PostgreSQL
Integrations:
GitHub (projects)
Google Meet (sessions)
LMS content
7. Feedback Loop (Self-Improving System)
Continuous loop:
Learn → Practice → Evaluate → Update Skill Graph → Recommend Next |
PragyanAI builds a unique, proprietary dataset centered on real student skills, behavior, and outcomes—something traditional EdTech platforms, resumes, or professional networks do not capture. At its core is a “Skill DNA” layer that generates a Proof of Competence (PoC) by analyzing GitHub activity (code quality, commits, contributions), real-world project outputs, AI-based viva and evaluation scores, assessments, and engagement signals such as interactions with AI mentors. This creates a deep, verifiable skill profile for every student. On top of this, PragyanAI continuously builds a proprietary knowledge graph that maps students to skills, projects, events, and career outcomes, evolving into a powerful talent intelligence layer and predictive hiring dataset. The platform also captures rich learning behavior data—how students learn, their weak areas, improvement curves, and interactions—enabling personalized learning, better recommendations, and outcome prediction. Through its Campus Intelligence Network (CaaS), it aggregates inter-college activity, competitions, and club performance to generate comparative talent insights across institutions. Additionally, agent interactions (Mentor, Evaluator, Career agents) capture student queries, concept gaps, and career intent, forming a longitudinal intelligence layer for each learner. This data advantage is highly defensible as it cannot be derived from resumes, LinkedIn, GitHub, or LMS platforms alone—it requires deep, full-stack integration across learning, projects, evaluation, and career pathways. Over time, this creates a powerful data flywheel where more students generate more projects and interactions, improving the skill graph, enhancing recommendations, driving better outcomes, and accelerating adoption.” |
PragyanAI’s AI component is defensible because it goes beyond generic AI upskilling and builds a deeply integrated, outcome-driven intelligence system. While many platforms offer AI courses or tools, PragyanAI combines AI upskilling with continuous skill validation, real-world project tracking, and career alignment. Its core strength lies in the “Skill DNA” engine, which uses agentic AI to evaluate actual student work—GitHub activity, projects, AI viva, and engagement—creating a verifiable Proof of Competence rather than self-reported skills. This is further strengthened by a proprietary knowledge graph that maps skills to projects and career outcomes, enabling predictive recommendations and hiring signals. Additionally, the platform’s multi-agent AI system (Mentor, Evaluator, Career, Matchmaker) operates across the full lifecycle—learning, assessment, and placement—making it difficult to replicate as it requires deep integration across content, behavior, and outcomes. Combined with a growing campus network and continuous data feedback loops, PragyanAI’s AI becomes smarter and more personalized over time, creating a strong data and system-level moat that competitors focused only on AI upskilling cannot match.” |
We evaluate model performance and reliability across multiple layers—content quality, retrieval accuracy, user outcomes, and system robustness—using a combination of offline benchmarks and real-world feedback loops. At the LLM level, we measure accuracy and correctness through human and rubric-based evaluation, along with relevance, coherence, completeness, and hallucination rates by validating outputs against trusted sources. For our RAG system, we track retrieval Precision@K and Recall@K, context relevance, answer groundedness (ensuring responses are supported by retrieved documents), and latency. In skill evaluation and AI viva, we assess scoring consistency, correlation with human evaluators, and alignment between project quality and assessment scores. At the user level, we focus on outcome-driven metrics such as skill improvement (pre vs post), quiz performance uplift, project completion rates, and engagement signals like interactions and session activity. For agent performance, we monitor task success rates, user satisfaction scores, and fallback or drop-off rates. System reliability is ensured through metrics like latency, uptime (SLA adherence), error rates, and safe fallback accuracy. All of this is continuously improved באמצעות human-in-the-loop evaluation, user feedback, and A/B testing of models and prompts, ensuring the system becomes more accurate, reliable, and effective over time. |
We take data privacy, compliance, and security seriously by designing PragyanAI with a privacy-first, secure-by-design architecture. All user data is encrypted both in transit (TLS) and at rest (AES-256), and access is strictly controlled through role-based access control (RBAC) and authentication mechanisms. We follow data minimization principles—collecting only what is necessary—and ensure clear user consent and transparency in how data is used. Sensitive data such as student profiles, project data, and evaluation results are securely stored and isolated, with audit logs to track access and changes.
From a compliance perspective, we align with global standards such as GDPR and other applicable data protection regulations, ensuring user rights like data access, correction, and deletion. For institutional use, we support data segregation at the college level (multi-tenant isolation) to ensure one organization’s data is never exposed to another.
On the AI side, we implement safeguards to prevent data leakage, including secure prompt handling, anonymization where required, and controlled use of third-party APIs. We also monitor for model misuse, enforce content safety filters, and maintain human-in-the-loop oversight for critical decisions such as evaluations.
Additionally, we follow best practices in infrastructure security—using secure cloud environments, regular vulnerability assessments, and continuous monitoring—to ensure system reliability and protection against threats. Overall, our approach ensures that user data is protected, compliant, and handled responsibly while enabling powerful AI-driven learning experiences.” |
Yes, there are potential regulatory risks, primarily around data privacy, AI governance, and education compliance, but we proactively design our platform to mitigate them. Since PragyanAI handles student data, project work, and behavioral insights, evolving data protection laws (such as GDPR and India’s DPDP Act) could impact how we collect, store, and process data. We address this through a privacy-first architecture, explicit user consent, data minimization, and strong encryption.
Another area is emerging AI regulations, especially around automated decision-making (e.g., AI-based evaluations, scoring, and career recommendations). To mitigate this, we maintain transparency in AI outputs, provide explainability where needed, and incorporate human-in-the-loop oversight for critical evaluations like grading or assessments.
In the education sector, institutional and accreditation requirements may vary across regions, particularly when integrating into formal curricula. We position PragyanAI as a complementary skill development and infrastructure layer rather than a replacement for accredited degrees, which reduces regulatory friction.
Additionally, there could be compliance considerations when integrating with third-party platforms (e.g., GitHub, payment systems, or communication tools), which we manage through secure APIs and adherence to platform policies. |
At 10x scale, the primary challenges in our AI system would be around latency, retrieval quality, evaluation consistency, and infrastructure cost—but these are known scaling constraints that we are proactively designing for.
First, RAG performance can degrade as data grows significantly—larger vector stores can reduce retrieval precision and increase latency. To address this, we plan to implement hybrid search (vector + keyword), better indexing strategies, and domain-specific chunking to maintain high-quality retrieval.
Second, LLM latency and cost can increase with higher usage, especially for real-time mentor interactions and evaluations. We mitigate this through model optimization (smaller fine-tuned models for specific tasks), caching, batching, and asynchronous processing where real-time response is not critical.
Third, consistency in AI evaluation (e.g., AI viva and scoring) may vary at scale. To handle this, we use structured rubrics, calibration datasets, and continuous benchmarking against human evaluations to maintain reliability.
Fourth, multi-agent orchestration can become complex with scale, leading to coordination overhead and potential failure points. We address this by modularizing agents, implementing clear task boundaries, and using orchestration frameworks with monitoring and fallback mechanisms.
Finally, infrastructure scaling—handling concurrent users, real-time sessions, and large volumes of data—requires robust cloud architecture, load balancing, and distributed systems, which we plan to handle through microservices and scalable cloud infrastructure.
In summary, while scaling introduces challenges in performance, cost, and consistency, our architecture is designed with modularity, optimization, and continuous feedback loops to ensure reliability and efficiency as we grow.” |
Yes, we have a strong in-house AI/ML team led by founders with deep expertise in Generative AI, Agentic AI systems, and applied machine learning. The team has hands-on experience in building end-to-end AI products, including LLM-based applications, RAG pipelines, multi-agent systems (using frameworks like LangChain, CrewAI, and LangGraph), and deploying scalable AI solutions using FastAPI and modern cloud infrastructure.
In addition to the core team, we have access to a continuously growing pool of skilled talent through our own ecosystem—students, researchers, and practitioners who actively contribute to projects, experiments, and product development. This gives us a unique advantage in both talent acquisition and rapid iteration, as we are constantly training and working with AI-skilled individuals within our network.
Overall, our strength lies not only in a capable core team but also in a scalable talent pipeline that ensures we always have access to skilled AI/ML resources as we grow. |
We use a combination of publicly available, licensed, and proprietary datasets, depending on the use case. For our AI learning and RAG systems, we primarily rely on user-provided content such as PPTs, PDFs, videos, code repositories, and curriculum materials, which are securely processed with user consent. We also leverage publicly available datasets and pre-trained models (e.g., from Hugging Face and open research sources) where applicable, ensuring compliance with their respective licenses.
Our key proprietary data comes from platform usage—student projects, GitHub activity, AI evaluations (viva, quizzes), engagement signals, and learning behavior—which we generate and own as part of our “Skill DNA” and knowledge graph. This data is anonymized and used to improve personalization, recommendations, and skill validation models.
In summary, we combine licensed/public data with a growing proprietary dataset built from real user interactions and outcomes, which becomes a core long-term advantage for the platform. |
We continuously improve model performance through a combination of data-driven feedback loops, evaluation pipelines, and model optimization strategies. At the core, our system learns from real user interactions—student queries, project work, AI evaluations, and engagement patterns—which feed into our “Skill DNA” and knowledge graph. This allows us to refine prompts, improve retrieval quality, and personalize responses over time.
We implement continuous evaluation using both automated metrics (accuracy, relevance, groundedness, hallucination rate) and human-in-the-loop validation, especially for critical components like AI viva and assessments. We also run A/B testing on prompts, model variants, and agent workflows to identify the most effective configurations.
On the RAG side, we improve performance by optimizing chunking strategies, embeddings, and hybrid retrieval (vector + keyword), ensuring more accurate and context-aware responses. For scalability and efficiency, we selectively fine-tune smaller models for specific tasks (e.g., evaluation, summarization) while using larger models for complex reasoning.
Additionally, our multi-agent system evolves through continuous monitoring of agent success rates, fallback behavior, and user feedback, enabling us to refine orchestration and task execution.
Overall, our approach combines proprietary data, continuous feedback, rigorous evaluation, and iterative optimization to ensure the system becomes more accurate, reliable, and personalized as it scales. |
We are building for both India and global markets, with India as our initial focus and launchpad. India has one of the largest student populations and a significant gap between education and industry readiness, making it an ideal market to validate our model at scale. By partnering with colleges, running bootcamps, and deploying our platform across campuses, we can rapidly iterate, generate proprietary data, and prove measurable outcomes.
Our platform, however, is designed to be globally scalable from day one. The core architecture—AI-powered learning, Skill Passport, agentic evaluation, and career intelligence—is not region-specific and can be adapted to different education systems and job markets. As we build strong traction and a campus network in India, we plan to expand into other emerging and developed markets where similar skill gaps exist.
In summary, India is our beachhead market for rapid adoption and validation, while our long-term vision is to become a global infrastructure layer for skill-based education and talent intelligence. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Su77ojzGz8Gq-PragyanAI-Product-RoadMap-Deck-Investor-Deck_Incubator.pdf, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/fY4ZshzcBV6B-PragyanAI_Scaling_Bharat_Pitch_Deck-Owl-VC-Investor-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Su77ojzGz8Gq-PragyanAI-Product-RoadMap-Deck-Investor-Deck_Incubator.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/fY4ZshzcBV6B-PragyanAI_Scaling_Bharat_Pitch_Deck-Owl-VC-Investor-Deck.pdf |
https://youtu.be/Ie252B0OfMc |
Yes, PragyanAI is a deeply mission-driven and impact-focused startup. Our core mission is to bridge the gap between education and employability by transforming how students learn, prove, and apply skills in the real world. Today, millions of students graduate without job-ready skills, while industries struggle to find qualified talent—this mismatch is what we are solving at scale.
We aim to democratize access to high-quality, AI-driven skill development by enabling every student—regardless of college tier or location—to learn cutting-edge technologies, build real projects, and receive verified proof of competence. Through our platform, students in Tier-2 and Tier-3 colleges can access the same opportunities, mentorship, and industry exposure as those in top institutions.
Our Campus Intelligence Network and AI-powered systems ensure that learning is not just theoretical but practical, personalized, and outcome-driven. By integrating real-world projects, continuous evaluation, and career pathways, we empower students to become globally competitive talent.
Ultimately, our impact is to create a skill-first ecosystem where opportunities are based on verified ability rather than background, enabling upward mobility for students and building a more efficient, merit-driven talent economy. |
Yes . North Karnataka and Women Led |
linkedin |
checked |
| Mar 20, 2026 @ 7:36 PM |
Rakshit Sawarn, Amruta Vivek Saharkar |
rakshitsawarn@gmail.com |
http://www.linkedin.com/in/rakshitsawarn |
https://www.definitisoftware.com/ |
+91 7479687456 |
Rakshit Sawarn is CMD of the company and an undergraduate B.Tech student. Amruta Vivek Saharkar is the CEO of the company and an undergraduate B.Tech student. Both of them are from Computer Science and Engineering background and speciallised in Artificial Intelligence. |
We first met during our undergraduate studies, where we were classmates. Over time, we collaborated on academic projects and developed a strong working relationship based on trust, complementary skills, and shared interest in solving real-world problems.
We have been working together closely for over 3 years, during which we have built and refined our startup idea, aligning on long-term vision and execution. |
All |
Our biggest strength as a team is our strong alignment on vision combined with complementary skill sets. While one of us focuses on technical development and product innovation, the other brings execution, problem-solving, and business perspective.
We are highly execution-driven and have consistently taken ideas from concept to implementation, including building our current product and working towards real-world impact. Our ability to make quick decisions, adapt, and stay committed through challenges sets us apart. |
DEFINITI SOFTWARE PRIVATE LIMITED |
https://www.definitisoftware.com/ |
New Delhi |
We are building a technology-driven solution to detect and prevent counterfeit products across supply chains, ensuring authenticity and trust for brands and consumers. |
Counterfeit products are a massive and growing global problem, causing significant revenue losses for brands, safety risks for consumers, and erosion of trust across supply chains. Existing solutions are often fragmented, expensive, or difficult to scale, making it hard for businesses and consumers to reliably verify product authenticity. As a result, counterfeit goods continue to circulate freely, impacting industries ranging from pharmaceuticals to consumer goods, with no seamless, scalable solution in place. |
Our solution follows a simple and user-friendly workflow through the DE-FAKE application.
The user begins by opening the DE-FAKE app and selecting a verification method. We provide three modes of authentication:
1. Barcode – The user scans the product’s barcode, which is then cross-verified with our database, blockchain, and secure backend to check authenticity.
2. Image Analysis – The user uploads or captures an image of the product, and our AI-based system analyzes it to detect signs of counterfeiting.
3. Manual Input – The user can manually enter product details, which are then verified against our database, blockchain, and backend systems.
In all three cases, the system processes the input and instantly displays whether the product is authentic or counterfeit.
Additionally, our system is capable of tracing the exact point of purchase, including the shop name and location, enabling brands to identify where counterfeit products are being sold and take targeted action. |
Our solution is built on proprietary technology that enables multi-layered counterfeit detection, making it significantly harder to replicate compared to traditional single-layer approaches like barcodes or QR codes. Unlike existing solutions, DE-FAKE does not require any changes to product packaging, allowing seamless adoption at scale. We combine this with high real-world accuracy (98%+) and sub-second response times, making the solution both practical and reliable for enterprises. Additionally, our patent-backed approach and continuous data-driven improvement create strong long-term defensibility, as the system becomes more accurate and robust with increased usage. |
Users |
Pilots |
Our primary target customers are brands and manufacturers |
₹15.71 lakh crore |
₹1.57 lakh crore |
₹5,000 |
Subscription based |
Sproxil, Poizon, Verify |
We acquire customers through direct on-ground enterprise outreach and targeted digital marketing. |
We go to market through pilot-led enterprise sales, scaling via partnerships, on-ground outreach, and targeted digital marketing. |
Our long-term vision is to make DE-FAKE the global standard for product authentication, eliminating counterfeit goods across industries and building trust in every transaction.
We aim to create a universally accessible verification layer integrated across supply chains, e-commerce platforms, and regulatory systems, ensuring that every product can be instantly verified anywhere in the world. |
Incorporated on 29-03-2025 |
200000 |
Yes |
We are applying to Avinya to accelerate our growth from pilot-stage to large-scale enterprise adoption through mentorship, strategic connections, and ecosystem support. |
We aim to convert pilot deployments into paying enterprise clients while refining our product and go-to-market strategy for scalable growth. |
virtual participation |
Yes |
Core engine |
Our AI architecture is built on a multi-layered framework combining computer vision and data-driven validation models. The system analyzes product images using deep learning to detect visual inconsistencies, while additional validation layers cross-check patterns and authenticity signals to improve accuracy.
It is designed for real-time inference with a lightweight, scalable pipeline that can be deployed via mobile applications and integrated with enterprise systems. The architecture continuously improves through feedback loops and data-driven model updates. |
We are building a proprietary dataset of real-world product scans across industries, including both authentic and counterfeit samples. This data is enriched with contextual insights such as point-of-sale information, including shop-level and location data, enabling deeper detection and traceability.
As adoption grows, each scan strengthens our dataset through continuous feedback loops, improving model accuracy and enabling us to identify counterfeit patterns across regions and distribution networks. This creates a compounding data advantage that is difficult for new entrants to replicate. |
Our AI component is defensible due to its combination of proprietary real-world data, multi-layered model architecture, and continuous learning loops. Unlike generic models, our system is trained on domain-specific datasets of authentic and counterfeit products, improving accuracy across industries.
Additionally, each scan strengthens our dataset, creating a compounding data advantage over time. Combined with our patent-backed approach and real-world deployment, this makes our AI system increasingly difficult to replicate. |
We evaluate model performance using accuracy, precision, recall, and F1-score to ensure reliable counterfeit detection while minimizing false positives and false negatives. In addition, we monitor real-world performance through live deployment metrics such as detection consistency, response time, and user feedback.
We continuously validate and improve the model using new data from real-world scans, ensuring robustness and reliability across different products and environments. |
We follow a privacy-first approach by minimizing user data collection and ensuring that sensitive information is securely stored. All data is encrypted both in transit and at rest, and access is strictly controlled through role-based permissions.
We are aligned with standard data protection practices and continuously work towards compliance with relevant regulations. Additionally, we implement secure APIs and infrastructure best practices to ensure data integrity and protection against unauthorized access. |
No major regulatory risks, as our solution works within existing frameworks, though we remain aligned with evolving compliance requirements. |
At 10x scale, potential bottlenecks could arise in infrastructure load, and maintaining consistent model performance across diverse product categories and environments.
To address this, we are building a scalable architecture with optimized inference pipelines, cloud-based infrastructure, and continuous model retraining using diverse real-world data to ensure performance and reliability at scale. |
Yes, we have an in-house AI/ML team with expertise in computer vision and deep learning. The team focuses on building and optimizing our multi-layered detection models, handling real-world deployment, and continuously improving accuracy using domain-specific data. |
We primarily use a combination of proprietary and curated datasets, including real-world product images collected through our pilots and testing, along with publicly available data for initial model training.
Our core datasets are increasingly proprietary, built from authentic and counterfeit samples, which are owned and continuously expanded as part of our deployment and feedback loop. |
We continuously improve model performance through a feedback-driven learning loop, where real-world scan data is used to retrain and refine our models. We incorporate new product variations, edge cases, and counterfeit patterns to enhance accuracy over time.
Additionally, we regularly monitor performance metrics, run validation cycles, and optimize our models to ensure consistent reliability as we scale across industries. |
We are building for both India and global markets. |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/uVI2yjzyEgVr-DEFINITI-SOFTWARE-PRIVATE-LIMITED.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/uVI2yjzyEgVr-DEFINITI-SOFTWARE-PRIVATE-LIMITED.pdf |
https://drive.google.com/file/d/1mjxcmxi2ZUVhzlihbH605iThNKkoGsLu/view?usp=drive_link |
Yes, we are building a mission-driven startup focused on eliminating counterfeit products and restoring trust across global supply chains. Counterfeiting not only causes significant financial losses for businesses but also poses serious risks to consumer safety, especially in sectors like pharmaceuticals.
Through DE-FAKE, we aim to create a transparent and reliable ecosystem where every product can be verified instantly, protecting both brands and consumers while contributing to a safer and more trustworthy marketplace. |
yes |
Ashok Mishra |
checked |
| Mar 18, 2026 @ 2:23 PM |
Ranjan Kumar Baisak |
ranjan@xhield.tech |
http://linkedin.com/en/ranjanbaisak |
https://xhield.tech |
9880398951 |
Ceo |
We are working together from last 10years or so |
two |
Compatibility |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 18, 2026 @ 12:51 AM |
Niharika Sri Parasa , Venkata Sridhar Parasa |
pnsri@advaita1.com |
https://linkedin.com/in/niharika-sri-parasa |
https://www.advaita1.com |
8247654106 |
CEO ( PhD in Data Science), Shareholder ( MBBS, MD) |
Family |
two |
We cover 2 domains |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Mar 17, 2026 @ 6:06 PM |
Patrick Joseph |
joseph.patrick87@gmail.com |
https://www.linkedin.com/in/patrick-joseph-7b691726/ |
https://www.learningwell.ai/ |
7738155269 |
Patrick Joseph – Founder. Patrick brings over14 years of experience in US Financial Service and EdTech space, with over 8+ years of experience in global EdTech operations, product workflows, and education program scaling. At Elevate K-12, he led large-scale operational and content initiatives supporting 500+ education programs, managing distributed teams across the US and India while working closely with teachers, curriculum teams, and school systems. Through this experience, he observed first-hand: • Teachers spending excessive time on manual content creation • Limited personalization tools for students • Significant gaps in technology supporting special education workflows Additionally, his personal journey as a parent of an autistic toddler exposed the significant challenges faced by families, therapists, and educators in accessing effective tools for special education support and individualized learning plans (IEPs). These insights directly shaped the vision behind Learningwell AI — a platform designed around real educator needs rather than generic AI tools. |
Learningwell AI is currently founder-led. The company was started by a founder with over 8 years of experience in the education sector, working closely with teachers, curriculum teams, and school systems across the US and India. Through this experience, the founder gained deep insight into the operational challenges educators face daily—particularly around lesson preparation, differentiated instruction, and special education support.
This direct exposure to real classroom workflows and large-scale education programs led to the vision for Learningwell AI: building an AI-driven operating system that supports teachers, students, and special education needs in one integrated platform.
As the product progresses through the MVP stage, we are actively expanding the team by bringing in AI engineers, education specialists, and product advisors who share the mission of transforming education with practical, workflow-driven AI solutions. |
1 |
Our biggest strength is the combination of deep education domain expertise and execution capability to build scalable solutions for real classroom problems.
Our founding team brings 8+ years of hands-on experience in large-scale education operations, working directly with teachers, curriculum teams, and school systems across the US and India. This has given us an insider’s understanding of how classrooms actually function — from lesson planning and assessments to differentiated instruction and special education support.
Because of this background, we are not building AI for education from the outside. We are building it from within the system.
We understand the workflows teachers follow, the time pressures they face, and the gaps that existing tools fail to solve. This allows us to design AI products that integrate naturally into teaching and learning processes rather than adding more complexity.
This combination of deep domain insight, operational scale experience, and a clear product vision for AI infrastructure in education allows us to move quickly, build with precision, and create solutions that schools and educators can adopt at scale. |
Learningwell AI Private Limited |
https://www.learningwell.ai/ |
|
The AI Operating System for Modern Education |
Education systems globally face structural inefficiencies that limit teaching effectiveness and student outcomes. 1. Teacher Workload Crisis Teachers spend 40–50% of their time on administrative and preparation tasks such as lesson planning, worksheet creation, and assessments instead of focusing on teaching and student engagement. 2. Limited Personalization in Classrooms Students learn at different speeds and styles, yet most classrooms still operate under one-size-fits-all instruction, leaving many learners unsupported. 3. Special Education Resource Gap Special education teachers, parents, caregivers, and therapists often struggle to create IEPs, adapted learning materials, and behavioral support plans due to time constraints and limited tools. 4. Fragmented EdTech Ecosystem Existing AI tools are mostly generic or disconnected, making it difficult for educators to integrate them into real classroom workflows. |
Learningwell AI is building an AI-powered education infrastructure platform designed to support teachers, students, and special education ecosystems through intelligent automation and personalized learning tools.
The platform integrates three AI-powered modules:
AI Teacher Magic Box – helps teachers instantly create lesson plans, presentations, worksheets, assessments, and classroom materials.
AI Student Magic Box – provides students with a 24/7 AI learning companion for personalized tutoring, homework support, and exam preparation.
AI SPED Magic Box – AI tools designed for special education to generate IEP goals, simplify learning materials, and support inclusive education.
Our goal is to embed AI directly into real education workflows, enabling teachers to save time, improve classroom effectiveness, and support diverse learners more efficiently. |
Our key differentiation is our strong focus on Special Education (SPED), a critical area that most AI education platforms have largely ignored. Learningwell AI integrates AI Teacher Magic Box, AI Student Magic Box, and a dedicated AI SPED Magic Box within one platform. While many tools help generate generic classroom content, our platform enables teachers to automatically adapt lessons, generate IEP-aligned materials, simplify learning content, and create differentiated worksheets for students with diverse learning needs. This SPED-first approach, combined with our deep understanding of classroom workflows, allows us to address a major gap in education. As teachers create and adapt content for different learning needs, the platform builds a unique dataset of differentiated learning materials, creating a strong long-term defensibility layer. |
MVP |
Pilots |
Educators, Learners, Parents/Caregivers of kids with special needs, SPED teachers. |
The global EdTech market is projected to exceed $400B+ by 2030, with AI-driven education tools and inclusive learning solutions growing rapidly. This represents the total market opportunity for AI-powered education platforms like Learningwell AI |
Our initial focus is on AI tools for teachers, students, and Special Education (SPED) in schools and learning institutions. This segment is estimated at $40B–$60B globally, including digital classroom tools and SPED learning support solutions. |
In the early stages, we aim to capture a small share by targeting schools, educators, NGOs, and therapy centers, starting with India and expanding globally. Our realistic initial market capture is $5M–$10M over the first few years through institutional partnerships and teacher adoption. |
The first layer is subscriptions from teachers, students, and SPED users. The second layer is pay-per-use AI credits, allowing users to access advanced content generation and learning tools. The third layer is an AI marketplace, where educators can access premium lesson packs, templates, and learning assets. This creates a high-margin SaaS business with scale |
Magicschool AI, Monsha AI, Buddiz AI, Teachmate AI. These are some of the applications which have both Teahcer and Student AI. But none of them are operating in to the SPED segment. |
First, we reach educators through teacher communities, social media, webinars, and educator networks, offering free access to core AI tools that help them save time in lesson planning and classroom preparation. As teachers begin using the platform, students naturally engage with the AI learning tools, creating organic adoption within classrooms. Once usage grows, we approach schools with institutional licensing, particularly highlighting our SPED-focused capabilities, which address a major unmet need in many schools. This teacher-first approach helps us build early traction, product validation, and scalable institutional partnerships. |
We begin by offering freemium AI tools that allow teachers to experience the platform and immediately see the value. From there, we grow adoption through teacher certification programs, webinars, and educator communities, where educators learn how to use AI effectively in the classroom. Students then join the ecosystem through learning communities, interactive learning tools, and peer engagement. A key growth driver will be our referral model, where teachers and students can invite others to the platform and earn rewards such as free AI credits, premium features, or access to exclusive learning resources. As the platform gains adoption and trust among educators, we expand into school partnerships and institutional adoption. “In simple terms, we first build a strong teacher and student community, and then scale through institutional contracts.” |
Our long-term vision is to build the AI infrastructure layer for education, operating in both B2C and B2B space, supporting teaching, learning, and inclusive education globally. |
Incorporated in Jan 2026 as a Private Limited Company |
Have not raised any funding yet |
Yes |
We are applying to Avinya to gain the right mentorship, ecosystem support, and strategic guidance needed to scale Learningwell AI |
Product & Technology Mentorship Guidance on building scalable AI infrastructure and refining the MVP. Market Validation Access to networks of educators, schools, and EdTech stakeholders for early product testing. Go-To-Market Strategy Support Mentorship on building effective strategies for user acquisition and platform growth. Startup & Fundraising Guidance Support in preparing the venture for early-stage funding and investor engagement. |
Virtual |
Yes |
Core engine |
Frontend: React / Next.js
Backend: Node.js
AI Layer: Open AI
Database: MongoBD
Cloud infra optimized for cost control |
Our proprietary data advantage will come from the differentiated learning and SPED-focused content generated and used on our platform.
As teachers create lesson plans, worksheets, assessments, and SPED-adapted materials such as simplified content and IEP-aligned resources, the platform builds a growing dataset of structured, differentiated learning materials across subjects and learning levels.
Over time, interactions from teachers and students will generate feedback signals and usage patterns, allowing us to improve personalization and adaptive learning. This creates a unique education and SPED-focused dataset that becomes increasingly valuable and difficult for competitors to replicate. |
Our AI defensibility comes from domain-specific intelligence and SPED-focused data.
Unlike generic AI education tools, our platform is designed around education workflows and Special Education (SPED). As teachers use the platform to generate lesson plans, differentiated worksheets, and IEP-aligned learning materials, the system continuously builds a unique dataset of structured education and SPED-adapted content.
This creates a proprietary learning dataset and feedback loop that improves the AI’s ability to generate personalized and differentiated educational resources. Over time, this education-specific and SPED-focused intelligence layer becomes difficult for competitors to replicate. |
We evaluate model performance using a combination of quality, accuracy, and reliability metrics.
For content generation, we measure relevance, curriculum alignment, and clarity through educator feedback and internal review scoring. For AI responses, we track accuracy, hallucination rate, and response consistency.
We also monitor user engagement metrics such as content usage, edits made by teachers, and completion rates of generated materials. For SPED-related outputs, we validate content against IEP alignment and readability levels to ensure materials are appropriate for diverse learning needs.
Continuous feedback from educators and usage analytics helps us fine-tune prompts, improve outputs, and maintain reliability over time. |
We prioritize data privacy, security, and compliance by following industry best practices for handling educational data.
The platform uses secure cloud infrastructure, encrypted data storage, and role-based access controls to protect user information. We minimize the collection of sensitive student data and ensure that any data used by AI models is anonymized and securely processed.
As we expand into institutional partnerships, we will align with relevant data protection and education compliance standards, ensuring that schools, educators, and students can use the platform safely and responsibly. |
Yes, there are potential regulatory considerations related to data privacy, student safety, and AI use in education.
We address this by designing the platform with strong data privacy and security practices, ensuring compliance with relevant regulations such as student data protection and institutional data policies. We also implement safeguards to ensure AI-generated content is used as a support tool for educators rather than replacing professional judgment.
As we expand into institutional partnerships, we will align with education compliance standards and responsible AI guidelines to ensure the platform remains safe, transparent, and reliable for schools and learners. |
If we scale 10x, the main challenges would be model inference costs, latency, and infrastructure load.
A rapid increase in users generating AI content could lead to higher API usage, slower response times, and increased compute costs. To address this, we plan to optimize through model caching, prompt optimization, batching requests, and gradually moving some workloads to fine-tuned or smaller specialized models.
We are also designing the system with scalable cloud infrastructure and microservices, allowing us to distribute workloads efficiently and maintain performance as usage grows. |
Currently, we are in the early stage of building our AI/ML team.
The core team brings strong education domain expertise and product vision, particularly in curriculum development, classroom workflows, and Special Education (SPED). As we develop the platform, we are working with AI engineers and technical collaborators experienced in LLMs, prompt engineering, and AI-powered content generation systems.
Our focus is to build a team that combines AI engineering expertise with deep education knowledge to create practical and scalable AI solutions for teachers, students, and special education support. |
Currently, our system uses publicly available educational resources, curriculum-aligned reference materials, and AI model training data provided through licensed APIs (such as LLM providers).
We also use curriculum frameworks, open educational resources, and structured prompts to generate lesson plans, worksheets, and assessments. As the platform grows, we will build our own proprietary dataset from teacher-generated content, differentiated learning materials, and SPED-adapted resources created within the platform.
This approach allows us to start with licensed and open datasets while gradually building owned data assets through platform usage. |
We plan to continuously improve model performance through iterative feedback, data refinement, and model optimization.
As teachers and students use the platform, we collect usage signals such as edits, ratings, and engagement patterns to understand how well the generated content performs. These insights help us refine prompts, improve retrieval pipelines, and optimize model outputs.
Over time, the differentiated content created on the platform—especially SPED-adapted materials and IEP-aligned resources—will help us build a proprietary dataset, enabling further fine-tuning and improving the accuracy, personalization, and reliability of our AI system. |
Both |
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/7XnLKltCp5W6-Learningwellai-Final.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/7XnLKltCp5W6-Learningwellai-Final.pdf |
http://The%20Product%20is%20currently%20in%20MVP%20and%20we%20do%20not%20have%20a%20demo%20video%20yet. |
Yes, Learningwell AI is a mission-driven startup focused on making education more inclusive and accessible through AI.
Our mission is to help educators provide personalized learning support for every student, especially those with diverse learning needs and special education requirements (SPED). Many classrooms lack the time and resources to create differentiated materials or IEP-aligned support, which leaves many students underserved. |
NA |
NA |
checked |
| Oct 6, 2025 @ 5:01 PM |
Core Team Members: Dr. Amit Kumar Jaiswal – Head of Technology Research Sharmistha Chatterjee – Head of Technology & Architect (6 patents) Aniruddha Choudhury – Lead Data Scientist (2 patents) Avik Bhattacharjee – Head of Go-To-Market (1 patent & 1 trade secret) Dr. Avidip De – Head of Clinical Research |
amit.chr@iitbhu.ac.in |
https://www.linkedin.com/in/amitkumarjaiswal1/ |
https://inceptragen.webflow.io/ |
9886181117 |
🧠 Dr. Amit Kumar Jaiswal Role: Head of Technology Research Background: Academic affiliation with IIT (BHU) Expert in AI/ML research and healthcare technology innovation Leads R&D for multimodal AI in mental health and oncology applications 💡 Sharmistha Chatterjee Role: Head of Technology & Architect Background: Alumni of London & Oxford School of Business Holds 6 patents in technology and AI-driven architecture Specializes in system design, scalable AI infrastructure, and enterprise architecture 📊 Aniruddha Choudhury Role: Lead Data Scientist Background: Alumni of the Indian School of Business (ISB) Holds 2 patents in AI and data science Expert in predictive analytics, machine learning, and multimodal data fusion 🚀 Avik Bhattacharjee Role: Head of Go-To-Market Background: Alumni of IIM Bangalore (Executive MBA) Former Amazon Healthcare GTM Lead Holds 1 patent and 1 trade secret Expert in AI/ML business strategy, healthcare GTM, and enterprise partnerships 🩺 Dr. Avidip De Role: Head of Clinical Research Background: Alumni of Calcutta Medical College and Tata Medical Center Consultant Laparoscopic & GI-HPB Onco Surgeon |
The team of OncoSaathi came together through a shared mission — to bridge the gap between mental health and oncology care in India.
Dr. Amit Kumar Jaiswal and Sharmistha Chatterjee first collaborated in early 2021 on an AI research project focused on multimodal speech and emotion recognition at the intersection of healthcare and technology.
Aniruddha Choudhury joined soon after, contributing his data science and model optimization expertise, forming the core technology trio behind InceptraGen’s AI research initiatives.
In 2022, Avik Bhattacharjee, then leading Healthcare GTM at Amazon, connected with the team through an industry–academia innovation forum, helping shape the business strategy, partnerships, and clinical adoption roadmap.
Dr. Avidip De joined as the clinical research lead, bringing his frontline oncology experience from Tata Medical Center and Calcutta Medical College, ensuring clinical depth and validation for the product.
Together, the team has been collaborating for over three years (since 2021) — combining AI research, business strategy, data science, and clinical expertise to build India’s first culturally aware, multimodal AI platform for mental health in oncology. |
No |
Our biggest strength lies in our complementary expertise and shared purpose — blending deep clinical insight, AI innovation, and business execution into one mission-driven unit.
Clinical Depth: Dr. Avidip De ensures every innovation is validated in real oncology settings, grounded in patient impact.
AI & Research Excellence: Dr. Amit Jaiswal, Sharmistha Chatterjee, and Aniruddha Choudhury bring cutting-edge research in multimodal AI, emotion recognition, and data science — transforming ideas into deployable healthcare solutions.
Market & Scale Experience: Avik Bhattacharjee bridges the innovation-to-adoption gap with his GTM and healthcare strategy experience from Amazon and industry partnerships.
Together, we combine academic rigor, clinical empathy, and enterprise execution — allowing us to move from prototype to scalable impact faster than traditional research or startup teams.
Our strength is not just what we build — but how seamlessly we integrate technology, care, and culture to solve one of India’s most urgent healthcare challenges. |
InceptraGen LLP |
https://inceptragen.webflow.io/ |
|
OncoSaathi is India’s first multimodal AI mental health companion for cancer patients — bridging clinical care and emotional healing through culturally aware, multilingual support. |
Cancer patients in India face a hidden mental health crisis — over 40% experience depression or anxiety, yet less than 1 psychiatrist per 100,000 people is available for support. Language barriers, social stigma, and fragmented clinical systems make it nearly impossible to provide timely, culturally sensitive mental health care. Hospitals lack tools for real-time emotional monitoring, early distress detection, and integrated mental health workflows. We’re solving this by creating India’s first multimodal, multilingual AI platform that understands speech, text, and emotion — enabling continuous mental health support, crisis prevention, and seamless integration with oncology hospitals and EMRs. |
OncoSaathi is India’s first multimodal, multilingual AI mental health companion designed specifically for cancer patients and caregivers.
It combines speech, text, and emotional cues to assess psychological well-being in real time — across 22+ Indian languages. The platform uses privacy-preserving, on-device AI to detect early signs of depression, anxiety, or distress and provide personalized, culturally aware support through empathetic conversational guidance and clinical escalation alerts.
For hospitals, OncoSaathi integrates seamlessly with EMR/EHR systems (FHIR, HL7, Epic, Cerner) to deliver real-time risk alerts, automate mental health assessments (PHQ-9, GAD-7), and improve treatment adherence.
Through this fusion of clinical AI, emotional intelligence, and cultural empathy, we enable healthcare systems to transform cancer care from reactive treatment to proactive emotional healing — saving lives, reducing costs, and improving patient outcomes. |
OncoSaathi stands apart through its deep clinical integration, cultural adaptation, and proprietary AI technology purpose-built for India’s oncology landscape. Multimodal Intelligence: Fuses speech, text, and emotional signals for real-time psychological insights — a first for Indian oncology mental health. Cultural & Linguistic Adaptation: Understands 22+ Indian languages and code-switching (Hindi–English mix) with emotion-mapping tuned to Indian speech and behavior patterns. Proprietary Clinical Dataset: Built from longitudinal oncology interactions and clinically annotated emotional datasets, providing a defensible data moat. Seamless Hospital Integration: Natively supports FHIR/HL7, Epic, Cerner, and other EMR/EHR systems — enabling plug-and-play adoption across NCG and private cancer centers. Responsible, Privacy-First AI: Uses federated learning, differential privacy, and on-device inference to ensure security, trust, and compliance with Indian healthcare data laws. First-Mover Advantage: Only platform combining AI mental health, multilingual capability, and oncology workflow integration — validated through pilot partnerships with NCG hospitals. |
Idea |
Users, Pilots |
Primary Customers – Healthcare Institutions Oncology Hospitals & Cancer Care Centers (NCG network + private hospitals) Diagnostic and treatment centers offering chemotherapy, palliative, or radiation therapy Hospital chains seeking integrated mental health screening and EMR-compatible AI solutions Secondary Customers – End Users Cancer Patients and Caregivers needing 24/7 emotional support, crisis detection, and culturally aligned guidance via mobile app Tertiary Customers – Ecosystem Partners Government Health Programs (e.g., Ayushman Bharat, IndiaAI initiatives) integrating mental health into oncology care Digital Health Platforms, Telemedicine Providers, and Insurance Partners leveraging AI-based emotional analytics and patient engagement tools |
₹3,500–5,000 crore ($400–600M) annual market Rapid 20%+ YoY digital health adoption |
Target: 1,000+ cancer-focused hospitals, >700,000 annual oncology admissions Early partnerships: 100+ NCG & private centers |
Urban, tier 1, tier 2, tier 3, and Rural popolation |
B2B ( Hospital Subscription, Integration/API Access) B2C ( Patient Premium Subscription, Pay-Per-Use Teletherapy) |
🧠 1. AI Mental Health Platforms (General) These companies provide AI-based emotional or psychological support, though not necessarily cancer-focused: Wysa – AI chatbot for mental health therapy, used globally (Indian origin). Youper – AI emotional health assistant using CBT principles. Woebot Health – Clinically validated AI companion for mental health. Replika – General-purpose emotional support AI companion. 🩺 2. Cancer-Focused Mental Health & Support Apps These focus specifically on oncology and patient emotional well-being: Cerebral – Offers mental health therapy, including for chronic illness patients. Careology (UK) – Digital cancer care platform with wellbeing tracking. Belong.Life – Community and support platform for cancer patients (popular globally). OncoConnect – Oncology patient engagement and support system (some Indian hospitals use similar systems). 🌐 3. Indian Healthtech / Oncology-Focused Startups Compete in overlapping areas like tele-oncology, patient tracking, and emotional support: Onco.com – India’s leading cancer care navigation platform. Karkinos Healthcare – Comprehensive oncology management platform (partnering with NCG hospitals). Navya Network – AI-assisted cancer treatment decision platform. HealthifyMe / Mfine / Practo – While not oncology-specific, they’re potential competitors for mental health and teleconsultation integrations. 💡 Summary of Competitive Advantage (vs. Others) OncoSaathi differentiates itself by: Being India’s first multimodal AI mental health platform for cancer patients. Offering multilingual, culturally-aware emotional support (Hindi, English, Hinglish, etc.). Integrating directly with hospital EMRs (NCG, FHIR, HL7, etc.). Running on-device AI (privacy-preserving) rather than cloud-only models. |
Customer Acquisition (OncoSaathi) Hospitals (B2B): Direct outreach to oncology hospitals and NCG centers Partnerships with EMR/HMS vendors for seamless integration Clinical pilots to demonstrate impact and build trust Referrals through existing hospital networks and government collaborations Patients (B2C): In-app referrals from partner hospitals Digital marketing via social media, cancer forums, and awareness campaigns Collaborations with NGOs, telemedicine platforms, and advocacy groups Word-of-mouth from patient communities and peer support features Focus: Build credibility through clinical results, strong partnerships, and culturally personalized engagement. |
Go-To-Market Strategy (OncoSaathi) Target Segments: Primary: Oncology hospitals & NCG/private cancer centers Secondary: Cancer patients & caregivers Tertiary: Government health programs & digital health initiatives Sales Approach: Direct enterprise sales to hospital decision-makers Partnerships with NCG and EMR vendors for credibility & integration Collaborations with NGOs and patient advocacy groups Marketing & Adoption: Pilot in 5–10 hospitals for validation and testimonials Scale to 50+ hospitals post-pilot Digital campaigns to drive patient app downloads Support & Training: Clinical and technical onboarding for hospitals 24/7 multilingual support and feedback-based updates Key Message: Culturally aware, multilingual AI support that improves mental health outcomes, integrates seamlessly, and saves lives. |
Long-Term Vision (OncoSaathi)
To become India’s leading AI-driven mental health companion for cancer care, empowering every patient with culturally aware, multilingual emotional support — from diagnosis to recovery.
We aim to:
Integrate with all major oncology hospitals nationwide through NCG and government partnerships.
Expand to global emerging markets with similar linguistic and cultural needs.
Advance research and policy by generating real-world mental health insights from oncology data.
Transform cancer care from reactive treatment to proactive, holistic healing — saving lives, reducing stigma, and making mental health support accessible to all. |
OncoSaathi is a project powered by InceptraGen LLP (IDN: ACQ-8228) — a legally registered Limited Liability Partnership (LLP) in India. It operates under InceptraGen LLP’s corporate and compliance framework, with partnerships across clinical, academic, and technology institutions for development, validation, and deployment. |
NA |
Yes |
We’re applying to Avinya to accelerate OncoSaathi’s scale and validation as India’s first AI-driven, multilingual mental health companion for cancer patients. Avinya’s ecosystem offers: Mentorship and technical guidance to strengthen our AI and healthcare integration. Access to investors and clinical networks to expand across NCG and private hospitals. Regulatory and go-to-market support to navigate India’s digital health ecosystem. Visibility and credibility through association with a premier innovation platform. With Avinya’s support, we aim to move from successful pilots to nationwide adoption, impacting millions of cancer patients and caregivers. |
During the programme, I want to accelerate OncoSaathi from validated prototype to scalable clinical deployment. My key goals are: Clinical Validation & Regulatory Readiness – Work with mentors and partners to finalize clinical validation protocols, achieve compliance benchmarks, and prepare for medical-grade AI certification. Business Model Refinement – Strengthen our dual B2B/B2C revenue strategy, refine pricing models, and align value propositions with hospital systems and patient needs. Network Expansion – Build partnerships with NCG hospitals, healthcare regulators, and investors who can help us expand reach across India. Impact Measurement – Develop robust success metrics for mental health outcomes and engagement, ensuring measurable clinical and economic impact. Investor Readiness – Prepare for our first institutional fundraise by solidifying financial projections, pitch narrative, and go-to-market roadmap. Ultimately, I aim to turn OncoSaathi into India’s trusted AI mental health companion for cancer care — a model that combines empathy, explainable AI, and real-world clinical impact. |
yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/nOd54Ps7zddg-ACSEL-submission-draft.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/nOd54Ps7zddg-ACSEL-submission-draft.pptx |
http://NA |
Yes — OncoSaathi is a mission-driven and impact-focused startup.
Our mission is to transform cancer care in India by making mental health support accessible, empathetic, and culturally relevant for every patient — regardless of language, location, or socioeconomic status.
In India, over 40% of cancer patients experience psychological distress, yet fewer than 1 psychiatrist per 100,000 people is available. OncoSaathi bridges this gap through an AI-powered, multilingual mental health companion that understands local languages, detects distress early, and connects patients to timely help.
Our impact is threefold:
Patients: Improved emotional resilience, early intervention, and better treatment adherence.
Hospitals: Enhanced patient outcomes, operational efficiency, and compliance with mental health mandates.
Healthcare System: Reduced overall treatment costs and a data-driven understanding of India’s oncology–mental health landscape.
We measure success through engagement, crisis intervention rates, and patient satisfaction — aiming not just for technology adoption, but for real emotional healing and measurable quality-of-life improvement.
Ultimately, we are driven by a simple belief: no one fighting cancer should ever feel alone or misunderstood. |
NA |
amit.chr@iitbhu.ac.in |
checked |
| Oct 3, 2025 @ 5:01 PM |
Aditya Nigam |
adityanigam003@gmail.com |
http://www.linkedin.com/in/aaditya-nigam |
http://www.obsydianclub.com/ |
9115375073 |
Founder & CEO of Obsydian Club – leading product vision, business development, and partnerships with clubs. Responsible for strategy, fundraising, and scaling operations. |
Obsydian Club was founded by me as a solo founder. While I currently lead the project independently, I have been working with partner clubs, early beta users, and mentors for over 6–12 months to validate the idea, build the MVP, and test initial traction. I’m now in the process of expanding the core team with a technical co-founder and early operators. |
Yes |
Our biggest strength as a team is the blend of vision, adaptability, and strong market validation. As founder, I bring persistence, digital marketing expertise, and execution drive, while our network of partner clubs, mentors, and early adopters provide continuous feedback and industry insights. This combination allows us to move fast, validate ideas quickly, and build a product deeply aligned with what both users and clubs want. |
Obsydian Club Pvt Ltd |
http://www.obsydianclub.com/ |
|
“The world’s first social booking app that combines real-time club reservations with AI-powered party squads and in-app vibes.” |
Party-goers waste hours searching for clubs, often face rejection at the door, and struggle to find like-minded groups or trusted event info. At the same time, clubs lose revenue from underfilled tables and missed global audiences. |
Obsydian Club is a members-only platform that lets users discover nearby clubs, instantly book seats, and connect with like-minded people worldwide. Using AI matchmaking, in-app chats/calls, and AR-powered venue previews, it creates a seamless nightlife experience—helping party-goers vibe together before, during, and after the party, while enabling clubs to fill tables and reach global audiences. |
Obsydian Club is the first platform to merge real-time club booking with social discovery. Unlike ticketing apps or generic social platforms, we offer AI-powered party squads, AR venue previews, and in-app vibe chats—features that competitors lack and are hard to replicate quickly. Our exclusive partnerships with clubs and viral referral loops further create defensibility by locking in supply and driving network effects. |
MVP |
Testimonials |
Our primary customers are 18–35-year-old social party-goers who actively visit clubs, festivals, and nightlife events, and use apps like Instagram/TikTok to discover experiences. Secondary customers are clubs and festivals that want to boost ticket sales, fill tables, and retain customers through data-driven insights and promotions. |
“The total addressable market for Obsydian Club includes ~30–35 million urban young adults (18–35 years) in India who actively engage in nightlife and clubbing, representing an annual spending potential of approximately ₹3.75 lakh crore (~$45–50B USD), with significant opportunities for global expansion.” |
“The serviceable available market for Obsydian Club focuses on urban young adults (18–35 years) in Tier-1 and Tier-2 Indian cities—approximately 10–12 million users—who actively use digital platforms to discover, book, and socialize at nightlife events, representing an annual spending potential of around ₹1.25–1.5 lakh crore (~$15–18B USD).” |
“The serviceable obtainable market for Obsydian Club is estimated at 1–2 million active users in key urban cities in India during the first 2–3 years, representing an annual revenue potential of approximately ₹12,500–25,000 crore (~$1.5–3B USD) through subscriptions, bookings, and partner collaborations.” |
Booking Fees & Commissions: Earn a percentage from club/event bookings made through the platform. Subscription Plans: Premium membership for exclusive access, discounts, and priority bookings. Advertising & Partnerships: Revenue from clubs, bars, and event organizers for featured listings and promotions. In-App Purchases & Virtual Experiences: Selling virtual party experiences, event passes, or merchandise. |
Insider.in / BookMyShow: Event discovery and ticketing platforms for nightlife and entertainment. Nearify / Fever: Platforms offering curated nightlife events and experiences. Dineout / Zomato (Nightlife Section): Restaurant and bar booking with nightlife/event integrations. Meetup / Social Networking Apps: Platforms helping people connect for events and parties. |
Digital Marketing: Targeted Instagram, Facebook, and YouTube ads aimed at young adults (18–35) in Tier-1 & Tier-2 cities. Influencer Partnerships: Collaborations with nightlife influencers, DJs, and event hosts to promote the platform. Referral Programs: Incentivize existing users to invite friends with discounts, free passes, or rewards. Campus & Community Engagement: Tie-ups with colleges, universities, and social clubs to onboard student users. SEO & Content Marketing: Blogs, guides, and social media content on nightlife trends to attract organic traffic. |
Target Key Urban Cities First: Launch in Tier-1 cities (Mumbai, Delhi, Bangalore) where nightlife culture is strongest, then expand to Tier-2 cities. Leverage Influencers & Partnerships: Collaborate with nightlife influencers, DJs, clubs, and event organizers for brand awareness and credibility. Digital Marketing Focus: Use paid ads, social media campaigns, and content marketing to reach 18–35-year-old urban users. Referral & Loyalty Programs: Encourage early adopters to invite friends with incentives and rewards to accelerate user growth. Event-Based Promotions: Sponsor or host exclusive events and parties to showcase the platform’s value and attract media coverage. |
“To become the world’s leading nightlife and social discovery platform, connecting party-goers globally, enabling seamless experiences—from discovering events and booking tables to socializing with like-minded people—while creating a thriving digital community for nightlife enthusiasts.” |
“Obsydian Club is a private limited company (Pvt Ltd), officially registered and compliant with Indian corporate regulations.” |
NA |
Yes |
“We are applying to Avinya to gain access to mentorship, funding, and a strong network that can accelerate Obsydian Club’s growth, refine our business model, and help us scale rapidly in the nightlife and social discovery ecosystem.” |
“During the Avinya programme, we aim to refine our business model, strengthen our go-to-market strategy, gain mentorship from industry experts, validate our product-market fit, and accelerate user acquisition and revenue growth for Obsydian Club.” |
“Yes, we are open to both hybrid and virtual participation, as it allows us to fully engage with Avinya’s mentorship, resources, and network while ensuring flexibility for our team’s operations.” |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/uBPwAet7tLrA-Obsydian-club-PITCH-DECK-2_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/uBPwAet7tLrA-Obsydian-club-PITCH-DECK-2_compressed.pdf |
https://www.instagram.com/reel/DPWHG79ihqZ/?igsh=dHRicGNhNXhjZXBl |
“Obsydian Club is a mission-driven startup focused on transforming the nightlife and social discovery experience. Our goal is to create safe, seamless, and engaging nightlife experiences, helping people connect, explore, and enjoy social interactions responsibly while supporting local clubs and event ecosystems.” |
“No, we are not part of an underrepresented group in tech/startups, but we are committed to building an inclusive platform that connects diverse communities and promotes safe, engaging social experiences for all users.” |
“Referred by i3 Foundation, IIT BHU, Varanasi.” |
checked |
| Oct 3, 2025 @ 4:12 PM |
Yash Meena |
yashcodes105@gmail.com |
https://www.linkedin.com/in/yash-meena-3b0748203 |
http://NA |
7067935654 |
CEO and IITBHU |
Brothers,forever |
Yes |
Aim is same , both come from same background.
Both want to solve for farmers |
KishanMitra |
NA |
|
Helping Farmers |
One stop solution for farmers , buy fertilizers ,seed and sell crops . Connect with farmers community. Personalised attention to your crops . Crop doctor all in a single app |
Single app for all farmers need . building a tech based solution .
App is half complete which supports : weather , news , mandi prices currently working on crop doctor and community support. |
Very less players in agri field. Having tech background can build it quickly once get funded |
MVP |
Users |
Farmers |
120M |
62M |
1.5M |
1.. Freemium + Subscription – Free basic services, paid premium advisory & insights (₹99–299/month). 2. Marketplace Commission – Earnings from sale of seeds, fertilizers, pesticides, and equipment via the app (5–15% margin). 3. Market Linkage Fees – Commission (1–5%) on connecting farmers with buyers, mandis, FPOs, and exporters. 4. Financial Services Partnerships – Referral/commission from banks, NBFCs, and insurers for loans, insurance, and credit scoring. 5. Ads & Sponsorships – Revenue from curated advertisements by agri-input companies and government/NGO projects. |
Dehaat |
1. Village-Level Partnerships – Collaborate with FPOs, cooperatives, local agri-dealers, and NGOs to onboard farmers in bulk. 2. On-Ground Agents (“Mitra Model”) – Train local youth as KishanMitra agents who demonstrate the app, build trust, and provide first-hand support. 3. Referral & Incentives – Farmer-to-farmer referral program (e.g., free premium days, input discounts for inviting friends). 4. Digital Outreach – Use WhatsApp groups, YouTube agri-influencers, and regional-language content to build awareness. 5. Government & Institutional Tie-ups – Partner with state agriculture departments and schemes to distribute the app during training camps or subsidy programs. |
1. Pilot in Key States – Launch in 2–3 agri-heavy states (e.g., UP, Bihar, Maharashtra) focusing on small & marginal farmers. 2. On-Ground Network – Partner with FPOs, cooperatives, and local agri-dealers; deploy trained “Mitra” agents for farmer onboarding. 3. Digital Outreach – Leverage WhatsApp groups, vernacular YouTube influencers, and regional content for awareness at scale. 4. Partnership Ecosystem – Collaborate with agri-input companies, banks, and government/NGOs to boost trust and adoption. 5. Monetization Rollout – First focus on adoption, then add premium subscriptions, marketplace commissions, and fintech services in phases. |
1. One-Stop Digital Platform for Farmers – Become the go-to app for advisory, inputs, finance, and market linkage.
2. Financial Inclusion at Scale – Enable affordable credit, insurance, and savings products tailored for farmers.
3. Data-Driven Agriculture – Use AI + farm-level data to provide predictive insights on crop health, yield, and pricing.
4. Global Expansion – Extend the model to other agri-heavy developing countries (Africa, SE Asia) facing similar farmer challenges.
5. Farmer Prosperity & Sustainability – Improve farmer income, reduce risk, and promote sustainable farming practices for the long run. |
“Pre-incorporation, in prototype / MVP development phase |
NA |
Yes |
1. Scaling Support – We want to leverage Avinya’s ecosystem to refine our model and scale KishanMitra from a pilot to a pan-India solution. 2. Mentorship & Expertise – Access to domain experts in agri-tech, rural markets, and fintech will help us avoid common pitfalls and grow faster. 3. Access to Networks – Avinya’s connections with investors, corporates, and government bodies can accelerate farmer onboarding and partnerships. 4. Technology & Innovation Guidance – Support in strengthening our AI-driven advisory, multilingual platform, and secure fintech integrations. 5. Impact Alignment – Avinya’s mission to enable impactful startups aligns with KishanMitra’s vision of improving farmer income and sustainability |
1. Refine Business Model – Validate and sharpen KishanMitra’s revenue streams (subscription, marketplace, fintech) with expert guidance. 2. Strengthen Product-Market Fit – Pilot with farmers in select regions, gather user feedback, and improve adoption/retention. 3. Build Key Partnerships – Connect with FPOs, agri-input companies, financial institutions, and government bodies to expand reach. 4. Investor Readiness – Develop a strong pitch deck, financial model, and fundraising strategy for scaling post-programme. 5. Impact Measurement – Establish clear metrics to track farmer income growth, crop yield improvement, and adoption rates. |
Virtual |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/iqEt6QhLDeho-KishanMitra-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/iqEt6QhLDeho-KishanMitra-Deck.pdf |
https://drive.google.com/file/d/1DXQFh0sojIxK9GJ8a6bteVCPBDcvCI1H/view?usp=drivesdk |
Yes — KishanMitra is a mission-driven, impact-focused startup. Our core mission is to empower small and marginal farmers, who form the backbone of India’s agriculture but often face low incomes, lack of reliable information, and limited market access.
We are building a platform that:
1. Increases farmer income by connecting them directly to markets and buyers.
2. Reduces risk through AI-powered crop advisory, weather alerts, and disease detection.
3. Improves access to affordable seeds, fertilizers, and financial products.
4. Promotes sustainability by encouraging data-driven, resource-efficient farming practices.
5. Drives inclusion by supporting multiple Indian languages and local farmer networks.
Our long-term vision is not just commercial success, but to make farming more profitable, resilient, and dignified for millions of smallholder farmers in India and beyond. |
Na |
IIT BHU |
checked |
| Oct 1, 2025 @ 4:44 AM |
ARUN PRASATH MUTHAIAH, VAIBHAV DAYANAND UCHIL, PRAVIN RAJAKUMARAN, SATISH KUMAR |
arun@pangun.tech |
http://www.linkedin.com/in/arun-prasath-muthaiah |
https://pangun.tech/ |
9980929590 |
CEO, IIT alumnus with over a decade of expertise in marketing, branding , business growth strategic positioning and ERP product development & consulting for the manufacturing and supply chain industry. |
All the co-founders have known each other for nearly half a decade, with some bonds going back much further. The CEO and the Co-founder (Tech & Product) share a friendship spanning over 15 years, having studied together during their school days. Meanwhile, the CEO and the Co-founder (Marketing) previously worked closely in a professional setting, building a strong foundation of trust and collaboration before coming together to launch the organization. The Founder and team members have been working full-time for the past 9 months, currently the organisation is in Revenue generation stage on product marketing, branding, customer projects, product customization, product implementation, operational revenue, financial models and annual operating plans. We have been conducting product demos for customers and recruiting talent for technology and strategic marketing for the upcoming projects. Our team has dedicated 36 months ( from 2021) to developing the product, including configuring setup, conducting market surveys, and utilizing business and customer test cases. We have focused on ensuring product-market fit, refining product pitches, showcasing company capabilities, and developing the business model and annual operating plans. Additionally, we have promoted the product at manufacturing and supply chain technology events, delivered investment pitches, and conducted customer demonstrations. |
No |
Our biggest strength as a team lies in the values we uphold — integrity, responsibility, resilience, and transparency. These principles guide the way we collaborate as co-founders and with our wider team, ensuring clarity and trust in every decision. They have helped us navigate challenges, including building sustainable business models even during periods without regular revenue flow. Moreover, our deep focus on customer satisfaction and a well-structured delivery process not only enables us to win new projects but also fosters long-term relationships, ensuring customer retention and growth. |
Pangun Infosol Pvt Ltd |
https://pangun.tech/about/ |
|
Streamlining manufacturing and enterprise operations with i AI-integrated ERP and ISO reporting tools |
Our startup addresses critical challenges in manufacturing and enterprise operations, including operational leakages caused by the absence of a centralized supply chain, manufacturing, planning, inventory, shipping tool. We solve the complexities in managing multilevel manufacturing with intricate BOM and sub-assembly data. Stockouts occur frequently due to improper inventory reorder points. Additionally, there is a significant gap in standardized ISO reporting and KPI templates, hindering compliance with global safety and quality standards. Finally, we tackle inefficiencies in product safety and quality management through seamless ISO report generation and AI/ RAG integrated Reliability Tools |
The integrated SCM system connects procurement, manufacturing, sales, and finance departments to ensure seamless data flow and minimize miscommunication risks. We offer robust manufacturing configuration alongside comprehensive financial management modules to track expenses, manage budgets, and streamline billing and payments. Precise inventory management tools enable accurate tracking and reduce stockouts. Our solution features an inbuilt ISO audit tool that simplifies generating ISO reports and KPIs essential for AS9100, IRIS, and ISO 9001 certifications.. Advanced AI/ RAG integrated manufacturing tools support FMEA, RAMS, FRACAS, LCC, RCA analysis for quality intelligence. Together, these capabilities empower organizations to achieve operational excellence and compliance with global standards. |
Our solution stands out due to its comprehensive integration of ERP, ISO audit software, and advanced reliability tools, including FMEA, RAMS, FRACAS, LCC, and RCA analyses, within a single platform. We are pioneering the development of a "Multi-Modal Agentic RAG Platform for Enhanced Product Risk & Reliability Management," leveraging cutting-edge GenAI technology specifically tailored for aerospace, defense, railway, and UAV sectors. This AI-powered FMEA and RAMS solution incorporates our deep expertise in AI/ML, MLOps, cloud computing, and data analytics, delivering unmatched efficiency and precision. Such an integrated suite of tools is exceptionally rare in both the Indian and global markets, providing us with a strong competitive moat and defensible market position. |
Revenue |
Revenue |
Our target customers include SMEs, OEMs, and enterprises across manufacturing and related industries. We primarily focus on sectors such as aerospace, defense, railways, oil & gas, industrial machinery, and other large-scale enterprises. Geographically, our focuses are concentrated in India, UAE, France, Canada, and the United States, where we aim to address complex operational and compliance needs through our integrated solutions. |
Our Total Addressable Market (TAM) spans key regions with significant SME presence and revenue generation potential. In India, there are approximately 1.5 million SMEs generating $200 billion in revenue. The Middle East, including UAE and other countries, hosts around 1.92 million SMEs with a combined revenue of $400 billion. France and Europe account for 13.365 million SMEs, producing $2.42 trillion in revenue. In North America, covering the US and Canada, there are 20.39 million SMEs generating $5.58 trillion in revenue. This diverse and expansive market forms the basis for our extensive growth and customer acquisition strategies |
Our Serviceable Addressable Market (SAM) represents a scalable 50% share of the total market, factoring in planned investments in marketing, recruitment, and operational capacity. Specifically, this includes: India with 1.5 million SMEs generating $200 billion in revenue, Middle East (UAE and others) with 1.92 million SMEs and $400 billion revenue, France and Europe with 13.365 million SMEs generating $2.42 trillion, North America (US & Canada) with 20.39 million SMEs and $5.58 trillion in revenue. By targeting this 50% realistic market penetration, we aim to effectively capture a substantial and achievable portion of these diverse and high-value markets. |
Our Serviceable Obtainable Market (SOM) reflects the realistic portion of the market we can capture with our current resources, investment, marketing efforts, and recruitment plans. We estimate scalability for approximately 10% of the total market share across key regions such as: India with 1.5 million SMEs generating $200 billion in revenue, Middle East (UAE and other countries) with 1.92 million SMEs and $400 billion revenue, France and Europe with 13.365 million SMEs generating $2.42 trillion, and North America (US & Canada) with 20.39 million SMEs generating $5.58 trillion in revenue. This focused market capture strategy targets achievable growth aligned with our business capabilities and market demand |
Our revenue model comprises multiple streams aligned with our product offerings. For our ERP, ISO audit software tool, and reliability solutions, we generate revenue through implementation charges, monthly subscription fees, and customization charges tailored to client needs. For our upcoming "Multi-Modal Agentic RAG Platform for Enhanced Product Risk & Reliability Management," we offer a freemium model—providing a free subscription for standard users and a monthly subscription plan for premium features. This hybrid approach allows us to cater to diverse customer segments while ensuring scalable and recurring revenue growth. |
Our main global competitors in the ERP and ISO audit software space include industry leaders such as IFS, Infor, Epicor, SAP, and Oracle. In the Indian market, notable competitors are Grow Smart, Tally, and Zoho, among others. For reliability tools covering FMEA, RAMS, and related analyses, global competitors include PHM Technologies’ MADE, Dassault Systèmes 3DEXPERIENCE (MODSIM, SIMULIA), IFS for Aerospace & Defense, RAM Commander, and Relyence. In India, BE Analytics is a key competitor in the reliability tools segment. |
We acquire customers through targeted branding activities including participation in tech expos, LinkedIn branding, SEO-driven content marketing, and leveraging referrals. Once leads are generated, our marketing team engages in a multi-phase process involving detailed customer requirement analysis, personalized pre-sales pitches, and real-time product demos tailored to each customer's specific business case. This structured approach ensures a high-touch engagement that builds trust and converts prospects into long-term customers. |
Our go-to-market strategy is rooted in comprehensive primary and secondary market research to deeply understand customer needs and market dynamics. We focus heavily on branding through tech expos, LinkedIn engagement, SEO-driven marketing, and referral networks to generate leads. Once leads are identified, our marketing team executes a multi-phase engagement process that includes detailed customer requirement analysis, personalized pre-sales pitches, and real-time product demos tailored to each customer's specific business case. This approach ensures high relevance, builds trust, and accelerates customer acquisition. |
Our long-term vision is to create a seamless, integrated AI and RAG-driven ERP, ISO audit, and reliability tool ecosystem. Central to this vision is the development of our "Multi-Modal Agentic RAG Platform for Enhanced Product Risk & Reliability Management," a GenAI-powered FMEA and RAMS solution tailored for aerospace, defense, railway, and UAV applications. Leveraging our expertise in AI/ML, MLOps, cloud computing, and data analytics, we aim to lead innovation in AI-driven reliability engineering. While we are currently progressing with synthetic datasets, access to real industry data and domain expertise will be crucial for advancing innovation and scaling our solutions. Ultimately, this initiative positions us to become the market leader in delivering significant value to safety-critical industries worldwide. |
Incorporated |
NA |
Yes |
Apart from seeking funding, we are applying to Avinya to gain strategic support for market expansion in key regions including the UAE, France, Japan, and Canada. We aim to leverage Avinya's expertise in go-to-market strategies to accelerate our entry and growth in these markets. Additionally, we require access to a tech incubator space that can support our team’s innovation efforts, complemented by funding allocations for critical resources such as GPUs, cloud infrastructure, systems, and human resources. This comprehensive support will enable us to scale effectively and enhance our technological capabilities. |
We are applying to Avinya not only for funding but also to gain essential support for expanding our market presence in the UAE, France, Japan, and Canada. Avinya’s structured mentorship and global network will provide us with strategic guidance and connections critical for executing projects in these regions. During the programme, we aim to secure opportunities to implement projects across these key markets and target generating $100 million in revenues within the next two years. Given the uniqueness and scalability of our product portfolio across aerospace, defense, railways, and oil & gas sectors, we believe Avinya’s ecosystem will accelerate our growth trajectory significantly. |
We are open for hybrid participation. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/h0picPJ0PaJ8-PanGun-Pitch-Deck-Sep-2025.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/h0picPJ0PaJ8-PanGun-Pitch-Deck-Sep-2025.pdf |
https://www.youtube.com/watch?v=5RDUysgRCHw |
Mission:
Our mission is to empower enterprises in both traditional and next-generation manufacturing and mobility sectors by delivering affordable, AI-integrated ERP, compliance, and reliability management solutions. These tools not only enable effective data and business process management but also help our customers attain critical certifications such as AS9100, IRIS, API Q1, and ISO. This certification support enables our customers to qualify as suppliers to major OEMs like Boeing, Airbus, HAL, NAL, Alstom, and Armco. Our solutions are designed to improve operational efficiency and simplify complex tasks.
Vision:
We aim to become a global leader in super AI-integrated enterprise software solutions, making advanced technology accessible to businesses across all industries. |
Yes |
Abhisek , Asst Professor, XLRI, India |
checked |
| Oct 1, 2025 @ 12:26 AM |
Biboswan Roy |
biboswan98@gmail.com |
https://www.linkedin.com/in/biboswanroy |
https://www.bibotecture.com/,%20https://biboswanroy.com |
+918820249026 |
Role: Tech Founder and CEO. Background: I have been a Senior Frontend Engineer at cybersecurity company SentinelOne Inc, where I led impactful products. Worked in various startups. I was recognised as a valuable open source Contributor at Mozilla and invited to a sponsored trip to their annual conference in Orlando, FL, USA. I have over 5 years of experience in Software Engineering building fullstack applications end to end. |
Working as a solo founder. Finding the right fit for a co-founder. |
Yes |
Having lived across multiple countries, I’ve gained firsthand insight into the diverse communication barriers that emerge in dating, friendships, and business. This experience—my ‘pseudo-PhD’ in communication and linguistics—uniquely equips me to design and build the product I’m working on. |
ChatPsyche Coach |
NA |
|
Realtime Chat Coach that improves conversations with psychology and EQ integrable like Grammarly across any chat platform. |
People often struggle to understand the emotional and psychological nuances behind what others try to convey and also their own self awareness. Current tools just provides robotic reply suggestions but don’t work on improving person’s emotional intelligence. They don’t consider the context of the conversation, the dynamics of the relationship and their individual backgrounds. As a result, communication can feel awkward, ineffective, or misaligned or maybe lead to unwanted conflicts. |
Our AI-powered Chat Coach helps people communicate clearly, empathetically, and authentically while preserving their personal style and improve their emotional intelligence. Available as a browser extension, desktop app, and mobile app, it can auto-integrate across any chat platform — Slack, WhatsApp, Instagram, Dating apps or others — providing real-time suggestions and the underlying reasons behind it.
Key features include:
* Style-preserving suggestions: Improves tone, clarity, and impact while keeping your unique voice intact.
* Context-aware guidance: Understands relationship dynamics, emotional cues, personal background, and the preferences of the person you’re communicating with (e.g., technical vs non-technical).
* Tutor clones & marketplace: Communication experts can create AI clones of themselves, allowing users to crowdsource communication styles and personas. These clones auto-train as the expert interacts, capturing subtle style patterns.
* Social integration: Connect and merge conversations across social media to maintain consistent communication.
* Split-screen mobile experience: Enables simultaneous chat and coaching on Android (testing in progress).
Think of it as “Grammarly for communication” — but for both consumers and B2B, helping users improve not just grammar, but persuasion, clarity, empathy, and style.
* Self-awareness: Recognizing and understanding your own emotions and how they influence your behavior.
* Self-regulation: The ability to manage your emotions and control your actions, especially in stressful situations.
* Empathy: Understanding and sharing the feelings of others.
* Relationship Management: Using emotional understanding to build strong relationships, communicate effectively, and resolve conflicts |
Current tools provides robotic reply suggestions but don’t work on improving person’s emotional intelligence. They don’t consider the context of the conversation, the dynamics of the relationship and their individual backgrounds and the goal of the conversation. They don’t work on long term relationship management. Our platform builds a personal fingerprint of your communication and psychology We will provide real-time holistic coaching experience by being at your side across any conversation across any chat communication platform. It will focus on fundament aspects of communication comprehension, conciseness, biaseness etc. |
MVP |
Testimonials |
Anyone who uses any Conversational Platform both for consumers and B2B |
4.5B messaging users globally (~$42B/year) |
English-speaking professionals in North America & Europe (~$10.5B/year) |
Initial realistic adoption ~60k paying users (~$7.2M ARR) |
Freemium AI chat coach with subscription tiers for higher usage limits and advanced style insights, plus optional marketplace purchases for personalized coaching clones and style packs |
Grammarly and Meta AI |
We acquire customers through targeted digital marketing, partnerships with professional communities and coaches, content marketing around personal growth and communication, and seamless freemium onboarding that encourages organic referrals. |
Acquire users via freemium apps that auto-integrate with chat platforms, grow organically, and monetize with subscriptions, teams, and future marketplace. |
Chat Coach Clones Marketplace: Communication experts can create AI clones of themselves, allowing users to crowdsource communication styles and personas. These clones auto-train as the expert interacts, capturing subtle style patterns. |
Bibotecture Ltd |
NA |
Yes |
Grant will help me stay focused and committed to my vision full time. Learn as much as I can from the accelerator program and navigate in the right direction. |
Find GTM and launch the first beta version |
Hybrid preferred open to both. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/YzCrHgjRuwwJ-ChatPsyche_Coach_PitchDeck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/YzCrHgjRuwwJ-ChatPsyche_Coach_PitchDeck.pdf |
https://drive.google.com/drive/folders/19OVwMiwVQ4rYsOg6W5NCWsS-XOfveHQ-?usp=drive_link |
Our mission is to reduce misunderstandings, bridge cultural and interpersonal gaps, and elevate emotional intelligence by building stronger personal and professional relationships worldwide. We recognize that each person’s personality, beliefs, experiences, and subconscious mind shape how they think, communicate, and perceive others. Our platform harnesses these insights to help people navigate conversations with empathy, clarity, and authenticity, while keeping their unique communication style intact.
Examples of Impact:
* Helping a sales manager craft messages that land with empathy across different cultures.
* Guiding a new graduate to communicate confidently with their boss on Slack.
* Helping someone navigate early-stage dating conversations with clarity and respect.
* Supporting cross-functional teams to avoid miscommunication when stakes are high.
* Noticing when a user’s tone turns unusually aggressive and gently prompting them to reflect on possible stress or emotional triggers. |
No |
Kamlesh Kumar IITACB (unokamlesh@gmail.com) |
checked |
| Sep 30, 2025 @ 8:42 PM |
Venkatesh Vadde; Kenneth Joel; Mohan Prabhakar |
vvadde@sensio-ai.in |
https://linkedin.com/in/vvadde |
https://sensio-ai.in |
9945715725 |
Dr. Venkatesh Vadde (B.Tech, IIT-Madras; MS IISc-Bangalore, PhD- Carnegie Mellon all in ECE disciplines, 25 years experience) is the Co-founder & CEO; he takes charge of overall Tech, Finance, IP, B2B marketing, and Strategy. Mr. Kenneth Joel (B.Tech, in EEE from PES University; 7 years experience) is the Co-founder & Director of SW; he looks after SW development, Algorithms, B2C marketing, bizdev, HR and Compliance; Mr. Mohan Prabhakar (B.Tech, in EEE from PES University; 7years experience) is the Co-founder and Director of HW; he is in charge of HW design, Prototyping, Design of experiments, Data collection, Certifications and Accounts. |
We all met while in the academia. The principal co-founder was a Professor, while the other two, Kenneth and Mohan, were blazing a trail in hackathons and class projects achieving things well beyond the typical brief. We were from different departments, never met in a classroom, always in a coffee place. Sensors, signals and solutions drew us together; we loved solving good problems with these ingredients. We came together to do a capstone project, their final year project. We saw real potential in a product startup. We decided to take the plunge because all of us had the startup bug in good measure. |
Yes |
Our biggest strength as a team is in the depth as well as breadth of skills we encompass.
Technical: We can tame a technical problem spanning both HW and SW. Our insights across these domains enable us to dimension a solution for a wide range of problems with sensors and AI.
Market: we are a cross-gen team that understand the market from a needs as well as a wants perspective. We balance our offerings and marketing campaigns with this in mind.
Chemistry: we have a value system that binds us together, we have known each other for close to a decade now; our shared values and love for India inspires us to build in India for the World. We love the work we do. |
Sensio Enterprises Pvt Ltd |
https://sensio-ai.in |
|
We are India's 1st deeptech wearables ODM creating original solutions |
Health is in short supply, the world is stressed and losing sleep. We are creating wearable solutions to measure and improve health, wellness and safety. |
We have developed sensor-enabled wearable solutions using 3 form-factors: a smart ring, a smart band and a chest patch. These are sensor-enabled smart wearables that collect human vitals and help people quantify health. As an ODM, our superpower lies in customizing any of these wearables to a particular use case. We use medical grade sensors, making our wearables accurate and reliable. Our smart ring has 4 sensors (ECG, PPG, Temp, IMU) when most in the market have just 3 (PPG, Temp, IMU). Our USP is in access to raw data, ultra-low power design, edge-AI algorithms and customization. We offer ground-up designs that match or exceed China in cost and capability. |
We have our unique IP and recipe to make our vital signs monitoring wearables entirely in India, we own our design and build the product hands-on. Our capabilities are also unique, and we do not see a peer like us in the wearable ODM space. |
Revenue |
Revenue |
We are an ODM, our target customers are businesses with wearable needs. They can be OEMs, service providers who use wearables, research groups etc. |
25Bn USD |
15Bn USD |
5Bn USD |
We customize wearables and sell them in MOQ volumes. We can also charge for algos licensed to the customer. |
Our main competitors would be other ODMS, mainly from China |
We present ourselves in B2B expos, display our wears and also do Linkedin digital marketing. These channels give us leads, which get qualified over B2B discussions and interactions leading to sales. |
We keep refining our focus on wearables. We have clients who have an order book approaching 70-80K units of wearables. Our current task is to execute with focus and keep delivering on those orders as we build the pipeline. We keep adding 1-2 new customers every month. |
Our long term vision is to be a unique sensors and AI driven company that can build on our solid foundation. We have a multi-wearable vision for the future, and we believe people will use multiple wearables to quantify human health and achieve longevity via preventative health. We wish to make Karnataka the base for a "build in India for the World" play in wearables. We want to be the top name that comes to mind anywhere in the world for accurate, research grade, customized solutions. We also want to play our bit in making India a healthier nation. |
We are incorporated as a private limited company |
We have raised over Rs.5.5Cr so far of which Rs.55L has been as govt grants in our early years. |
Yes |
There was a vibe that attracted us when we heard it is IITians backed and also has a Karnataka backing. |
We want to de-risk our GTM and scale-up so we can keep growing and riding on the global health opportunity. We also want to quickly raise funds to do the same. We hope our traction and progress will convince you. |
Yes, we are open to a hybrid or virtual participation model. However, we are very ready for an infusion of capital, so please consider accelerating our cycle. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/80gkHuw5Afi5-Sensio-ODM-pitch-deck-brief-Aug2025.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/80gkHuw5Afi5-Sensio-ODM-pitch-deck-brief-Aug2025.pdf |
https://drive.google.com/file/d/1Tx6pwK7ZShRjSrYPXAOXvjwKeNeygdV2/view?usp=sharing |
We are building a mission driven startup. We are playing the slightly longer game here. However, we will carve out a way to find an exit for investors based on our smart ring IP which we think is the hardest to replicate and unique with value in both health and wellness. |
Not in a manner that is typical; we are mainstream guys but uniquely different in our own ways. |
Mr. TR Anand, TIE Chartered Member |
checked |
| Sep 30, 2025 @ 8:07 PM |
Naveen Kumar |
naveen@hyperalive.in |
https://www.linkedin.com/in/naveen-choudhary-29902a224/ |
https://hyperalive.in/ |
7905445367 |
Founder — Naveen Kumar Education: M.Tech (Design Engineering), NIT Allahabad, 2019. Experience: ~8 yrs across machine design, FEA/simulation, systems research, design innovation, and full-stack software development. Ex-GS1 India. Co-Founder — Abhishek Mahajan Education: B.Tech (Biotechnology), Lovely Professional University, 2020. Experience: ~4 yrs in medical devices; exposure to clinical workflows and regulatory basics. Co-Founder — Prashant Singh Education: B.Tech (Mechanical), AKTU Lucknow, 2017. Experience:~7 yrs Machine design, CAD/FEA & simulation; BOM & manufacturing; ex-Hungama Tech; full-time on HyperAlive. |
Founders’ chemistry & timeline
Naveen & Prashant: college friends for ~13 years; co-building HyperAlive from inception (PoC → V2.3), working hands-on across design/FEA and fabrication.
Abhishek: professional colleague who joined later; we’ve worked together through current R&D, clinical workflow, and GTM. |
Yes |
13-year founder trust + complementary deep-tech skills (design/FEA, mech/controls, software) with clinician advisors. We execute fast and frugally—prototype, validate, and iterate to clinic-grade, export-ready products. |
HyperAlive |
https://hyperalive.in/ |
|
HyperAlive is an AI-enabled HBOT platform with dual-shell safety delivering hospital-grade care at 45–60% lower CapEx, with built-in telemetry and EMR logs—available via purchase or performance lease. |
India faces a rising burden of chronic and acute conditions where hyperbaric oxygen therapy (HBOT) is a proven or emerging treatment—but it remains inaccessible due to cost, safety, and infrastructure barriers. HBOT supports: diabetic foot ulcers (100M Indians with diabetes at risk; leading cause of amputations), stroke and traumatic brain injury/neuro-rehab, non-healing wounds, burns, radiation injuries, post-COVID fatigue, and sports injuries. Yet hospital-grade chambers cost ₹45 lakh–₹90 lakh and are mostly in metros; sessions run ₹3,000–₹5,000. Cheaper soft chambers often fail safety norms and are banned in many countries. HBOT is clinically powerful but logistically broken in India—HyperAlive exists to fix this. |
HyperAlive is a state-of-the-art, next-generation, AI & IoT-enabled HBOT platform for chronic conditions (diabetic wounds, radiation injury, stroke rehab) and longevity/anti-ageing protocols. Our dual-shell, load-sharing design lowers wall stress (~60%), delivering cutting-edge affordability (CapEx ↓45–60%) with higher safety.
Semi-autonomous control: closed-loop ramps/holds; safety stack (BPR, dual PRVs, ΔP limiter, emergency dump).
Digital layer: real-time telemetry, clinician dashboard, remote alerts, and auto PDF/EMR logs with protocol presets.
Lines: Soft (≤1.5 ATA) for clinics/sports; Hard (≤3 ATA) for hospitals — turnkey install, training, support.
Commercial: Direct sale + software or Performance Lease (per-session; try-before-buy; upgrade Soft → Hard).
Export-ready: PVHO → CE pathway; India-first supply chain for rapid rollout in MENA, Africa, SE Asia, LATAM. |
Architecture edge: Dual-shell, semi-autonomous HBOT delivers clinic-grade safety at 45–60% lower CapEx—hard to copy without our tuning. Data moat: IoT telemetry + EMR logs → protocol optimization, sticky SaaS, compounding outcomes data. Regulatory & IP: PVHO→CE track, ISO-aligned QMS, provisional patents + control-stack know-how raise barriers. GTM rails: Performance lease and Soft→Hard upgrade ladder reduce friction and lock in lifetime value; India-first manufacturing boosts speed & margins. |
MVP |
Pilots |
Primary customers: Anti-aging/longevity clinics and therapy centers, physiotherapy/sports-medicine chains, wound-care & plastic-surgery units, neuro-rehab centers, and small/mid-size hospitals—especially in tier-2/3 cities—that can’t justify ₹45–90L HBOT capex or lack monitored, reportable dosing. Economic buyers & champions: Clinic owners/administrators (P&L), heads of rehab/wound care/sports medicine/diabetology, and hospital biomedical teams. End users (patients): Middle-class adults seeking recovery and longevity—post-surgery/orthopedic rehab, chronic or diabetic wounds, radiation tissue damage, sports injuries, and performance/healthspan optimization—who are priced out of current HBOT or offered low-fidelity substitutes. |
5000 cr |
3000cr |
400cr |
Device sales: Soft (≤1.5 ATA) and Hard (≤3 ATA) units. Performance Lease: low MMG + per-session fee; try-before-buy; upgrade Soft → Hard. Software (SaaS): telemetry/EMR dashboard, tiered plans. Services: AMC (~3%/yr), training, accessories/spares. Exports: distributor margins + service packs. |
Oxycea, Baromedic healthcare (India), macy-pan (china), OxyHealth (USA) |
Segments: wound/DFU, oncology, neuro, physio. Channels: founder-led enterprise + distributors. Motion: demo → pilot (performance lease) → convert to sale/lease → upsell SaaS/AMC & Soft→Hard. Proof via telemetry-backed outcomes. |
Lead with a try-before-buy Performance Lease (MMG+per-session) to run reference pilots, then convert to sales via founder-led enterprise + regional distributors with NBFC financing, upsell SaaS/AMC and Soft→Hard, and replicate in export clusters. |
Make HBOT as routine as imaging—affordable, safe, data-driven. Build an AI-monitored platform (Soft ≤2 ATA, Hard ≤3 ATA) for chronic care and longevity, scale across India, then export via distributors, with recurring SaaS and growing clinical evidence. |
Incorporated in India as Secondbrain Ventures Pvt Ltd (Private Limited). CIN: U63999UP2023PTC185461. Status: Active and compliant. |
Non-dilutive grants ₹7.5 lakh committed (₹3.0 lakh disbursed via StartInUP; ₹4.5 lakh in pipeline incl. BFI Biome). Equity: ₹0. Debt: ₹0. Bootstrapped by founders to PoC/bench validation/pilot-ready stage. |
Yes |
To fast-track HyperAlive’s commercialization: seed capital, regulatory (ISO 13485, PVHO→CE) guidance, lab access, hospital pilot connects, NBFC/EXIM links, and mentor/distributor networks—to launch Soft HBOT and advance Hard HBOT R&D for export. |
Commercialize Soft HBOT to TRL-8 with 2–3 pilots; stand up ISO 13485 and PVHO→CE pre-compliance; complete DFM/DFx & vendor lock-ins; launch AI/IoT telemetry v1; secure NBFC lines and roll out Performance Lease; close seed & export MoUs. |
Hybrid or virtual both |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/4CQ6Dcwe5xhj-HyperAlive-Avinya-pitch-deck_final_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/4CQ6Dcwe5xhj-HyperAlive-Avinya-pitch-deck_final_compressed.pdf |
https://www.youtube.com/watch?v=M2_PrI7oenM |
Yes—mission-driven. We democratize HBOT for chronic care and longevity: clinic-grade therapy at 45–60% lower CapEx, reaching tier-2/3 India. Impact: fewer amputations, faster healing, lower OOP costs, better outcomes. Made-in-India, export-ready. |
Yes. We’re first-generation, non-metro founders building deep-tech medical hardware—an underrepresented segment in India’s startup/VC landscape. We’ve been bootstrapped, lack legacy networks, and operate in a capital-intensive, regulated category that’s traditionally overlooked. |
self-applied |
checked |
| Sep 30, 2025 @ 6:20 PM |
Robin Singh , Chitranjan Singh |
robin@cluix.in |
https://www.linkedin.com/in/robin-singh-b9952028 |
https://www.cluix.in/ |
09560589782 |
Chitranjan Singh (Founder & Chairman)- Chitranjan Singh boasts an impressive career spanning over 30 years. He graduated with a degree in Civil and Structural Engineering from the College of Military Engineering in Pune in 1995. Since then, he has served as a distinguished bureaucrat within the Indian defense sector. His extensive experience covers numerous projects of national significance across India. Notably, he is known for his creative problem-solving skills and his proven track record in executing large-scale projects Robin Singh (Co-Founder) - Robin Singh is a visionary entrepreneur and technical innovator with a strong background in Electrical and Electronics Engineering and over a decade of experience driving impactful technological solutions. He specializes in automation, embedded sensors, electrochemical applications, and advanced machine design, with hands-on expertise in software development (C, C++, Python, Java) and microcontroller and microprocessor-based solutions. Under his leadership, CLUIX developed a Hand-held IoT Digital Water Quality Analyzer that empowers rural communities, Gram Panchayats, and industries to conduct real-time water quality assessments. His commitment to sustainable innovation was recognized with the Start-up India Challenge award, and he secured more than INR 2crore grant from Invest India under the Jal Jeevan Mission and several other government and international schemes. |
As a father–son duo, our journey has always been intertwined. Robin has been deeply inspired by his father, Chitranjan, whose guidance encouraged him to experiment, innovate, and build technical skills from a young age. This mentorship shaped Robin’s professional and personal growth, and together we formally started working on CLUIX over 3 years ago, combining experience with new-age innovation. |
Yes |
Strong family trust and alignment, complemented by a mix of technical expertise and business strategy, enabling rapid decision-making and innovation. |
CLUIX Private Limited |
https://www.cluix.in/ |
|
CLUIX is a cleantech deeptech startup, who has developed IoT-enabled water technology providing real-time, field-ready water quality monitoring solutions. |
In India, over 70% of water sources are contaminated, yet testing remains centralized, costly, and slow, leaving rural and underserved communities without timely information. Traditional lab testing takes weeks, creating a dangerous gap between contamination and corrective action. This lack of accessible, reliable, and real-time monitoring fuels waterborne diseases, loss of trust in public supply, and unnecessary spending on bottled water. CLUIX addresses this by enabling easy to use, portable, rapid, and affordable testing at the source, ensuring that communities, industries, and end-users can make immediate, data-driven decisions to safeguard health. |
Our innovation integrates advanced sensors, IoT connectivity, and data analytics into a compact water quality analyzer. The device measures multiple critical parameters in real time and transmits results instantly to a digital dashboard for users and governing bodies. This decentralized approach removes dependency on distant labs, while embedded AI-driven analytics detect anomalies and predict contamination trends with historical data. By combining portability with digital traceability, technology ensures transparency, trust, and scalability. It bridges the gap between rural communities, government programs, and industries by delivering actionable insights for safe water management. |
Portable, field-ready lab-grade accuracy Multi-parameter testing in one device Smart connectivity for real-time monitoring Scalable and adaptable for local and international water systems Proprietary combination of hardware and software algorithms |
Product-Market Fit |
Users, Revenue, Pilots |
• Government departments (PHED, Jal Nigam, Health, and Rural Development) • CSR and foundations working in WASH and public health • Water utilities and private operators • NGOs and international development agencies focused on last-mile delivery • Enterprises offering water as a service or running decentralized treatment plants |
46.04 Bn globally / $2.38 Bn in India — representing all possible use cases of water quality monitoring solutions. |
$2.04 Bn — focused and realistically addressable Indian market. |
$40.9 Mn – $102.25 Mn — the market share CLUIX could realistically capture. |
B2B & B2G |
Elico, Hanna, Hach, Xylem |
Through B2B sales, government procurement channels (GeM), CSR partnerships, and direct outreach to water authorities. |
Outbound Sales & Partnerships Collaboration and PR Marketing Online Marketing Event Marketing |
Our vision is to make safe, real-time water quality monitoring accessible globally, directly contributing to SDG 6 by enabling clean water and sustainable water management for all. By ensuring proactive monitoring and early detection of contamination, we also advance SDG 3 by reducing waterborne diseases and promoting healthier communities worldwide. |
private limited |
234.5 lacs |
Yes |
To gain mentorship, expand our AI and IoT capabilities, and connect with investors and global networks to scale our water quality solutions. |
Validate AI/ML features for water quality prediction and analytics Build global partnerships for pilots and scaling Gain strategic mentorship for product-market fit and commercialization |
yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/vOwQONqdOydr-CLUIX_2025_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/vOwQONqdOydr-CLUIX_2025_compressed.pdf |
https://youtu.be/U1KmccovdaM?feature=shared |
Yes, Cluix is a mission-driven startup dedicated to addressing two of the world’s most pressing challenges: access to safe water and prevention of waterborne diseases. Our work directly aligns with the UN Sustainable Development Goals, particularly SDG 6 (Clean Water & Sanitation) and SDG 3 (Good Health & Well-Being). By making real-time, lab-grade water quality monitoring affordable and accessible, we empower municipalities, corporates, and communities to detect contamination early, prevent health risks, and manage water resources more sustainably. Beyond technology, our mission is impact-first: to create healthier communities, support climate resilience, and ensure that safe water is not a privilege but a universal right. |
na |
na |
checked |
| Sep 30, 2025 @ 6:13 PM |
Robin Singh |
robin@cluix.in |
https://www.linkedin.com/in/robin-singh-b9952028%20 |
https://www.cluix.in/ |
09560589782 |
Chitranjan Singh (Founder & Chairman)- Chitranjan Singh boasts an impressive career spanning over 30 years. He graduated with a degree in Civil and Structural Engineering from the College of Military Engineering in Pune in 1995. Since then, he has served as a distinguished bureaucrat within the Indian defense sector. His extensive experience covers numerous projects of national significance across India. Notably, he is known for his creative problem-solving skills and his proven track record in executing large-scale projects Robin Singh (Co-Founder) - Robin Singh is a visionary entrepreneur and technical innovator with a strong background in Electrical and Electronics Engineering and over a decade of experience driving impactful technological solutions. He specializes in automation, embedded sensors, electrochemical applications, and advanced machine design, with hands-on expertise in software development (C, C++, Python, Java) and microcontroller and microprocessor-based solutions. Under his leadership, CLUIX developed a Hand-held IoT Digital Water Quality Analyzer that empowers rural communities, Gram Panchayats, and industries to conduct real-time water quality assessments. His commitment to sustainable innovation was recognized with the Start-up India Challenge award, and he secured more than INR 2crore grant from Invest India under the Jal Jeevan Mission and several other government and international schemes. |
As a father–son duo, our journey has always been intertwined. Robin has been deeply inspired by his father, Chitranjan, whose guidance encouraged him to experiment, innovate, and build technical skills from a young age. This mentorship shaped Robin’s professional and personal growth, and together we formally started working on CLUIX over 3 years ago, combining experience with new-age innovation. |
Yes |
Strong family trust and alignment, complemented by a mix of technical expertise and business strategy, enabling rapid decision-making and innovation. |
CLUIX Private Limited |
https://www.cluix.in/ |
|
CLUIX is a cleantech deeptech startup, who has developed IoT-enabled water technology providing real-time, field-ready water quality monitoring solutions. |
In India, over 70% of water sources are contaminated, yet testing remains centralized, costly, and slow, leaving rural and underserved communities without timely information. Traditional lab testing takes weeks, creating a dangerous gap between contamination and corrective action. This lack of accessible, reliable, and real-time monitoring fuels waterborne diseases, loss of trust in public supply, and unnecessary spending on bottled water. CLUIX addresses this by enabling easy to use, portable, rapid, and affordable testing at the source, ensuring that communities, industries, and end-users can make immediate, data-driven decisions to safeguard health. |
Our innovation integrates advanced sensors, IoT connectivity, and data analytics into a compact water quality analyzer. The device measures multiple critical parameters in real time and transmits results instantly to a digital dashboard for users and governing bodies. This decentralized approach removes dependency on distant labs, while embedded AI-driven analytics detect anomalies and predict contamination trends with historical data. By combining portability with digital traceability, technology ensures transparency, trust, and scalability. It bridges the gap between rural communities, government programs, and industries by delivering actionable insights for safe water management. |
Portable, field-ready lab-grade accuracy Multi-parameter testing in one device Smart connectivity for real-time monitoring Scalable and adaptable for local and international water systems Proprietary combination of hardware and software algorithms |
Product-Market Fit |
Users, Revenue, Pilots |
Who It’s For • Government departments (PHED, Jal Nigam, Health, and Rural Development) • CSR and foundations working in WASH and public health • Water utilities and private operators • NGOs and international development agencies focused on last-mile delivery • Enterprises offering water as a service or running decentralized treatment plants |
46.04 Bn globally / $2.38 Bn in India — representing all possible use cases of water quality monitoring solutions. |
$2.04 Bn — focused and realistically addressable Indian market. |
$40.9 Mn – $102.25 Mn — the market share CLUIX could realistically capture. |
B2B & B2G |
Elico, Hanna, Hach, Xylem |
Through B2B sales, government procurement channels (GeM), CSR partnerships, and direct outreach to water authorities. |
Outbound Sales & Partnerships, Collaboration and PR Marketing, Online Marketing, Event Marketing |
Our vision is to make safe, real-time water quality monitoring accessible globally, directly contributing to SDG 6 by enabling clean water and sustainable water management for all. By ensuring proactive monitoring and early detection of contamination, we also advance SDG 3 by reducing waterborne diseases and promoting healthier communities worldwide. |
Private Limited |
234.5 lacs |
Yes |
To gain mentorship, expand our AI and IoT capabilities, and connect with investors and global networks to scale our water quality solutions. |
Validate AI/ML features for water quality prediction and analytics Build global partnerships for pilots and scaling Gain strategic mentorship for product-market fit and commercialization |
yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/AR9SrzGhkwQ7-CLUIX_2025_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/AR9SrzGhkwQ7-CLUIX_2025_compressed.pdf |
http://%20https://youtu.be/U1KmccovdaM?feature=shared%20%20%20%20 |
Yes, CLUIX is a mission-driven startup dedicated to addressing two of the world’s most pressing challenges: access to safe water and prevention of waterborne diseases. Our work directly aligns with the UN Sustainable Development Goals, particularly SDG 6 (Clean Water & Sanitation) and SDG 3 (Good Health & Well-Being). By making real-time, lab-grade water quality monitoring affordable and accessible, we empower municipalities, corporates, and communities to detect contamination early, prevent health risks, and manage water resources more sustainably. Beyond technology, our mission is impact-first: to create healthier communities, support climate resilience, and ensure that safe water is not a privilege but a universal right. |
NA |
NA |
checked |
| Sep 30, 2025 @ 4:59 PM |
Rahul Raj & Nitish Kamal Chaudhary |
resumeupgrader.rn@gmail.com |
https://www.linkedin.com/company/resume-upgrader-official/about/ |
https://www.resumeupgrader.in/ |
+918905936141 |
Rahul Raj (CEO), B.Tech in CSE from IIT Dharwad. Nitish Kamal Chaudhary (CTO), B.Tech in CSE from IIT Dharwad |
We both met each other during our undergraduate program; we are batchmates.
We have been working together for the past year. |
No |
As a team we are able to handle all the stressful situations and help each other to perform the tasks more effieciently. |
Resume Upgrader |
https://www.resumeupgrader.in/ |
|
Resume Upgrader is an AI-powered platform that creates tailored, ATS-friendly resumes, provides actionable insights, and enables one-click job applications for freshers and early professionals. |
Most job seekers, especially freshers and early professionals, fail to land interviews because their resumes are generic, misaligned with job descriptions, or rejected by ATS systems, and they lack the time, expertise, and feedback to improve them |
Resume Upgrader uses advanced AI and LLMs to instantly analyze resumes against job descriptions, provide visual match scores, and generate tailored, ATS-friendly resumes with keyword optimization. It also offers actionable insights, one-click job applications, personalized upskilling suggestions, and interview prep resources—helping job seekers maximize their chances of landing interviews. |
My solution stands out with AI-driven, per-role resume tailoring, ATS optimization, and one-click job applications—capabilities traditional resume builders and job boards lack. We combine individualized resumes, actionable feedback, and personalized upskilling in one platform, supported by a freemium model that ensures accessibility and scalability. |
Users |
Pilots |
Freshers, Early-career professionals, Universities, Training & Placement Cells. |
$27.6 Billion |
$2.76 Billion |
$ 220 Million |
Freemium model |
Kickresume, Zety, VisualCV, Novoresume, Resume.com |
I acquire customers through digital outreach, sponsorship activations, and community collaborations, along with strong partnerships with colleges and placement cells for bulk adoption. |
My GTM also includes academic network expansion, alumni success amplification, and funnel conversion to drive sustained growth. |
My long-term vision is to become the go-to AI-powered career platform that not only tailors resumes but also streamlines the entire hiring lifecycle—resume shortlisting, candidate testing, and interview scheduling—through products like HireSmarts (future scope).
Ultimately bridging the gap between job seekers and employers at scale. |
Incorporated |
NA |
Yes |
For getting Financial Support and Mentorship |
I aim to accelerate product development, strengthen our AI-driven resume and hiring solutions, and validate our business model through expert mentorship and milestone-based execution. I want to leverage the funding, co-working ecosystem, and network access to scale user adoption, establish institutional partnerships, and position Resume Upgrader as market-ready. |
Yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/b4qTEe3rDP8K-Pitchdeck_Resume_Upgrader.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/b4qTEe3rDP8K-Pitchdeck_Resume_Upgrader.pdf |
https://drive.google.com/file/d/1GTklQKroqbw7Ak-PSssQI5ihI0Bkv2K9/view?usp=sharing |
We are building a mission-driven, impact-focused startup that empowers students, freshers, and early professionals to access better career opportunities. Resume Upgrader uses AI to create personalized, ATS-friendly resumes, provide upskilling suggestions, and enable one-click applications, ensuring affordability and accessibility for all. By partnering with universities and organizations, we aim to improve employability outcomes at scale and level the playing field for job seekers. |
NA |
IIT Dharwad |
checked |
| Sep 30, 2025 @ 4:43 PM |
Adithya Moodbdri. Vrandha Nabhiraj |
beylesys@gmail.com |
http://na |
http://beylesys.com |
09886211222 |
i am the main founder of this company although officially this is in my mothers name , i drive the operations. I am a commerce graduate and have had persued chartered accountancy and have done upto the articleship. i was earlier the managing director of a private limited company formed by myself and this company was into consumer durable and white goods trading business. i have had a team of more than 120 people working in the organisation then. |
NA |
No |
we are family |
beylesys technologies OPC private limited |
https://www.beylesys.com |
|
we are building the future of accounting with AI- automating book keeping and delivering predictive insights for growing businesses. |
entrepreneurs are held back by outdated, manual accounting tools that waste time , increase errors and deny founders the real-time financial insights they need to grow. |
we are providing an AI native accounting platform that automates bookkeeping, ensures compliance and delivers real-time financial insights, empowering entrepreneurs to save time , cut costs, and make smarter growth decisions. |
unlike lagacy tools retro fitting with AI features, our product is AI native at its core , built to sutomate accounting end to end, deliver predicitve insights, and learn continuously from data. our defensibility comes from the proprietory AI/ML pipeline that is trained on custom financial data. |
MVP |
Pilots |
SMEs MSMEs, startups, freelancers, individuals and professionals. |
USD 11 Billion |
USD 3.3Billion |
USD 30 million |
Saas |
tally , ZOHO books |
digital and content marketing, CA partnerships, targeted customer outreach. |
We will launch by targeting SMEs and startups seeking smarter, simpler accounting solutions, positioning itself as an AI-native alternative to legacy software. Our go-to-market strategy combines direct digital campaigns (LinkedIn, Google Ads), partnerships with accelerators and accounting firms, and a freemium model to drive adoption and upsell premium features. We’ll build trust through content marketing on AI in finance, leverage referral incentives, and focus on retention via predictive insights and seamless integrations, scaling from early adopters to broader SME markets. |
Our long-term vision is to evolve from an AI-native accounting platform into a comprehensive, AI-driven ERP suite that not only automates finance and compliance but also powers office tools, CRM, supply chain management, and other core business operations. We aim to become the intelligent backbone for SMEs and startups, enabling them to operate efficiently, make strategic decisions, and scale seamlessly while AI handles insights, processes, and optimization across the entire organization |
incorporated under the companies act 2013 on 17th february 2025 |
bootstrapped |
Yes |
for technical assistance , mentorship in the area of product and market strategy , seed grants |
we aim to validate our AI-native accounting platform with SMEs, accelerate user acquisition, refine our product - market fit and establish strategic partnerships setting the foundation to scale and evolve into a full AI driven ERP suite. |
virtual prefereed |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/9TC4zJv8S6gB-Beylesys-Technonlogies-OPC-private-limited.txt, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/9TC4zJv8S6gB-Beylesys-Technonlogies-OPC-private-limited.txt |
https://drive.google.com/file/d/15--Rzjyp_84zsFC7sgGIys04HvFlbsWP/view?usp=sharing |
Both mission-driven and impact-focused. Our mission is to empower SMEs and startups with AI-driven financial tools that simplify accounting, enhance efficiency, and provide actionable insights—leveling the playing field for smaller businesses that often lack access to sophisticated finance solutions. By automating tedious processes and enabling smarter decision-making, we aim to reduce operational inefficiencies, increase financial transparency, and help entrepreneurs focus on growth and innovation, creating tangible economic and social impact across the SME ecosystem |
Yes, since i come from a commerce background and the founder my mother is also from a non tech background, the tech space is very much new and exciting to us. |
NITK STEP. |
checked |
| Sep 30, 2025 @ 1:16 PM |
Suyash Nahar , Vishruth V Srivatsa , Vatsal Gandhi |
naharsuyash@gmail.com |
https://www.linkedin.com/in/suyash101 |
https://lawveo.ai/ |
7899777020 |
Suyash Nahar - CEO & Director 3rd year NITK EEE , Vatsal Gandhi - COO & Director 3rd year NITK CSE , Vishruth V Srivatsa - CAIO & Director 3rd year NITK CDS |
We all met during our time in college. Suyash and Vatsal became close friends from the very first year, while Vishruth was initially just an acquaintance. In our second year, Suyash and Vishruth became roommates, and that’s when their friendship really grew. Over time, the three of us started working together, participating in 7–8 hackathons as a team. Our very first hackathon project, Lexify AI, eventually grew into the company we’re now building. We’ve been collaborating closely for over two years, not only through hackathons but also in multiple clubs we share on campus. |
Yes |
Our biggest strength lies in the strong chemistry we’ve built over the years and the way our skills complement each other. Vishruth is a strong AI developer, Vatsal excels in backend development and seamlessly integrates Vishruth’s models, while Suyash brings frontend expertise along with business insights from his background. These clearly defined yet perfectly aligned roles make us highly effective. We are all deeply passionate about building, and with each hackathon and project, our collaboration improves, our bond strengthens, and our ability to execute as a team grows even stronger. |
Lexify AI Private Limited |
https://lawveo.ai/ |
|
Lawveo.ai is an AI-powered platform that simplifies legal workflows by automating research, drafting, Translation , and case management for faster, smarter legal outcomes. |
We are solving the problem of time-consuming, complex, and expensive legal processes by streamlining research, drafting, translation, virtual courtroom, and case management with AI-powered automation. |
Scan and translate any legal document instantly with advanced OCR and translation. No more language barriers-just clear, accessible legal information wherever you are.
Research cases, draft arguments, and find precedents with Lexi. She works like a 24/7 paralegal-fast, accurate, and always ready to support you
Create legal documents instantly with our AI drafter. From contracts to petitions, reduce turnaround time without compromising accuracy.
Step into a virtual courtroom and sharpen your advocacy skills. Immersive simulations help young lawyers prepare for real-world proceedings.
Keep every case safe in an encrypted ledger. Organize, track, and preserve sensitive data with full confidentiality built in
All your tools, OCR, lexi, ledger, drafting, and courtroom-connected in one platform. A truly seamless workflow for the modern legal professional. |
Lawveo.ai unifies all essential legal tools like Translation&OCR, Lexi (AI assistant), Ledger (case management), automated drafting, and a Virual courtroom connected module into one platform. This creates a truly seamless workflow for the modern legal professional, saving time, reducing costs, and increasing accuracy and locking them in an end to end eco system. |
MVP |
Pilots, Signups, Testimonials |
Law schools , Legal Professionals ,End Users that need Legal remedy(in a later stage) |
Rs 40,000 Crores |
Rs 25,000 Crores |
Rs 1000 Crores |
Pay as you go |
jhanna.ai,lexlegis.ai,jurisphere,casemine |
Court hall stalls, Social media marketing , Word of mouth through referal |
NA(Still devising a solid plan before implementation) |
Our long-term vision is to transform the legal industry by creating a fully AI-powered, end-to-end legal platform that empowers lawyers, law firms, and corporates to handle legal work faster, smarter, and more efficiently. We aim to be the go-to platform for legal professionals globally, seamlessly integrating AI-driven research, drafting, case management, and courtroom tools to make legal services accessible, accurate, and cost-effective. |
We are incoperated in Manglore |
NA |
Yes |
We are applying to Avinya Ventures because of their strong focus on early-stage technology startups and their hands-on approach to supporting founders. Beyond capital, Avinya provides access to a network of experienced mentors, strategic investors, and a community of innovators, which aligns perfectly with our goal to scale Lawveo.ai efficiently while building a global, AI-powered legal platform. |
During the Avinya programme, we aim to refine our product-market fit, accelerate user acquisition, and strengthen our go-to-market strategy with guidance from experienced mentors. We also want to leverage Avinya’s network to build strategic partnerships, gain early adopters, and position Lawveo.ai for sustainable growth in the Indian legal-tech market and beyond. |
open to both |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/9XbVUnKdMwMs-Lawveo.aiStartupPitchDeckupdated-compressed-1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/9XbVUnKdMwMs-Lawveo.aiStartupPitchDeckupdated-compressed-1.pdf |
https://app.supademo.com/demo/cmdox0hhx00lz0k0isao76z1l?utm_source=link |
Yes, Lawveo.ai is a mission-driven startup committed to transforming the legal industry in India. Our platform leverages AI to streamline legal research, drafting, and case management, making legal services more accessible, efficient, and affordable. By reducing the time and cost associated with legal processes, we aim to empower legal professionals and enhance access to justice, particularly for underserved communities and small businesses. |
NA |
NITK STEP (Praseeda Kalkur) |
checked |
| Sep 29, 2025 @ 10:55 PM |
Saundarya Tyagi, Aaradhya Saxena |
saundarya.tyagi@bettermeals.in |
https://www.linkedin.com/in/saundaryatyagi/ |
https://bettermeals.in/ |
7023554282 |
Aaradhya Saxena (Co-founder): Systems engineer and applied ML researcher. Background in large-scale personalization and search at JioMart and TIRA. Researcher at CERN on code generation for deep-learning inference. Brings deep expertise in decision engines, scalable infra, and applying AI to real-world systems. Competitive athlete (swimming, rowing) with firsthand experience of disciplined nutrition. Saundarya Tyagi (Co-founder): Senior engineer with prior experience at Wayfair and Goldman Sachs. Awarded Wayfair’s global People Principle Award (“Adapt & Grow”) for driving high-impact initiatives. Expertise in building resilient infrastructure, productizing AI systems, and scaling consumer-facing tech. Athlete (basketball) with a lived understanding of structured eating and where it breaks down. |
We met at Wayfair India, where we collaborated on engineering projects and hackathons. That experience built trust and complementary working styles, which we extended beyond work — volunteering with People+AI to build Cheenikum.ai, a health agent for women with gestational diabetes. Over the past year, we’ve worked together on multiple health-tech prototypes and hackathons, and for the last two months we’ve been full-time on BetterMeals. Our journey reflects not just shared skills, but a shared conviction that food execution is the missing link in health. |
Yes |
Our biggest strength is that we’re both builders at heart, hands-on, and fast. Aaradhya thinks in systems and has scaled ML engines for millions; Saundarya thrives on infrastructure and getting products production-ready. Together, we’ve shown we can turn an idea into something real in days, whether it was winning hackathons at Google, Wayfair, and South Park Commons, or building a health agent with People+AI for women with GDM. What really holds us together is a shared bias for action and the grit we’ve carried from our sports backgrounds, rowing, swimming, basketball where discipline and teamwork mattered every day. That mix of speed, technical depth, and lived conviction around why healthy eating fails without infrastructure makes us unusually aligned, and determined to build. |
BetterMeals |
https://bettermeals.in/ |
|
BetterMeals turns lab reports and household quirks into a 3 dish, 45 minute cook plan- every day, auto groceried, cook ready. |
Eating healthy isn’t a knowledge problem — it’s an execution problem. Most people already know they should eat better, but they fail in the messy day-to-day: planning varied meals, buying the right groceries, and coordinating with whoever cooks at home. Diet charts, apps, and trackers don’t solve this last mile — they add more work. In India alone, 100M+ people live with diabetes and millions more with PCOS, fertility, and lifestyle diseases, yet compliance to meal plans is near zero because there’s no infrastructure to enforce them. We’re solving that gap: the missing execution layer that takes health goals, lab reports, and household quirks, and quietly turns them into consistent meals on the table. |
BetterMeals is the execution layer for food, an AI-driven health automation engine that translates lab reports, biometrics, and household quirks into daily, personalized, cook-ready meal plans. Our core is a multi-objective solver that balances nutrient targets, health protocols, preferences, and grocery constraints to generate 3 dishes/day (≤45 mins cook time). We map meals to raw ingredients, auto-optimize grocery orders with perishability-aware inventory management, and deliver instructions to home cooks via WhatsApp and voice agents.
The system is built on a modular architecture: a nutrient reasoning engine (Python/FastAPI) orchestrated with GPT-4 and LangChain, Qdrant for profile and lab-based retrieval, n8n for agentic workflows, and ElevenLabs/Sarvam for cook-facing voice guidance. This ensures that dietary science, household logistics, and real-world constraints interlock seamlessly.
Unlike apps that give advice, BetterMeals enforces compliance in messy, real homes: skipping meals triggers rebalancing, missing ingredients prompt substitutions, and cook absences are handled with automated fallbacks. At scale, the same infrastructure powers B2B integrations (digital clinics, wellness apps, insurers), residential communities (bulk-optimized grocery flows), and remote care (ensuring elderly parents eat healthily).
We’re not just solving “what to eat”, we’re automating the entire loop of planning, procurement, and execution, making healthy eating default at population scale. |
Most health apps stop at advice trackers, nudges, or static diet charts. BetterMeals is unique because it executes. Our defensibility comes from three layers working together: Full-stack execution loop — We own the end-to-end chain: nutrient reasoning, meal generation, grocery mapping, and cook communication. This allows us to enforce compliance in real households, not just suggest it. Technical moat — Our solver combines retrieval (lab reports, preferences, inventory) with multi-objective optimization for nutrient balance, perishability, and cultural fit. We’ve built agentic WhatsApp + voice flows that adapt dynamically (e.g., meal skipping, missing ingredients, cook absences). Each solved edge case compounds into a knowledge base that’s hard to replicate. Distribution moat — By embedding in existing household flows (WhatsApp, local vendors, cooks), we avoid app fatigue and become infrastructure, not another app competing for attention. Scalability — The same engine powers multiple markets: B2C households, B2B clinics/dietitians, residential communities, and NRI elderly care. What makes us defensible is not a single algorithm, but the closed-loop system we’ve built, one that integrates dietary science, AI agents, and messy real-world household behavior into a seamless execution engine. |
MVP |
Pilots, Signups |
Our initial target customers are urban Indian households with cooks, where meal decisions are daily pain points but execution still runs through kitchens. Within these, we focus on health-conscious families managing lifestyle conditions (diabetes, PCOS, fertility, postnatal recovery) and fitness-focused professionals who want structured nutrition without micromanaging. |
~$15-18B |
~$1B+ |
~$100-300M |
BetterMeals monetizes through three streams: Grocery margin/commission — We manage household grocery flows (₹6,000–₹16,000/month per household) and take a 5–10% commission (~₹1,200/HH/month). Subscription tiers — Free basic meal plans; paid tiers from ₹1,500–₹15,000/month for advanced features like diagnostics-linked personalization, providing cook professional cook services as well. B2B licensing — White-label infra for dietitians, digital clinics, wellness insurers, and residential communities. |
This is a very new category with few direct incumbents in India yet. The closest local player is Kookar, which focuses on cook enablement, while in the US apps like Eat This Much automate parts of healthy meal planning. Broader alternatives include health apps like HealthifyMe (advice and tracking) or meal-kit services (convenience but not personalization). BetterMeals is unique because we don’t stop at recipes or advice, we own the full execution loop: meal planning, grocery mapping, and cook guidance. |
We started with a pilot waitlist of 500+ households in just 4 days, driven entirely by word-of-mouth and health-conscious communities existing in our own network. Adoption spreads virally inside families and neighborhoods since multiple members and cooks experience the product together. Beyond organic growth, we partner with digital clinics, dietitians, and diagnostic clinics who need meal execution for their patients, and with residential communities where one pilot unlocks hundreds of households via shared vendors. Because BetterMeals integrates into existing family and cook workflows, acquisition is low-CAC, sticky, and referral-driven. |
We’re starting with urban Indian households that employ cooks — ~20M+ homes where the pain of daily meal planning and grocerying is highest. Our GTM is bottom-up virality through pilots with families, cooks, and residential communities, where adoption spreads within extended families and across neighbors sharing vendors. In parallel, we’re building B2B2C channels with digital clinics, diagnostics, and dietitians who need meal execution for their patients. This dual approach combines direct consumer traction with enterprise integrations for scale. |
Our long-term vision is to build a Jarvis-like home assistant that doesn’t just advise but executes, starting with the kitchen. Food is the hardest, most frequent decision families face, so we begin by automating meal planning, grocerying, and cook guidance. Over time, the same framework extends outward: a cook asking our system for a recipe or substitutions, groceries auto-reordered as inventory runs low, deliveries coordinated without user involvement, and contextual health advice generated directly from lab reports. At its core, we are building the execution layer for home. Intelligent agents that quietly manage daily living. What begins as kitchen automation evolves into a full-stack home operating system that ensures healthier, smarter, and frictionless living. |
Done |
N/A |
Yes |
We’re at an inflection point where we’ve validated demand and now need the right ecosystem to scale. Avinya’s network and credibility can help us form critical partnerships with diagnostics labs, healthcare providers, and other startups far faster than we could alone. Beyond capital, we’re looking for guidance from mentors who’ve scaled in India before, and a peer community that sharpens our thinking. Our mission is simple but ambitious — to improve the quality of life in India by making healthy eating effortless — and we believe Avinya can help us accelerate that mission with speed and trust. |
During the programme, our goal is to sharpen our go-to-market playbook and build the right partnerships that will help us scale faster — especially with diagnostics labs, wellness providers, and residential communities. We want to refine our business model and unit economics with the guidance of mentors who’ve scaled in India before. Another key milestone is to establish early credibility through Avinya’s network, so that adoption with healthcare partners and early customers becomes smoother. By the end of the programme, we aim to move from early pilots to a repeatable, scalable model with clear traction. |
Hybrid |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/kJK22suPjxcx-BetterMealsPitchDeck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/kJK22suPjxcx-BetterMealsPitchDeck.pdf |
https://drive.google.com/file/d/1vxGjRv6KSmkH42H8fbp5AS0gy9oH5eBG/view?usp=sharing |
Yes, BetterMeals is fundamentally mission-driven. India is in the middle of a nutrition crisis: over 100M people live with diabetes, 57% of women are anaemic, and nearly 70% of the population lacks Vitamin D. Protein and fiber intake across households is far below basic requirements, while traditional foods like millets and pulses are rapidly disappearing from our plates. At the same time, the modern food environment makes eating healthy harder than ever, fragmented apps, generic diet charts, and endless recipes put the burden on individuals who are already stretched thin. The result is a nation that knows what to eat but consistently fails at execution. BetterMeals is building the infrastructure to fix this, turning lab reports, health goals, and household realities into daily meals that actually get cooked and eaten. By solving compliance at scale, we aim to improve the baseline quality of life for millions of Indian families. |
Yes. As a woman founder in health-tech, I represent a group that remains significantly underrepresented in India’s startup ecosystem. |
Praseeda Kalkur |
checked |
| Sep 29, 2025 @ 3:29 PM |
Meenatchi Udayasankar |
meena.femicura@gmail.com |
https://www.linkedin.com/in/meena-udayasankar-ab04b8100?utm_source=share&utm_campaign=share_via&utm_content=profile&utm_medium=android_app |
https://femicura.com/ |
7339279069 |
1. Meenatchi .U, Founder. Biomedical Engineer; 9+ years of experience in developing medical and healthcare products |
My husband Mr. Madhan .S is the co-founder of the company. Together we started this company. Mr. Tamil Venthan And Miss. Salma. S are full employers. we have been working together for past 2 years |
Yes |
Our biggest strength as a team is collaboration. We bring diverse skills together, support each other, and work toward a common goal with shared commitment. |
FemiCura HealthTech Private Limited |
https://femicura.com/ |
|
Leveraging Green Innovations to nurture Women and Babies’ wellness |
Despite growing awareness around health and sustainability, the hygiene product market for women and babies remains saturated with synthetic, chemical-laden, and non-biodegradable products that can cause skin irritation, infections, and long-term environmental harm. Women in their postpartum phase and newborns are especially vulnerable to harsh ingredients and poor-quality materials used in mainstream products. Additionally, there is a lack of certified, clinically tested, and eco-friendly options that cater to their sensitive needs while also addressing environmental concerns. |
FemiCura HealthTech bridges this gap by designing and clinically testing gentle, organic, patent filed hygiene products Green Seal certified, biodegradable, and free from petrochemicals or harsh preservatives providing safe, eco friendly choices for mothers and babies across global markets |
Unlike mainstream brands that focus on cost efficiency over health and sustainability, CutetotZ prioritizes safe, non-toxic, and eco-conscious solutions. By integrating plant-based materials, dermatologically tested ingredients, clinically tested and advanced absorption technology, we offer products that truly cater to modern, health-conscious, and environmentally responsible consumers. All our products comply with global standards like FDA, WHO, CIR. |
Revenue |
Revenue |
Eco conscious parents, New Mothers, Health Conscious parents |
5-7 billion USD |
3-5 billion USD |
30-50 million USD |
B2B, B2C, Hospitals, Pharmacies, Distributors |
Baby wipes- Mother sparsh, Himalayas, Johnson and Johnson. Maternity pads- Sparkle, New Mom, Dignity, |
Social Media Marketing, Hospitals Tie ups, Marketing Campaigns. |
ur go-to-market strategy focuses on targeting women through digital platforms and healthcare networks, offering affordable and eco-friendly products. We will sell via e-commerce, pharmacies, and hospital tie-ups, supported by awareness campaigns and distributor partnerships to scale across regions. |
Our long-term vision is to become a trusted global leader in women’s and children's health and wellness, delivering innovative, affordable, and sustainable solutions that improve lives and create lasting social impact. We aim to expand access to quality healthcare products across underserved communities. By combining technology and empathy, we aspire to set new standards in preventive care and women-centric innovation worldwide. Our focus is on developing organic, biodegradable products that meet global safety and quality standards, ensuring both health and environmental well-being. |
06/09/2023 |
70 Lakhs |
No |
Marketing and commercialization support |
Market Support |
Yes. Virtual Participation |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/6HFTYVvMcvKt-compressed_Meenatchi-U_FemiCura-7-mins.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/6HFTYVvMcvKt-compressed_Meenatchi-U_FemiCura-7-mins.pptx |
http://Our%20long-term%20vision%20is%20to%20become%20a%20trusted%20global%20leader%20in%20women’s%20health%20and%20wellness,%20delivering%20innovative,%20affordable,%20and%20sustainable%20solutions%20that%20improve%20lives%20and%20create%20lasting%20social%20impact.%20We%20aim%20to%20expand%20access%20to%20quality%20healthcare%20products%20across%20underserved%20communities.%20By%20combining%20technology%20and%20empathy,%20we%20aspire%20to%20set%20new%20standards%20in%20preventive%20care%20and%20women-centric%20innovation%20worldwide.%20Our%20focus%20is%20on%20developing%20organic,%20biodegradable%20products%20that%20meet%20global%20safety%20and%20quality%20standards,%20ensuring%20both%20women’s%20health%20and%20environmental%20well-being. |
Yes. We are building an impact-focused, mission-driven startup. Our mission is to make women’s health and hygiene products affordable, accessible, and sustainable, especially for underserved communities. We focus on creating organic, biodegradable, and safe solutions that not only improve women’s health outcomes but also reduce environmental waste.
By integrating technology, healthcare, and sustainability, we aim to break barriers in access to preventive care and empower women with dignity and choice. Our impact goes beyond business , it is about improving lives, protecting the environment, and setting global standards in women’s wellness. |
NA |
TECHIN-IIT PALAKKAD |
checked |
| Sep 29, 2025 @ 2:54 PM |
Dr. Shivaji K. Jadhav |
scientistshivaji@gmail.com |
https://www.linkedin.com/in/dr-shivaji-k-j-43125338 |
https://suvihaanbionobel.com//index |
6363786540 |
Dr. Shivaji K. Jadhav (Founder & CEO): Ph.D. in Molecular HIV Virology with 22 years of experience in Molecular Diagnostics, Oncology, and Infectious Diseases. Former Head of Molecular Laboratory at Orange Health and Senior Scientist at NIMR. Awarded 21 international awards including Fogarty NIH Fellowship and Bill & Melinda Gates Global Award. |
Dr. Shivaji K. Jadhav (Founder & CEO): Ph.D. in Molecular HIV Virology, 22 years in Molecular Diagnostics, Oncology, and Infectious Diseases. Led labs at Orange Health & NIMR; 21 international awards including Fogarty NIH Fellowship and Bill & Melinda Gates Global Award.
Varsha Shivaji Jadhav (COO & Co-founder): M.Tech. (Ph.D.) in Information Science/Computer Science, 15 years in AI/ML for healthcare, specializing in cervical cancer diagnostics.
Research Associate & 2 Interns: Support AI model development, data curation, and testing.
Administrative Assistant: Manages operations and coordination.
Advisors: Experts in Oncology, AI, and Regulatory Affairs guiding product development, clinical validation, and compliance.
How We Met & Worked Together:
Dr. Shivaji and Varsha Shivaji Jadhav have been collaborating professionally for over 5 years, combining medical research and AI technology to build scalable solutions in diagnostics. |
Yes |
Our biggest strength is the combination of deep expertise in healthcare and oncology with advanced AI/ML skills, enabling us to develop a scalable, accurate, and real-time cervical cancer diagnostic platform. This innovation helps women by enabling early detection, accurate staging, and timely treatment, especially in underserved areas. Guided by expert advisors in oncology, AI, and regulatory affairs, our team can effectively tackle technical and clinical challenges. |
SUVIHAAN BIONOBEL SOLUTIONS PRIVATE LIMITED |
https://suvihaanbionobel.com//index |
|
Suvihaan Bionobel Solutions develops an AI-powered platform for real-time detection and staging of cervical cancer, improving early diagnosis and treatment for women, especially in underserved areas. |
We are solving the problem of late and inaccurate cervical cancer diagnosis, which is time-consuming, expert-dependent, and often inaccessible in rural and underserved areas, leading to delayed treatment and high mortality among women. |
Suvihaan Bionobel Solutions offers an AI-powered diagnostic and staging platform that enables early, accurate, and automated detection of cervical cancer from Pap smear and histopathological images.
1. AI-powered platform for analyzing Pap smear and cytology images.
2. Detects abnormal cells and accurately stages cervical cancer.
3. Provides real-time, standardized, and consistent results.
4. Optimized for telemedicine and low-resource settings.
5. Easily integrates into hospitals, diagnostic labs, and screening programs.
“With this technology, healthcare providers can:”
1. Diagnose and stage cervical cancer in real time.
2. Reduce dependence on manual interpretation and expert cytologists.
3. Accelerate treatment planning through actionable insights based on cancer stage.
4. Reach underserved populations with cost-effective, scalable AI diagnostics. |
Our solution is the first integrated AI platform that not only detects cervical cancer but also stages it in real time. It minimizes expert dependence, delivers consistent and rapid results, adapts to low-resource settings, and is scalable across hospitals and screening programs, making it both unique and defensible. |
MVP |
Pilots |
Hospitals and diagnostic labs looking for faster, accurate, and scalable cervical cancer screening. Government health programs and NGOs conducting mass screening. Clinics, gynecologists, and telemedicine providers serving rural or underserved populations. |
11,17,94,10,01,380.00 |
3,54,90,19,05,200.00 |
44,38,76,00,000.00 |
Subscriptions from hospitals and labs Licensing to medical device manufacturers Pay-per-use diagnostic tests Government/NGO bulk screening partnerships Selling anonymized clinical data insights |
Not In india |
We acquire customers through partnerships with hospitals, labs, clinics, and government health programs, complemented by targeted outreach, workshops, and pilot deployments. |
We will launch pilots in hospitals and labs, collaborate with government and NGO screening programs, and use targeted outreach and medical workshops to scale adoption through subscriptions and licensing. |
To revolutionize cervical cancer diagnostics worldwide by providing AI-powered, accurate, and affordable detection and staging solutions, ensuring early diagnosis and better treatment outcomes for women, especially in underserved regions. |
CIN NO: U82990KA2024PTC186028 |
NA |
Yes |
We are applying to Avinya to access grant funding, mentorship, and industry networks that will help us accelerate development, validate our AI-powered cervical cancer diagnostic platform, conduct clinical trials, and expand access to underserved populations. |
During the programme, we aim to refine and validate our AI models, complete regulatory approvals, deploy pilot tests in hospitals and labs, and strengthen our go-to-market strategy to scale access to cervical cancer diagnostics, especially for underserved women. |
Yes, we are open to both hybrid and virtual participation to fully engage with the programme and leverage its resources. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/ywBkBJ2EuReD-Pitch-Deck-of-SUVIHAAN-ACSEL-1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/ywBkBJ2EuReD-Pitch-Deck-of-SUVIHAAN-ACSEL-1.pdf |
https://www.youtube.com/watch?v=rTViRKW4PIU&list=PL9S6xGsoqIBUGffj1BWAMAGePr_MnREtv&index=14 |
Yes, our startup is mission-driven and impact-focused. We aim to save lives and improve women’s health by providing AI-powered cervical cancer detection and staging. Our solution increases early diagnosis, reduces mortality, and makes high-quality screening accessible and affordable, especially in underserved areas. |
NA |
NA |
checked |
| Sep 29, 2025 @ 2:39 PM |
Dr. Shivaji K Jadhav |
scientistshivaji@gmail.com |
https://www.linkedin.com/in/dr-shivaji-k-j-43125338 |
https://suvihaanbionobel.com//index |
6363786540 |
Founder and CEO: Ph.D. in Molecular HIV Virology with 22 years of experience in Molecular Diagnostics, Oncology, and Infectious Diseases. Former Head of Molecular Laboratory at Orange Health and Senior Scientist at NIMR. Awarded 21 international awards including Fogarty NIH Fellowship and Bill & Melinda Gates Global Award. |
including Dr. Shivaji K Jadhav
Varsha Shivaji Jadhav (COO & Co-founder): M.Tech. (Ph.D.) in Information Science/Computer Science, 15 years in AI/ML for healthcare, specializing in cervical cancer diagnostics.
Research Associate & 2 Interns: Support AI model development, data curation, and testing.
Administrative Assistant: Manages operations and coordination.
Advisors: Experts in Oncology, AI, and Regulatory Affairs guiding product development, clinical validation, and compliance.
Dr. Shivaji and Varsha Shivaji Jadhav have been collaborating professionally for over 5 years, combining medical research and AI technology to build scalable solutions in diagnostics. |
Yes |
Our biggest strength is the combination of deep expertise in healthcare and oncology with advanced AI/ML skills, enabling us to develop a scalable, accurate, and real-time cervical cancer diagnostic platform. This innovation helps women by enabling early detection, accurate staging, and timely treatment, especially in underserved areas. Guided by expert advisors in oncology, AI, and regulatory affairs, our team can effectively tackle technical and clinical challenges. |
SUVIHAAN BIONOBEL SOLUTIONS PRIVATE LIMITED |
https://suvihaanbionobel.com//index |
|
Suvihaan Bionobel Solutions develops an AI-powered platform for real-time detection and staging of cervical cancer, improving early diagnosis and treatment for women, especially in underserved areas. |
We are solving the problem of late and inaccurate cervical cancer diagnosis, which is time-consuming, expert-dependent, and often inaccessible in rural and underserved areas, leading to delayed treatment and high mortality among women. |
Suvihaan Bionobel Solutions offers an AI-powered diagnostic and staging platform that enables early, accurate, and automated detection of cervical cancer from Pap smear and histopathological images.
1. AI-powered platform for analyzing Pap smear and cytology images.
2. Detects abnormal cells and accurately stages cervical cancer.
3. Provides real-time, standardized, and consistent results.
4. Optimized for telemedicine and low-resource settings.
5. Easily integrates into hospitals, diagnostic labs, and screening programs.
“With this technology, healthcare providers can:”
1. Diagnose and stage cervical cancer in real time.
2. Reduce dependence on manual interpretation and expert cytologists.
3. Accelerate treatment planning through actionable insights based on cancer stage.
4. Reach underserved populations with cost-effective, scalable AI diagnostics. |
Our solution is the first integrated AI platform that not only detects cervical cancer but also stages it in real time. It minimizes expert dependence, delivers consistent and rapid results, adapts to low-resource settings, and is scalable across hospitals and screening programs, making it both unique and defensible |
MVP |
Pilots |
Clinics, gynecologists, and telemedicine providers serving rural or underserved populations. Hospitals and diagnostic labs looking for faster, accurate, and scalable cervical cancer screening. Government health programs and NGOs conducting mass screening. |
₹ 11,18,43,51,84,720.00 |
₹ 3,55,10,07,11,200.00 |
₹ 44,38,76,00,000.00 |
Subscriptions from hospitals and labs, Licensing to medical device manufacturers, Pay-per-use diagnostic tests, Government/NGO bulk screening partnerships, Selling anonymized clinical data insights |
NA |
We acquire customers through partnerships with hospitals, labs, clinics, and government health programs, complemented by targeted outreach, workshops, and pilot deployments. |
We will launch pilots in hospitals and labs, collaborate with government and NGO screening programs, and use targeted outreach and medical workshops to scale adoption through subscriptions and licensing. |
To revolutionize cervical cancer diagnostics worldwide by providing AI-powered, accurate, and affordable detection and staging solutions, ensuring early diagnosis and better treatment outcomes for women, especially in underserved regions. |
COI NO: U82990KA2024PTC186028 |
NA |
Yes |
We are applying to Avinya to access grant funding, mentorship, and industry networks that will help us accelerate development, validate our AI-powered cervical cancer diagnostic platform, conduct clinical trials, and expand access to underserved populations. |
During the programme, we aim to refine and validate our AI models, complete regulatory approvals, deploy pilot tests in hospitals and labs, and strengthen our go-to-market strategy to scale access to cervical cancer diagnostics, especially for underserved women. |
Yes, we are open to both hybrid and virtual participation to fully engage with the programme and leverage its resources. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/zBPkU3LNdxnv-Pitch-Deck-of-SUVIHAAN-ACSEL-1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/zBPkU3LNdxnv-Pitch-Deck-of-SUVIHAAN-ACSEL-1.pdf |
https://www.youtube.com/watch?v=rTViRKW4PIU&list=PL9S6xGsoqIBUGffj1BWAMAGePr_MnREtv&index=14 |
Yes, our startup is mission-driven and impact-focused. We aim to save lives and improve women’s health by providing AI-powered cervical cancer detection and staging. Our solution increases early diagnosis, reduces mortality, and makes high-quality screening accessible and affordable, especially in underserved areas. |
NA |
NA |
checked |
| Sep 29, 2025 @ 1:09 AM |
Ritesh Appasaheb Belgudri; Dr. Koteswararao Kondepu; Dr. Nagaveni S |
ritesh.belgudri@entra-in.com |
https://www.linkedin.com/in/riteshbelgudri |
https://entra-in.com/ |
9964865457 |
Ritesh Belgudri is the CEO of Entra Innovations, with over a decade of experience in the FPGA industry, including previous roles at Intel and Altera. Dr. Koteswararao Kondepu is the Chief Technology Officer (CTO) of Entra Innovations and an Associate Professor at IIT Dharwad. He received his PhD and completed postdoctoral research at Institute for Advanced Studies IMT Lucca, Lucca, Italy, and his work focuses on the implementation of networks and communications for 5G and beyond. Dr. Nagaveni S is the Chief Product Officer (CPO) of Entra Innovations and an Assistant Professor at IIT Dharwad. She earned her PhD from IIT Hyderabad, and her research specializes in Analog and RF Integrated Circuit design. |
Ritesh Belgudri is a PhD scholar under the supervision of Prof. Nagaveni and co-supervision of Prof. Koteswararao Kondepu. Over the past year, he has been working with them on FPGA design and open-source tool development. |
No |
Our team brings together a unique blend of research excellence and industry experience, creating the ideal foundation for a deep-tech company. This combination equips us with the expertise and practical insight needed to tackle complex technical challenges and deliver innovative solutions. |
Entra Innovations LLP |
https://entra-in.com/ |
|
Entra Innovations LLP is a semiconductor design company developing Programmable Integrated Circuits and Intellectual Property for Automotive, Telecom and Defense Solutions. |
Reducing India’s reliance on foreign semiconductor supply is a critical step towards achieving technological self-reliance and national security. y strengthening domestic design, and innovation capabilities, we aim to build a robust semiconductor ecosystem that not only diminishes import dependency but also fosters indigenous innovation, economic growth, and global competitiveness. |
Entra Innovations' Akriti: Integrated Programmable Logic Generator (IPL-G) provides comprehensive solution for programmable logic (fabric and interconnect) IP generation leveraging the fully open-source Verilog-to-Routing (VTR) backend, open-source benchmark circuits and open-source EDA tools using Agentic AI to deliver cost-effective, customizable programmable logic solutions. USP: Reduce development cost by 50% and development time to weeks.
Our approach combines proven open-source synthesis and place-and-route tools (Yosys, ABC, VPR) with innovative programmable fabric architectures, enabling rapid deployment of tailored programmable logic IP cores for diverse applications including aerospace, defence, automotive, and AI/ML acceleration.
Building on our comprehensive programmable logic solution, we present an enhanced approach that leverages programmable logic hard IP integration into customer SOCs using commercial EDA tools (Cadence & Synopsys) while incorporating our unique selling proposition (USP): the latest interconnect routing technology, Reinforcement Learning (RL), and Deep Learning (DL) algorithms from VTR 9 for unprecedented routing optimization and placement efficiency. |
USP: Reduce development cost by 50% and development time to weeks. |
MVP |
Pilots, Signups |
B2B customers developing their custom SoC - Astra Microwave Products Ltd, DRDO, ISRO, Bharat Electronics Limited, Radisys, Tejas Networks, Tata Advanced Systems, L&T Semiconductors, Morphing Machines, Mindgrove |
$10.5B |
$330M |
$6M |
Fixed Licensing & Royalty per chip (depending on technology node & device density) |
QuickLogic, Achronix, Flex Logix (Acquired By Analog Devices) (all based out of USA), none in India |
Sales & Distribution Channels: Direct Sales Team, Channel Partners, Academic Partnerships and Online Platform |
Phase 1: Early adopters (Defense, Automotive Tier-1) Phase 2: Mid-market expansion (Telecom, AI/ML companies) Phase 3: Enterprise scaling (Global semiconductor companies) |
We aspire to become the number one semiconductor company in India and a global leader in providing cutting-edge programmable solutions. Our long-term vision is to deliver end-to-end indigenous programmable logic solutions that reduce dependency on foreign technologies while enabling cost-effective FPGA (Field Programmable Gate Array) integration. |
Incorporated on 28 August 2025, Status: Active |
zero |
Yes |
We are applying to Avinya (ACSEL) to accelerate the growth of Entra Innovations through access to world-class mentorship, resources, and industry networks. Avinya’s deep-tech focus and proven support for innovative startups align perfectly with our mission to build indigenous programmable logic solutions and strengthen India’s semiconductor ecosystem. Participation in Avinya will enable us to refine our business strategy, scale our technology, and connect with industry leaders, ultimately helping us drive impact both nationally and globally. |
During the programme, our primary objectives are: To refine our business and technology strategy with guidance from mentors and industry experts. To accelerate the development and commercialization of our indigenous FPGA solutions. To build strategic partnerships for both market access and technology co-development. To attract early customers through pilot deployments and targeted outreach. To strengthen our team’s entrepreneurial and leadership skills for scaling Entra Innovations. Participating in Avinya (ACSEL) will help us lay a strong foundation for sustainable growth, drive innovation, and position ourselves as leaders in India’s deep-tech semiconductor industry. |
Yes, open to both hybrid or virtual participation |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/6CDP7DlbuybT-Entra_Akriti_2025-Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/6CDP7DlbuybT-Entra_Akriti_2025-Pitch-Deck.pdf |
http://NA |
Entra Innovations is both mission-driven and impact-focused. Our core mission is to empower India’s technology sector by developing indigenous programmable logic solutions that reduce reliance on foreign semiconductor imports. We address critical national priorities, including technological self-reliance, economic growth, and global competitiveness in the semiconductor domain.
Our impact extends beyond product innovation. By fostering local talent, advancing domestic research, and enabling cost-effective access to advanced FPGA technologies, we aim to build a self-sustaining ecosystem that benefits not only industry but also academia and society at large. Our commitment to solving complex engineering challenges and nurturing deep-tech capabilities makes Entra Innovations a catalyst for positive change in India’s semiconductor landscape. |
We are based in Dharwad (North Karnataka), a region that faces challenges in accessing diverse and deep-tech talent pools compared to established tech hubs. By building our startup here, we aim to challenge the status quo, broaden regional representation, and demonstrate that world-class innovation can arise from emerging ecosystems outside major metropolitan areas. |
IIT Dharwad's dhaRti Foundation, Point of Contact: Rakshit Kalyani, COO |
checked |
| Sep 29, 2025 @ 1:08 AM |
Himanshu Gupta |
renergizr.industries@gmail.com |
https://www.linkedin.com/company/renergizr/ |
https://www.linkedin.com/company/renergizr/ |
8970537156 |
Himanshu Gupta : Founder and CEO : Himanshu Gupta is a clean-energy entrepreneur and mechanical engineer with over 12 years of international experience in geothermal power systems, hydrogen technologies, and industrial energy solutions. He earned his B.Tech in Mechanical Engineering from the University of Alberta (Canada), where he first worked on renewable energy and waste-heat recovery projects. Before founding Renergizr Industries in 2021, Himanshu worked on geothermal and hydrogen energy projects across Canada, Germany, Iceland, Kenya, and India, gaining hands-on expertise in power plant design, fuel cell integration, and large-scale pilot deployments. He is the co-inventor of Renergizr’s patented clean energy genset technology (Indian Patent No. 548040), which produces 24/7 zero-emission electricity and green hydrogen using wastewater and industrial waste heat. Under his leadership, Renergizr has: Built and commissioned 5 kW and 20 kW pilot systems in Delhi and Telangana. Secured ₹3 crore in non-dilutive grants from DST, MNRE, PFC, and other national programs. Received Letters of Intent from SCCL and the Indian Armed Forces for MW-scale deployments. Won four national clean-energy innovation awards. Himanshu is a full-time founder focused on product development, strategic partnerships, and scaling Renergizr’s operations in India and international markets such as the UAE. |
NA |
Yes |
Our biggest strength as a team is the rare combination of deep technical expertise and practical commercialization experience in clean energy.
The founding group blends multi-disciplinary engineering (hydrogen, fuel cells, geothermal, and power electronics) with hands-on project execution in India and overseas.
This allows us to design a patented clean-energy genset from first principles, rapidly build and deploy working pilots, and simultaneously navigate regulatory, utility, and industrial-scale partnerships.
Beyond technical skill, the team shares a mission-driven culture focused on solving real energy-reliability problems, not just developing lab prototypes.
Members bring prior experience in international renewable projects (Canada, Germany, Iceland, Kenya), large-scale industrial operations in India, and early-stage startup growth—giving Renergizr both global insight and strong local execution capability.
This blend of innovation, field experience, and shared purpose enables us to convert a complex technology into a scalable, bankable product that can replace diesel gensets in demanding markets like India and the UAE. |
Renergizr Industries Private Limited |
https://www.linkedin.com/company/renergizr/ |
|
Renergizr Industries delivers patented clean-energy gensets that generate 24/7 zero-emission electricity and green hydrogen, replacing polluting diesel generators. |
Renergizr is eliminating dependence on diesel generators, which are costly, noisy, and major sources of carbon and air pollution, by providing a reliable, 24/7 clean power alternative that meets industrial energy needs without fossil fuels or expensive battery storage. |
Renergizr has developed a patented clean-energy genset system that produces uninterrupted, 24/7 electricity and high-purity green hydrogen using treated wastewater and waste heat instead of fossil fuels.
Our modular containerized units integrate an electrolyzer, hydrogen storage, and a high-efficiency fuel cell to deliver grid-quality power with >98% uptime, zero emissions, and up to 80% lower operating cost than diesel gensets.
The system can be deployed in sizes from 5 kW to 1 MW, making it ideal for industries, smart cities, defense bases, and remote communities seeking a dependable, scalable, and sustainable alternative to diesel. |
Renergizr’s solution is protected by an Indian patent (No. 548040) and combines on-site green hydrogen production from wastewater with a modular fuel-cell genset, enabling continuous 24/7 power without fossil fuels or costly battery storage. This closed-loop design lowers operating costs by up to 80% and achieves >98% uptime—performance that current solar-plus-battery or diesel systems cannot match. The proprietary integration of waste-heat recovery, hydrogen generation, and compact containerized deployment creates a high barrier to entry, making the technology both technically differentiated and defensible for industrial and infrastructure markets. |
MVP |
Pilots |
B2B industrial customers : Our target customers are industries and institutions that require reliable, round-the-clock power but face rising costs and sustainability pressure from diesel use, including: Industrial & Commercial Facilities: mining operations, cement and steel plants, oil & gas installations, chemical factories, and large manufacturing units. Critical Infrastructure: hospitals, data centers, telecom towers, R&D labs, airports, and smart-city utilities that need uninterrupted backup or off-grid power. Government & Defense: PSUs (e.g., SCCL, NTPC), military bases, border outposts, and disaster-prone regions seeking secure clean energy. Geographically we focus on India and the UAE/GCC markets where diesel genset dependence is high and green-hydrogen mandates are emerging. |
Global diesel generator replacement + distributed clean-power market Global DG market ≈ $20–22 B/yr (~₹1.7–1.8 Lakh Cr) with ~7–8% CAGR. India + GCC/ME focus ≈ $6 B/yr (~₹50,000 Cr) for gensets >10 kW. Represents all potential demand for clean, 24/7 distributed power that could switch from diesel over the next decade. |
Industrial & commercial segments where hydrogen/fuel-cell solutions are technically and economically viable in the next 5–7 years High-load industries, critical infrastructure, and remote/off-grid sites in India + UAE/GCC. Estimated $1.5–2 B/yr (~₹12–16,000 Cr) today, growing to $3 B (~₹25,000 Cr) by 2030. Includes power ranges from 30 kW to 1 MW, our current sweet spot. |
Realistic market share Renergizr can capture within 3–5 years Target penetration of 1–2% of SAM through pilots, multi-MW rollouts, and early export projects. $30–40 M (~₹250–330 Cr) in cumulative installations achievable by Year 5 with current scaling plan (approx. 20–25 MW deployed across India and GCC). |
Renergizr follows a hybrid revenue model designed to capture both upfront equipment sales and recurring income streams: Direct Equipment Sales – Sale of modular clean-energy genset units (5 kW–1 MW) to industrial and infrastructure customers. Revenue from hardware + installation + commissioning. Power-as-a-Service / OPEX Model – Renergizr finances, installs, and operates the system at customer sites. Customers pay a per-kWh tariff or monthly service fee, similar to a PPA (Power Purchase Agreement). Generates steady recurring revenue and long-term contracts. Green Hydrogen Sales / Offtake – Surplus hydrogen produced is sold to nearby industrial users (refineries, mobility pilots, chemical plants). Optional buy-back arrangements with government or private offtakers. Annual Maintenance Contracts (AMC) – Service packages for preventive maintenance, monitoring, and stack replacement, typically 2–3% of plant CAPEX per year. This blended approach—hardware margin + recurring energy and service income—creates predictable cash flows, supports rapid scaling, and allows customers to choose CAPEX or OPEX depending on their financial and ESG requirements. |
Kirloskar Cummins Caterpillar Ashok Leyland Sudhir Power Competitive Landscape Summary Renergizr’s clean energy genset: Zero emissions, 80% lower OPEX, 30% higher energy density, no noise. Competitors (Kirloskar, Cummins, Caterpillar, Ashok Leyland, Sudhir Power): High carbon emissions (2.7 kg CO₂/liter), high OPEX, lower energy density, significant noise and pollution. |
Customer Acquisition Strategy for Renergizr Industries 1. Go-To-Market (GTM) Phases Phase 1: Pilots & Proof of Concept (PoC) Deploy 0.1–1 MW systems with early adopters, especially in defense and public sector units (PSUs). Phase 2: Commercial Sales Scale up to 1–10 MW systems with industrial and commercial customers. Phase 3: Nationwide Rollout Expand to 10 MW+ systems across multiple sectors and geographies. [Renergizr-...h Deck (2)] 2. Sales Models CAPEX: Direct sales to PSUs and large corporates. OPEX/Lease: Monthly fee model for mid-size industries. PPA (Power Purchase Agreement): Pay-per-kWh with zero customer CAPEX. [Renergizr-...h Deck (2)] 3. Target Customers Confirmed: SCCL (Singareni Collieries Company Limited) / Ministry of Coal. Targeting: Armed Forces, NTPC, HPCL, BPCL, and other large industrial and commercial entities. Selection Criteria: Focus on organizations with ESG goals and hydrogen presence. [Renergizr-...h Deck (2)] 4. Channels & Partnerships Direct B2B Sales: Engaging directly with industrial and commercial clients. EPC/System Integrator Partnerships: Collaborating with engineering, procurement, and construction firms and electrolyzer OEMs for deployment support. Demonstrations: Showcasing technology via IndianOil, NTPC, and other partners. Government & Research Support: Leveraging support from MNRE, GEDA, UPNEDA, and incubation at IIM Sambalpur and IIT Kanpur. [Renergizr-...h Deck (2)] 5. Marketing & Customer Relationships Customized Solutions: Designing gensets based on specific customer requirements. After-Sales Support: Providing ongoing O&M (operations and maintenance) and free clean fuel for the first 3 months. Brand Building: Participating in trade shows, conferences, and online marketing (company website, LinkedIn). Distributor/Retailer Collaboration: Expanding reach through third-party logistics and distribution partners. [Renergizr-...h Deck (2)] Summary Table: PhaseApproachTarget SegmentChannel/ModelPilots/PoCEarly deploymentsDefense, PSUsDirect, DemoCommercialScale-up salesIndustrial, CommercialDirect, EPC PartnersNationwideBroad rolloutAll sectorsDistributors, PPA Key Takeaway: Renergizr Industries acquires customers through a phased GTM strategy, direct B2B sales, strategic partnerships, pilot projects, and a mix of CAPEX, OPEX, and PPA models—targeting organizations with strong ESG mandates and a need for clean, reliable backup power. |
1. Phased Approach Phase 1: Pilots & Proof of Concept (PoC) Deploy 0.1–1 MW clean energy genset systems with early adopters, especially in defense and public sector units (PSUs). Phase 2: Commercial Sales Scale up to 1–10 MW systems with industrial and commercial customers. Phase 3: Nationwide Rollout Expand to 10 MW+ systems across multiple sectors and geographies. [Renergizr-...h Deck (2)] 2. Sales Models CAPEX: Direct sale to PSUs and large corporates. OPEX/Lease: Monthly fee for mid-size industries. PPA (Power Purchase Agreement): Pay-per-kWh with zero customer CAPEX. [Renergizr-...h Deck (2)] 3. Target Customers Confirmed: SCCL/Ministry of Coal. Targeting: Armed Forces, NTPC, HPCL, BPCL, and other large industrial and commercial entities. Selection Criteria: Focus on organizations with ESG goals and hydrogen presence. [Renergizr-...h Deck (2)] 4. Channels & Partnerships Direct B2B Sales: Engaging directly with industrial and commercial clients. EPC/System Integrator Partnerships: Collaborating with engineering, procurement, and construction firms and electrolyzer OEMs for deployment support. Demonstrations: Showcasing technology via IndianOil, NTPC, and other partners. Government & Research Support: Leveraging support from MNRE, GEDA, UPNEDA, and incubation at IIM Sambalpur and IIT Kanpur. [Renergizr-...h Deck (2)] Summary Table: PhaseApproachTarget SegmentChannel/ModelPilots/PoCEarly deploymentsDefense, PSUsDirect, DemoCommercialScale-up salesIndustrial, CommercialDirect, EPC PartnersNationwideBroad rolloutAll sectorsDistributors, PPA Key Takeaway: Renergizr Industries’ GTM strategy is a structured, phased rollout—starting with pilots, then scaling to commercial sales, and finally expanding nationwide. The approach leverages direct B2B sales, strategic partnerships, and flexible sales models (CAPEX, OPEX, PPA) to target organizations with strong ESG mandates and a need for clean, reliable backup power. |
1. Decarbonizing Off-Grid Power at Scale
Renergizr Industries aims to replace diesel generators in industrial, commercial, and defense sectors with zero-emission, clean energy genset systems powered by waste water and green hydrogen. [Renergizr-...h Deck (2)]
The goal is to provide clean, reliable, and cost-effective electricity to B2B customers, helping them meet ESG mandates and reduce their carbon footprint. [Renergizr-...h Deck (2)]
2. Global and National Impact
The company targets a $35 billion global market and a $3 billion Indian market for clean backup power by 2030, with a focus on rapid scaling and adoption. [Renergizr-...h Deck (2)]
Vision includes nationwide rollout of modular genset systems, expanding from pilot projects to large-scale commercial deployments across India and globally. [Renergizr-...h Deck (2)]
3. Environmental and Social Transformation
Renergizr’s technology is designed to eliminate 99% of CO₂ emissions compared to fossil fuel-based plants, save thousands of trees, and improve air quality in deployment zones. [Renergizr-...h Deck (2)]
The company is committed to supporting India’s goal of carbon neutrality by 2070 and contributing to the United Nations Sustainable Development Goals (SDG 7: Affordable and Clean Energy, SDG 13: Climate Action). [Renergizr-...h Deck (2)]
4. Technological Leadership
The vision includes continuous innovation in clean energy, leveraging patented technology, AI-based monitoring, and smart grid integration. [Renergizr-...h Deck (2)]
Renergizr aspires to be a leader in circular resource use (wastewater + hydrogen), and to integrate green hydrogen and geothermal energy for round-the-clock clean power. [Renergizr-...h Deck (2)]
5. Economic and Industrial Growth
By promoting indigenous manufacturing and job creation, Renergizr aims to support the Make in India initiative and reduce India’s dependence on imported fossil fuels. [Renergizr-...h Deck (2)]
The company’s business model is built for sustainable growth, with diversified revenue streams from genset sales, leasing, power purchase agreements, and carbon credits. [Renergizr-...h Deck (2)]
In summary:
Renergizr Industries’ long-term vision is to drive a global transition to clean, decentralized, and sustainable energy—replacing polluting diesel gensets with innovative, zero-emission solutions that benefit the environment, economy, and society. |
Registered Private Limited |
Rs 20 Lakhs |
Yes |
We are applying to the Avinya Cohort because it is the ideal platform to accelerate Renergizr Industries’ mission of decarbonizing off-grid and backup power in India and beyond, using AI-driven, patented clean energy genset systems. AI at the Core: Our technology leverages AI for predictive maintenance, remote monitoring, and smart grid integration—directly aligning with Avinya’s focus on AI-based startups in Karnataka. Scalable Impact: With a vision to replace diesel generators in industrial, commercial, and defense sectors, we are ready to scale from successful pilots to commercial rollouts. Avinya’s curated incubation, mentorship, and network can help us rapidly expand our reach and impact. Collaborative Ecosystem: We value Avinya’s emphasis on a collaborative, ethical, and productive environment. Our team is committed to knowledge sharing, responsible AI, and contributing to Karnataka’s innovation ecosystem. Market Access & Validation: The program’s connections to industry, government, and research institutions will help us validate our technology, access early adopters, and refine our go-to-market strategy for the Indian and global markets. Alignment with Karnataka’s Vision: As a Karnataka-based deep-tech startup, we are eager to contribute to the state’s leadership in clean energy and AI, supporting both economic growth and sustainability goals. In summary: Avinya offers the resources, mentorship, and ecosystem we need to accelerate our growth, maximize our positive impact, and establish Renergizr Industries as a leader in AI-powered clean energy solutions. |
During the Avinya Cohort, we want to: Accelerate Commercialization: Move from successful pilots to large-scale commercial deployments of our AI-powered clean energy genset systems, especially in industrial and defense sectors. Strengthen AI Capabilities: Refine and scale our AI-driven predictive maintenance, remote monitoring, and smart grid integration features with expert mentorship and technical support. Validate Product-Market Fit: Engage with Avinya’s network of industry partners, mentors, and early adopters to validate our solution, gather feedback, and optimize our go-to-market strategy for Karnataka and beyond. Build Strategic Partnerships: Connect with government agencies, corporates, and research institutions to accelerate adoption, secure pilot projects, and explore joint development opportunities. Enhance Team & Business Model: Leverage Avinya’s resources to strengthen our team, improve business processes, and refine our revenue models (CAPEX, OPEX, PPA, carbon credits). Contribute to Karnataka’s AI & Clean Energy Ecosystem: Share our learnings, collaborate with fellow startups, and support the state’s vision for sustainable, AI-driven industrial growth. In summary: We aim to emerge from the Avinya Cohort as a market-ready, scalable, and investible AI-powered clean energy company, ready to make a significant impact on decarbonizing backup power in India and globally. |
Virtual |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/OlyOmA0XhxrX-Renergizr-Industries-Pvt-Ltd-Himanshu-Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/OlyOmA0XhxrX-Renergizr-Industries-Pvt-Ltd-Himanshu-Pitch-Deck.pdf |
https://drive.google.com/file/d/1d3Dy4ZYDTeKva0XMKHeBwFArZSZdUM0v/view?usp=sharing |
Absolutely—we are building a mission-driven and impact-focused startup.
Renergizr Industries was founded with the core mission to decarbonize off-grid and backup power for industrial, commercial, and defense sectors by replacing polluting diesel generators with our patented, AI-enabled clean energy genset systems. Our technology generates zero-emission electricity using waste water and green hydrogen, directly addressing urgent global challenges like climate change, air pollution, and energy security. [Renergizr-...h Deck (2)]
Our impact focus is threefold:
Environmental Impact:
Our solution eliminates 99% of CO₂ emissions compared to fossil fuel-based plants, removes 600 metric tonnes of greenhouse gases per MW, and saves 1,200 trees per MW system annually.
We promote circular economy principles by utilizing waste water and reusing waste heat, and our systems operate with the lowest carbon and land footprint among renewable energy technologies. [Renergizr-...h Deck (2)]
Economic & Social Impact:
We create direct and indirect employment, with a focus on inclusivity (40% of our skilled workforce is from ST/SC/OBC communities).
Our technology reduces India’s dependence on imported fossil fuels, saving foreign exchange and supporting the “Make in India” initiative.
By improving air quality and reducing respiratory illnesses, we directly enhance community health and well-being. [Renergizr-...h Deck (2)]
Alignment with Global Goals:
We are committed to supporting India’s carbon neutrality target by 2070 and contributing to the United Nations Sustainable Development Goals (SDG 7: Affordable and Clean Energy, SDG 13: Climate Action).
In summary:
Renergizr Industries is not just a business—we are a purpose-driven movement to deliver scalable, sustainable, and inclusive clean energy solutions that create measurable positive impact for the environment, economy, and society. |
NA |
IPTIF - IIT Palakkad |
checked |
| Sep 28, 2025 @ 10:44 AM |
Pradeep Singh |
pradeep@devalayas.com |
https://www.linkedin.com/in/pradeep-singh-9b60a8180/ |
http://devalayas.com |
9448144311 |
Pradeep Singh – Founder & CEO Pradeep brings over 35 years of experience in IT, project management, and strategic leadership. He has led large-scale technology deployments including nationwide university examinations and rural broadband connectivity projects impacting 3,000+ Gram Panchayats. As Founder & CEO of Anek Technologies Pvt Ltd, he drives vision, partnerships, and growth for Astro Devalaya. Under his leadership, the company is backed by CSC e-Governance, the Government of Karnataka (Muzrai Dept.), and has won the Elevate 2023 award. |
NA |
Yes |
NA |
Anek Technologies Pvt Ltd |
www.anektech.com |
|
AI-enabled platform connecting devotees with temples, pujas & prasadam online |
Millions of devotees across India and abroad face challenges in accessing their ancestral Kuladevata and Gramadevata temples due to geographical distance, lack of time, physical limitations, or financial barriers. Even when they wish to perform pujas or receive prasadam, they often depend on intermediaries or unreliable arrangements, leading to lack of trust, transparency, and continuity in spiritual practices. Temples, on the other hand, struggle with digital adoption, efficient service delivery, and reaching devotees beyond their immediate geography. This leads to underutilization of temple resources, declining engagement with younger generations, and reduced economic sustainability for temple ecosystems. We solve this problem by creating a digital bridge between devotees and temples. Through Astro Devalaya, devotees can book online pujas, prasadam delivery, and astrology consultations directly from authentic temples. Partnered with CSC e-Governance and the Muzrai Department, Govt. of Karnataka, our platform ensures trust, transparency, and inclusivity. By digitizing lesser-known Kuladevata and Gramadevata temples, we preserve spiritual heritage while making rituals accessible globally. For temples, we enable new revenue streams and digital reach, strengthening local communities. In short, we solve the accessibility gap for devotees and the sustainability gap for temples, using technology as a medium to revive and sustain India’s rich spiritual ecosystem. |
Devalaya is a SpiritualTech platform that digitally connects devotees with their temples and rituals. Through our mobile app and website, devotees can seamlessly book online pujas, homas, and prasadam delivery directly from authentic temples. Personalized astrological consultations and a curated Astro-Mall for devotional items further enhance the experience.
We ensure authenticity and trust through formal partnerships with CSC e-Governance and the Muzrai Department, Govt. of Karnataka, along with direct collaborations with temple authorities. Logistics is streamlined via India Post, enabling prasadam delivery to devotees worldwide.
Our platform uniquely digitizes Kuladevata and Gramadevata temples—often inaccessible or lesser-known—preserving cultural heritage while making spiritual services accessible at scale. For devotees, this means convenience, transparency, and the ability to stay connected with their roots from anywhere in the world. For temples, we provide a digital revenue model, increased outreach, and community engagement.
Future-ready features like AI-enabled personalization, multilingual access, and global temple onboarding will ensure Astro Devalaya becomes the one-stop spiritual ecosystem for millions of devotees worldwide. |
Devalaya is unique because it blends technology, authenticity, and government partnerships to solve a centuries-old challenge. Unlike generic astrology or e-commerce apps, we focus on Kuladevata and Gramadevata temples, which are deeply personal and culturally significant but largely underserved in the digital space. Our defensibility lies in: Official recognition & backing – Partnerships with CSC e-Governance and the Muzrai Department, Govt. of Karnataka create strong entry barriers. Trust & authenticity – Direct tie-ups with temples and verified priests ensure rituals are performed genuinely, building credibility with devotees. Integrated ecosystem – We combine puja bookings, prasadam logistics, astrology, and devotional commerce in one platform, making us a comprehensive solution rather than a niche service. Scalability – Technology-driven model with India Post logistics allows seamless expansion across India and global Hindu diaspora. Cultural moat – By digitizing ancestral temples and localized traditions, we tap into a highly emotional, non-replicable segment that creates long-term stickiness. Together, these factors make Astro Devalaya not just another online service, but a defensible, impact-driven platform with deep cultural, social, and technological value. |
Users |
Users, Signups, Testimonials |
Our target customers are devotees aged 25–65, living in urban and semi-urban areas across India and the global Hindu diaspora. These are individuals and families who are tech-savvy yet culturally rooted, seeking convenient and trustworthy access to their Kuladevata and Gramadevata temples, pujas, prasadam, and astrology services. We specifically cater to: NRIs and global Hindus who wish to stay connected to their ancestral temples despite distance. Working professionals and families with limited time for travel but strong devotion. Senior citizens and physically challenged devotees who face mobility issues. Temples that need digital reach, revenue sustainability, and connection with scattered devotees. This audience values authenticity, convenience, and cultural continuity, making Astro Devalaya their trusted spiritual bridge. |
$52 billion |
10 billion |
$50–100 million |
Astro Devalaya follows a multi-stream revenue model designed for sustainability and scale: Online Pujas & Rituals – Devotees book temple-specific pujas through our platform; we charge a service fee/commission per booking. Prasadam Delivery – Prasadam is delivered via India Post logistics; revenue generated through margin on each prasadam order. Special Homas & Assist Pujas – Premium rituals (Chandi Homa, Taila Abhisheka, etc.) booked at higher ticket sizes, with shared revenue between temple and platform. Astrology Services – Personalized consultations offered on a pay-per-session basis. Subscription Plans – Devotees can subscribe for monthly/annual packages covering recurring pujas, prasadam, or astrology. Astro-Mall (Devotional E-commerce) – Commission on sales of temple-related items (rudraksha, idols, puja kits, etc.). This diversified model ensures steady cash flow, scalability, and long-term defensibility while directly benefiting temples and devotees alike. |
Direct Competitors SmartPuja, ePuja, OnlinePrasad – Platforms offering online puja bookings and prasadam delivery. AstroTalk, AstroYogi – Astrology-focused apps with limited temple integration. Indirect Competitors E-commerce platforms (Amazon/Flipkart – Devotional category) – Selling puja kits & idols but without temple authenticity. Local agents/intermediaries – Traditional but unorganized, lacking transparency or digital convenience. |
Astro Devalaya acquires customers through a multi-channel strategy: Government-backed networks – Integration with CSC e-Governance (4 lakh+ VLEs) and Muzrai Department temples gives us direct access to millions of devotees. Temple Partnerships – Onboarding Kuladevata & Gramadevata temples ensures trust; devotees discover and adopt our platform via their own temple. Digital Marketing – Targeted campaigns on social media, SEO, and festival-driven content to engage tech-savvy devotees in India & abroad. Word of Mouth & Community Trust – Religious rituals and prasadam inherently create repeat usage and organic referrals. NRI & Diaspora Outreach – Partnerships with global Hindu associations and diaspora communities to tap international devotees. Subscription & Loyalty Programs – Encourage retention by offering recurring puja packages and exclusive benefits for frequent users. This mix of institutional tie-ups, grassroots temple presence, and digital-first outreach ensures both scale and sustainability. |
Our go-to-market strategy focuses on temple-led adoption and government-backed scale: Temple Onboarding (Supply-Side First) Digitize 150–300 Muzrai temples in Karnataka (covering ~90% of the state’s spiritual population). Extend to Kuladevata & Gramadevata temples, ensuring strong cultural connect. Government Partnerships (Trust & Distribution) Leverage CSC e-Governance (4 lakh+ VLEs) and Muzrai Department tie-ups for instant credibility and rural-urban penetration. Devotee Acquisition (Demand-Side Growth) Target festivals & auspicious days with puja/prasadam campaigns (e.g., Shravana, Navratri). NRI Outreach through diaspora associations, social media, and temple events abroad. Retention & Expansion Launch subscription packages for recurring pujas & prasadam. Introduce Astro-Mall & astrology services to deepen engagement. Use India Post logistics for nationwide & global prasadam delivery. This dual strategy of building trust via temple & government partnerships and driving adoption via digital & diaspora outreach ensures scalable and defensible market entry. |
Our long-term vision is to make Astro Devalaya the global spiritual gateway—a one-stop platform where every Hindu devotee, regardless of geography, can stay connected to their ancestral temples, rituals, prasadam, and astrology.
We aim to:
Digitize 10,000+ temples across India, with a special focus on Kuladevata & Gramadevata temples.
Build a trusted global ecosystem for pujas, prasadam, astrology, and devotional commerce.
Empower temples with digital infrastructure & sustainable revenue models, strengthening local communities.
Serve the 300M+ global Hindu diaspora, giving them seamless access to authentic traditions.
Become the leading SpiritualTech brand backed by AI, multilingual access, and cultural depth.
In essence, our vision is to preserve and modernize India’s spiritual heritage while making it universally accessible, sustainable, and future-ready through technology. |
“Anek Technologies Pvt Ltd, incorporated on 12th November 2021, registered in Karnataka, India. Recognized as a DPIIT Certified Startup (Certificate No. DIPP101384), registered under MSME, Startup India & Startup Karnataka.” |
NA |
Yes |
We are applying to Avinya because our mission aligns with your vision of backing founders with strong market fit, unique differentiation, and scalable impact. Astro Devalaya is India’s first SpiritualTech platform, digitizing ancestral temples, pujas, and prasadam with government partnerships (CSC e-Governance & Muzrai Dept). With early traction and a culturally defensible moat, we are ready to scale across India and the global Hindu diaspora. Avinya’s strength in early-stage support, mentorship, and community-driven growth makes it the right partner to accelerate our journey from a Karnataka-based startup to a nationwide and global leader in this space. We seek Avinya’s ecosystem to help us sharpen our go-to-market, unlock capital access, and build at scale, creating a transformative category and long-term socio-cultural impact. |
During the Avinya programme, we aim to: Refine our business model & financials – sharpen unit economics, revenue streams, and scalability plan. Strengthen go-to-market strategy – accelerate temple onboarding, build NRI outreach, and optimize digital marketing. Leverage Avinya’s mentor network – gain insights on fundraising, compliance, and investor readiness. Access capital & investor connects – prepare for our ₹2 Cr fundraise to expand beyond Karnataka. Build community partnerships – collaborate with like-minded founders and ecosystem partners to drive growth. Scale operations & tech – enhance our platform with AI, multilingual features, and seamless logistics. In short, we want to emerge from the programme as a funding-ready, scalable SpiritualTech venture, equipped with the right strategy, network, and tools to expand pan-India and globally. |
We kindly request permission to participate virtually, as we have a lean team managing multiple responsibilities. Given our multitasking requirements, it would be challenging to be physically present throughout, but we assure you of our full commitment and active engagement in all sessions online. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/bYWr5cGqJDnl-ASTRODEVAALAYA-VC.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/bYWr5cGqJDnl-ASTRODEVAALAYA-VC.pdf |
https://www.youtube.com/watch?v=MJMTjNfCvHo |
Yes, Devalaya is a mission-driven and impact-focused startup. Our mission is to digitally revive and sustain India’s spiritual heritage by connecting devotees worldwide with their Kuladevata and Gramadevata temples through online pujas, prasadam delivery, and astrology.
The impact we create is two-fold:
For devotees: We remove barriers of distance, time, and mobility, ensuring that millions—including NRIs, working families, and senior citizens—can stay connected to their roots.
For temples & communities: We enable temples to adopt digital infrastructure, generate sustainable revenue, and strengthen local economies by engaging priests, vendors, and logistics partners.
Backed by CSC e-Governance and the Muzrai Department, Govt. of Karnataka, we are not just building a business but a cultural movement that preserves traditions, empowers communities, and makes spirituality inclusive and accessible in the digital age. |
Yes. We are based out of Belagavi, a Tier-2 city in North Karnataka, which is an underrepresented region in India’s startup ecosystem that is largely concentrated in metro hubs like Bengaluru, Delhi, and Mumbai. Building a DPIIT-recognized and government-backed startup from this region reflects our commitment to proving that innovation and impact can emerge beyond metros. Additionally, by working at the intersection of spirituality and technology (SpiritualTech), we represent a niche and underexplored sector within the startup ecosystem, creating new opportunities for cultural, social, and economic impact. |
Machayya C M (GINSERV Mentor) |
checked |
| Sep 28, 2025 @ 5:09 AM |
Shreyas Iyer, Udit Jindal |
shreyas@admitra.in |
https://www.linkedin.com/in/shreyasiyer14/ |
https://admitra.in |
+91 9782719420 |
Shreyas: Co-Founder (Ex-Engineering @Google), Udit: Co-Founder (Ex-Engineering @Indeed), Mihir: Founding Engineer (Engineering @Sprinklr) |
Shreyas and Udit have known each other for 10 years. They were batchmates at IIIT Bangalore and have collaborated on multiple projects since. They are living in the same apartment after graduating from university. They've witnessed each other's coding meltdowns, questionable life choices and also career growth in their respective companies. Basically, they're battle-tested co-founders.
Mihir and Shreyas are childhood friends, having known each other since their JEE coaching days. This long-standing friendship adds another layer of trust and collaboration to the founding team.
For the past 4 months, both founders have been working together on AdMitra, initially building it alongside their full-time jobs at Google and Indeed respectively. They have now transitioned to working full-time on AdMitra. |
Yes |
Google, Indeed, and Sprinklr—where the team has worked—are all heavily invested in advertising. This common ground helped the team deeply understand the true potential of the industry, spot gaps in the current landscape, and see the real opportunity to solve these with advanced AI-powered solutions.
Shreyas has always been curious hard-core engineer, working on scalable architecture and complex systems. Additionally, he has always been passionate about video games, and, since college, worked on developing his own rendering engines. This background makes video generation and creative ad-tech a natural extension of his interests and skills, providing unique technical depth for our AI-driven creative solutions.
Udit comes from an SMB family and understands the pain of advertising for SMB users in India, who are price sensitive and not always tech savvy. Being from a business community, he has a knack for talking to and connecting with business owners, which helps him spot real gaps and needs. He's always been curious about product, and is an avid reader of the business world. He brings a product-focused mindset, connecting how technology can help solve real-world problems.
Mihir, our founding UI engineer, offers top-tier design and engineering leadership (Director at Sprinklr). Having worked at Sprinklr since pre-IPO, he brings insider knowledge of how large-scale marketing enterprise solutions operate and deep expertise in crafting exceptional user experiences that scale.
All in all, our biggest strength is the combination of deep advertising industry insight, engineering excellence, and direct SMB empathy. We understand how AI can revolutionize creative workflows for businesses, not just in theory, but from real market feedback. |
AdMitra |
https://admitra.in |
|
A SaaS platform that enables brands to easily create co-branded, hyperlocal ads for all their local retail partners and outlets. |
Brands today don’t use their dealer network for hyperlocal advertising—even though dealers are the final and most trusted touchpoint in the sales cycle. These dealers lack digital support, leaving a gap where local advertising potential is wasted and dealers struggle to connect with their customers online. |
AdMitra lets brands instantly create hyperlocal video and poster ads for their entire dealer network—each ad carries the dealer’s own identity, making it truly local, lovable and shareable. It is India’s first brand-to-dealer hyperlocal advertising platform, giving national brands a scalable way to activate their dealers with co-branded video and poster campaigns, with video as the standout USP.
How It Works:
Brand Side:
- Campaign Configuration: Brands define campaign goals, target regions and creative strategy.
- Scene Library Creation: Brands upload their high-quality video assets, which our AI decomposes into a library of voiceless, semantically-tagged scenes. In parallel, our AI Scene Creator can generate entirely new scenes aligned with the campaign’s objectives and target geographies. These reusable scenes become the backbone of the campaign: later we add different narrations and dealer details to create hundreds of unique ads at a fraction of traditional costs.
Dealer Side:
- Through a simple mobile app, dealers add their store name, address and local photos or clips, then select the target audience, preferred language and campaign. AdMitra’s AI Director understands the dealer’s location, target audience and language and automatically blends the brand’s polished scenes with the dealer’s authentic content to generate co-branded video and poster ads that feel truly local—carrying the dealer’s own identity and encouraging them to proudly share.
This centralised, AI-driven process cuts per-video costs by 70–80% and unlocks a free distribution channel for brands—their dealers themselves, to share hyperlocal ads, who function as nano-influencers with 500–5,000 personal contacts—giving brands authentic, scalable reach while maintaining brand consistency. |
AdMitra’s defensibility comes from the depth of its AI engine and the network effects it creates. The AI Scene Creator goes beyond simply slicing existing videos—it can generate entirely new, high-quality scenes tailored to each campaign’s goals and regions, giving brands a constantly expanding library without the cost or time of fresh shoots. The AI Director then understands each dealer’s location, audience and language to assemble polished, hyperlocal ads that stay perfectly on-brand. This dual AI system makes it possible to produce and refresh thousands of dealer-specific ads in days instead of months—something agencies or basic editing tools simply can’t match. Dealers, in turn, act as trusted nano-influencers with hundreds to thousands of personal contacts, creating authentic reach that grows with every new dealer. The result is a self-reinforcing advantage: 70–80% lower production costs, 2–4× higher engagement, and a hyperlocal distribution channel that becomes stronger as the network expands. |
Users |
Pilots |
Brands that sell through dealer or distributor networks—such as appliances, automotive or building materials—looking to run consistent, hyperlocal digital campaigns at scale. |
$3-5B / year |
$1-2B / year |
$20-50M / year |
Pay per creative |
There are no direct competitors yet, who are working on creating hyperlocal, co-branded ads at scale |
Initially we are trying to get warm connects with marketing agencies and brands to use or SaaS platform. |
We are starting by reaching out to marketing agencies and offering our product as a SaaS solution to help them serve their existing brand clients. Marketing agencies usually manage brands that don’t have large marketing teams, and agencies also have more appetite for trying new products. At the same time, we are also directly contacting brands who're looking to experiment with marketing-tech. |
Our long-term vision is to expand our platform so brands can not only provide creatives to their dealer networks, but also actively manage and optimize dealer-level marketing. Currently, we deliver creatives to dealers and enable easy auto-sharing across platforms, supporting dealers who may not be very tech-savvy. Going forward, we plan to empower brands to set marketing budgets at the dealer level and use our platform to optimize ad delivery for each dealer, identifying the best channels for inorganic promotion so that every dealer gets targeted, boosted exposure.
At present, we create hyperlocal co-branded ads tailored for brand dealers, where the expected ad quality can be lower since these are not national-level campaigns. In near future, once our creative engine will be able to reliably generate television-quality ads that meet national brand standards, we will enable brands to design and run distinct hyperlocal ads for themselves. This will allow brands to serve different ads in different localities, delivering highly personalized advertising campaigns at scale. |
Incorporated as AdMitra Private Limited |
NA |
Yes |
We are applying to Avinya because it offers a unique combination of funding, world-class mentorship, and access to a powerful network of industry leaders and IIT alumni—all tailored for AI startups. The program’s focus on structured guidance, founder well-being, and real-world pilot opportunities is perfectly aligned with our needs as we scale our platform. We believe that participating in Avinya will help us refine our product, accelerate our go-to-market execution, and connect with potential partners and investors in the AI ecosystem. |
During the programme, we want to 1) Validate our solution with pilot customers and partners, leveraging the mentorship and feedback from industry experts. 2) Strengthen our product offering and enhance our AI capabilities for creative ad generation. 3) Build connections with agencies, brands, and the broader AI startup community to drive early adoption. 4) Improve our operational readiness and pitch for follow-on funding to support future growth |
Yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/QnBgvhCx6XXX-AdMitra-Hyperlocal-Dealer-Advertising-at-Scale.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/QnBgvhCx6XXX-AdMitra-Hyperlocal-Dealer-Advertising-at-Scale.pdf |
https://storage.googleapis.com/admitra-tools/AdMitraDemo.mp4 |
We are a mission-driven startup. Our journey began with a clear focus on empowering small and medium businesses (SMBs) through automated advertising—this vision is reflected in our original SMB-focused platform (as seen at https://admitra.in/admitra-smb). We wanted to make ad creation easy and accessible so that every SMB could promote itself without needing a marketing team or agency.
Over time, we realized that directly serving SMBs is challenging because they tend to be highly price-sensitive and less open to adopting new technology. Onboarding them requires significant capital and hands-on education, which is resource-intensive. To overcome these barriers and continue serving the SMB sector, we adapted our model so that brands pay for their dealers’ ads and manage creative strategy. This drastically reduces both the financial and learning burden on SMBs, while still helping them grow.
In the long term, as our platform matures and we gather more resources, we plan to return to our founding mission: offering SMBs a direct, simple, and affordable ad solution—where creating campaigns is just a few clicks away and becomes a true business utility, not a creative hurdle. Our purpose remains unchanged: making high-quality advertising and growth accessible to every small business. |
NA |
Praseeda Kalkur (works at ACSEL) |
checked |
| Sep 27, 2025 @ 11:07 PM |
Rathishchandra R Gatti , Rohanchandra R Gatty, Ajith B Suresh, Navaneeth Bhaskar |
airatsystemsllp@gmail.com |
http://linkedin.com/company/airat-systems/?originalSubdomain=in |
http://www.airatsystems.com |
9481847572 |
Dr.Rathishchandra R. Gatti, Role: CEO- “Why & Where” Formulate mission, strategy, and execution; aligns research with a winnable market (e.g., RaaS, HW+SaaS). Raises capital, sets milestones, manages board/investors, and allocates resources across research, product, and go-to-market. Decides “build vs. partner,” regulates risk posture (AI safety, robotics compliance), and picks initial beachhead use-cases. Key KPIs: runway & growth, product-market fit, gross margin path, hiring quality, strategic partnerships. Background: BE from NITK, MBA,PGDGSM ,PhD from Curtin University, Australia.in mechatronics domain, 20 years of industrial experience from General Electric,Meritor Automotive, Rapsri ,Curtin University and Sahyadri College of Engineering & Management. has 17 patents,40+ research publications and 8 books in robotics and AI.His research focuses on Physical AI, Reinforcement learning, Robotics and Agentic AI. Dr.Navaneeth Bhaskar,CTO Role -CTO “What & How (Tech)” Sets technical roadmap across AI/ML, perception/control, simulation, and embedded systems; balances research vs. production. Owns architecture, MLOps, data pipelines, reliability, security, and IP/patent strategy. Establishes evaluation protocols (accuracy, latency, safety), and leads model/robot validation (sim-to-real). Key KPIs: model/stack performance (accuracy, latency), deploy frequency, reliability (MTBF/uptime), security incidents, IP created. Background: from Ph.D. in Biomedical Signal Processing and Machine Learning from Vellore Institute of Technology (VIT) and a B.E. & M.Tech. in Electronics from Visvesvaraya Technological University. His research focuses on Artificial Intelligence, Machine Learning, Data Analytics, IoT, and Signal Processing. Dr.Ajith BS, Role - CMO “Who & How to Reach” Defines ICP/segments, positioning, pricing/packaging (hardware, software, service), and demand generation. Builds pipeline via partnerships, content, events, community/dev-relations (if platform/SDK), and sales enablement. Closes feedback loops from market to product/research; owns brand and category creation. Key KPIs: qualified pipeline, CAC:LTV, win rate, activation/retention, NPS/case studies. Background:B.E-Rajeev Institute of Technology, Hassan,M.Tech- VTU PG Studies, Mysore,Ph.D- VTU RRC, Belagavi- Research Focus is on Agricultural Robotics and Energy Automation. Dr.Rohanchandra R Gatty Role - COO — “Make it Work at Scale” Runs operations: manufacturing/supply chain, vendor selection, field deployment, maintenance, and customer success. Drives certifications and safety compliance (e.g., IEC 61508/ISO 10218 as applicable), QA, and incident response. Optimizes unit economics and delivery: from prototype → pilot → fleet scale; manages SLAs. Key KPIs: on-time delivery, cost/unit & gross margin, uptime/SLAs, safety incidents, install time, churn. Background: MBBS from VIMS, MS in general Surgery from MMC,Mysore, McH in Surgical Oncology,30+ years of experience in surgical oncology. has operated about 15000 patients. His Research focuses on Surgical methods, Oncology and Applied AI in Oncology. |
Ajith, Rathish and Navaneeth were colleagues working on Oncoinsights—a drug-drug interaction app—with Rohan ,the oncosurgeon and managing partner at MIO. Then they decided to have an AI startup. There was one more product.—a laparoscopic surgical simulator using computer vision and RL. This led us to build an AI and robotics startup—AIRAT SYSTEMS LLP at NITK STEP. |
No |
Complementary skills
Ajith is great with networking and patents.
Rathish is very good with product design and AI for physical layers
Navaneeth is very good with software design and computer vision
Rohan is very good with functional expertise of medical requirements for medical AI and robotics |
AIRAT SYSTEMS LLP |
www.airatsystems.com |
|
An Deeptech AI and Robotics startup |
Small langauage Model based Agentic Robots |
We are currently building customizable and modular small language model-based agentic robots that can be used across industries in a secure and private offline mode. Unlike software robots that are used in web based apps, these are physical robots that have multimodal LLM capabilities. Typical applications we are envisaging are customer service robots in hospitals and polyclinics. |
Many robots use online servers and LLMs thereby compromising on data privacy. We are building edge robots that have localized LLM thus workable even in remote areas. training and inputs, |
MVP |
Users |
Hospitals, Clinics, Nursing Homes |
NA |
NA |
NA |
Product B2B, SAAS for Ai software |
AcYut, Unitree |
Product Demonstrations, mangalore city has several hospitals |
Target Hospital chains, local hospitals first |
To be a market leader in Edge robotics and AI |
MCA Registered,Startup India Registered, MSME Udhyam registered |
NA |
Yes |
To be part of the Avinya Accelerator Program |
traction to our existing startup and funding for product development |
yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/EfZtJbuDwCEp-Pitchdeck_AIRAT_Satva.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/EfZtJbuDwCEp-Pitchdeck_AIRAT_Satva.pdf |
https://www.youtube.com/watch?v=pOFcwcwtv3k |
We are a mission driven startup with values - Our mission is to make the edge AI based robotics solutions very affordable for local hospitals and even clinics for B2B. We are also envisaging to build edge Ai based personal patient healthcare robots for B2C. |
NA |
Dr.Subraya Hegde , Director -NITK STEP , NITK Surathkal |
checked |
| Sep 26, 2025 @ 5:29 PM |
Chaitali Diwan & Chittaranjan Savadi |
chaitali@kazunov1ai.com |
https://www.linkedin.com/in/chaitali-diwan/ |
https://www.kazunov1ai.com |
9071256378 |
Dr. Chaitali Diwan : Ph.D. (AI) Gold Medal, M. Tech Data Sci, Technology - 14 yrs : Responsible for Technology & AI; Chittaranjan Savadi : PGDM IIM Lucknow; 2 decades Insurance experience; responsible for Product, Business & Strategy |
Chaitali and I are more than just co-founders in our startup; we're partners in life, for over 15 years!
Our combined strengths are highly complementary: one brings cutting-edge AI and technology knowledge, while the other offers extensive insurance domain expertise and the ability to attract additional specialist input. Our collaborative deep-thinking sessions have fostered innovative product ideas and guided us to valuable expert insights. |
Yes |
Chaitali, a gold medallist with a Ph.D. in AI and an M.Tech in Data Science from IIIT Bangalore, brings a decade of experience in technology and development from her time at Samsung Research, CenturyLink, InsurFin, and Accenture. Her expertise encompasses AI, Machine Learning, technology, and product development. As a founding team member at InsurFin, she also has proven experience in building and mentoring high-performing startup teams. Chaitali will lead our efforts in AI, technology, and product/process design.
With over 20 years in the Indian insurance industry post-MBA (IIM Lucknow), I've held senior roles in Product, Marketing, and Distribution at ICICI Lombard and Pramerica Life. This has given me a deep understanding of the insurance domain, especially health and life insurance, and a robust industry network. I will oversee Finance, Subject Matter Expertise, Operations, and Business Development.
Our combined strengths are highly complementary: one brings cutting-edge AI and technology knowledge, while the other offers extensive insurance domain expertise and the ability to attract additional specialist input. Our collaborative deep-thinking sessions have fostered innovative product ideas and guided us to valuable expert insights. |
Kazunov 1AI |
https://www.kazunov1ai.com |
|
We automate the health insurance underwriting and health claims adjudication process, with a unique risk assessment model for the customers for underwriting & an industry first treatment verification engine. |
The Health Insurance Operations space (underwriting & claims adjudication) is a manual process, and involves assessment and evaluation of multiple documents, and workflow is a linear manual process. The OPEX cost for this is 5%+. The manual processes adds to the processing time, & transaction processing costs for insurers. The manual oversight and workflow leads to error rate, frauds, which has implications for the industry. As per reports, ~5% - 7% premiums losses are in wrong risk and fraud, and also leads to high customer dissonance. |
Our solution automates the underwriting and claims adjudication process, and also builds in tools to identify fraud, and wastage in the process. We are building a unique risk assessment model for the customers, with specific workflows to manage for scale, and right profiling of customers. On the claims adjudication side, we are building an industry first treatment verification engine to weed out unrelated expenses in a claim. Currently, the product is under development. |
We are building a unique risk assessment model for the customers, with specific workflows to manage for scale, and right profiling of customers. On the claims adjudication side, we are building an industry first treatment verification engine to weed out unrelated expenses in a claim. Currently, the product is under development. |
MVP |
Testimonials |
Insurance Companies, HealthCare Administrators (TPAs), Healthcare Fintechs |
USD 900 mn |
USD 300 mn |
NA |
Transaction Fees |
Vitraya, NVest, Metamorphosys |
We sell to Insurance Companies to identifying gaps in their process, and positioning our product solution, which will reduce their processing time and cost, with greater accuracy |
We have identified 3 companies basis on vintage, openness to innovate, and have pitched them. We are in the process to getting PoC active with 1 user in this quarter. Post 2 successful PoCs, we want to read |
We want to build GenAI based solutions in insurance Underwriting and Claims adjudication, and be the leading player in the APAC region |
Yes |
NA |
Yes |
We are looking for relevant mentorship and workshops which we will help improve our product |
We want to build product with strong feedback loops, and improvement basis a strong bouncing board |
Yes, open to Hybrid |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/QQxsV4uasGnr-Kazunov1AI-Pitch.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/QQxsV4uasGnr-Kazunov1AI-Pitch.pdf |
https://www.loom.com/share/edafc929d3874968a4e30274e9782f0c?sid=b9f34ef9-4991-425d-b7ac-18c46878af01 |
We are building in Insurance, with the focus wherein customers with good health will get insurance at a better rate, whcih si currently not the case |
Yes, Women led startups |
Praseeda K |
checked |
| Sep 25, 2025 @ 2:49 PM |
Dr. Deepak K T , Mr. Mahidhar Sastry |
deepak@iiitdwd.ac.in |
https://in.linkedin.com/in/deepak-k-t-92270a9 |
https://deepakthotappa.wixsite.com/mysite |
9900456936 |
CEO, Vocab.AI, Associate Dean (R&D) and CEO IDRP, Department of ECE , IIIT Dwd, CO-FOUNDER, Wyzmindz Solutions, Bangalore |
We are long lasting friends since 30 years, have worked together since 2023 on bringing our expertise together to solve problems pertaining to customer experience in the CRM processes using AI based analytical framework. |
No |
At Vocab.AI, we engineer cutting-edge products designed to revolutionize call center operations and significantly elevate the customer experience through advanced audio, text, and image analytics. Our team is proficient in AI that can handle audio, text, and image analytics, and stand out as leaders in what's known as multimodal AI. Our team excels at processing and understanding information from various data types simultaneously, allowing for richer insights and more human-like interactions.
The team comprises motivated and seasoned computer science graduates, who have a commendable track record of building industry grade solutions. |
VOCAB.AI |
www.vocab-ai.com |
|
Vocab.AI is committed to transform the business ecosystem and improve customer experience while harnessing the power of AI's generative and analytical capabilities. |
Vocab.AI builds advanced AI solutions for improving customer experience interacting with agents at call centers. Businesses struggle with fragmented Customer Relationship Management (CRM) due to the lack of automated and integrated analytics capabilities, leading to gaps in delivering high-quality customer experience. The interactions between customers and customer-care executives incurs volumes of multimodal data pertaining to various issues, on a daily basis. It is critical for customer retention and revenue generation, that these conversation must adhere to the prescribed quality guidelines issue for the process. Quality auditing of these interactions is a tedious task which is also quite restraining financially. Involving humans in quality auditing of such interactions yield a subjective perspective. Furthermore, call centers also suffer with the problem of high attrition rate, and low volume of well-trained and proficient agents who can yield a significant throughput. It is therefore critical to devise solutions which can ensure quality adherence and process compliance leading to customer satisfaction during such interactions. |
The technology proposes solutions towards audio, text, and image analytics for interaction between customers and customer-care executives (agents), happening at call centers. Our product SARATHI-AI efficiently extract information from customer agent interaction, and performs automatic quality analysis (AutoQA) pertaining to several quality based parameters. The product can assess agent's performance and also highlight critical and actionable insights which can help manage call centers effectively and poise business growth. Our product DISHA-AI is another product which acts like an agent assist, and performs real time quality analytics while providing real-time suggestions and action inputs to the agent, during a live conversation with the customer. Our product VAANI-AI is an AI based conversational engine which can act like a virtual agent and perform human-like conversation with customers, assess their complaints and requirements, and help with their queries while adhering to an underlying process SOP document. |
Deep tech is fundamental to the architecture of all the products developed at Vocab.AI. The products developed at Vocab.AI involve state-of-art deep architectures pertaining to audio , text, and image processing domains. We have utilized several pre-trained and untrained models, and have tuned it with real-time call center data from different domains, such as financial organizations, major online gaming platforms, public service and convenience, ecommerce groups, etc. The solutions form Vocab.AI are mostly hosted on cloud architecture and can be accessed using APIs. Our team is proficient in Deep Tech and pioneers in technologies for speech recognition, speech synthesis, natural language processing, large language models etc. |
Users |
Revenue, Pilots, Testimonials |
eCommerce , BFSI , Healthcare , etc |
2.6 Billion Dollars |
500 Million Dollars |
100 Million Dollars |
Pay-per-use model : where customers are charged per transaction (data processed), Deployment model : where customers gets the models deployed on their end, is charged one-time for deployment and additional maintenance and improvement charges as per their requirements |
ZOHO, SALESFORCE, FRESHWORKS, HUBSPOT, CONVIN, LEADSQUARED |
Vocab.AI is positioned as the premier AI solution for Indian businesses seeking to enhance their CRM and QA processes. By combining advanced text, speech, and vision analytics, Vocab.AI enables clients to gain a 360-degree view of customer interactions, ensuring accurate feedback, improved customer satisfaction, and personalized experiences. Unlike generic, one-size-fits-all solutions, Vocab.AI is tailored to the linguistic and cultural complexities of the Indian market, making it the ideal choice for businesses that demand precision, flexibility, and localized expertise. |
Target markets : Businesses such as ecommerce, public convenience, food and beverages etc. with significant public interaction are potential avenues relying on CRM processes. Assessment of quality assurance is one critical sector which determines the growth of these businesses. Initial market penetration : Device faster, accurate, multimodal, multilingual, and flexible solutions, which are customizable according to customer and process requirements. Channel/Partners : Partner up with Wyzmindz for leverage upon their rich clientele dealing with CRM solutions. The present solutions are based on manual efforts which are slow and cumbersome. Launch Team : Vocab.AI plans to develop AI based speech, text, and image analytics to automate quality auditing process and improve customer experience. We design low cost, highly accurate, customized solutions which can propel the growth of business and mitigate risks. |
To develop solutions to improve customer experience during their interaction with call centers, leading to business growth in different horizontals (sales, support, retention etc.,).
To deliver products efficient and cost effective products, which cater to the diversity in India and abroad. |
NA |
INR 1.4 Crores |
Yes |
The program is a pioneering initiative under the IIT Alumni Centre Bengaluru (IITACB) and supported by the Government of Karnataka, which provides significant credibility, networking opportunities, and potential access to public-sector resources and problem statements. ACSEL Technology Forum acts as a dynamic hub offering support through accelerator programs, talent development, and driving research, providing a multi-faceted approach to growing your venture or research project, with an ultimate goal to empower India's AI ecosystem in creating scalable and impactful solutions. Avinya will prove a perfect opportunity to establish our startup in the diverse market scenario, and foster collaboration with crucial stakeholders for partnership and scaling. |
From this program , we aim at achieving scalability for our products, and build strategic connections and partnership with the elite ecosystem comprising industry, academia, and the government. The program will also help us access skill and talent development programs from ACSEL, to gain expertise on cutting-edge real world problems. |
YES |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Pcc9V5xIm7HC-PitchDeck_VocabAI_2025_ASCEL.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Pcc9V5xIm7HC-PitchDeck_VocabAI_2025_ASCEL.pdf |
https://drive.google.com/file/d/1LJZ_fgwpDJyhRyhKFK80ORpp0Ig0PooT/view?usp=sharing |
At Vocab.AI we are aiming at building solutions which can impact businesses and improve their customer experience, in a way benefitting both parties to create a symbiotic relationship. We aim at developing our solutions in major Indian languages which can expand the outreach of business, and help the larger population to access their benefits. |
NO |
MS. SUMAN DASH |
checked |
| Sep 24, 2025 @ 10:46 PM |
Shivaram, Anuva Kapoor, Lakshminarayanan Subramanian |
shivaramvelayutham3223@gmail.com |
https://www.linkedin.com/in/shivaram-velayutham/ |
http://NA |
7548801657 |
Shivaram(CEO, CTO), M.Tech + B.Tech in Biological Engineering, Dr.Anuva Kapoor(CMO), Junior Resident in AIIMs Nagpur, Prof. Lakshminarayanan Subramanian(Co-Founder), Professor at NYU Courant |
We first came together through a shared commitment to improving healthcare in India. During the COVID-19 pandemic, Lakshminarayanan collaborated with Dr Bhanu Duggal (HoD of Cardiology, AIIMS Rishikesh) and Dr Mona Duggal (Director, ICMR Digital Health) on projects around hospital preparedness. This collaboration revealed a strong alignment in our vision to transform healthcare delivery in India.
Building on this, Lakshminarayanan co-advised my (Shivaram’s) M.Tech thesis with Prof. Balaraman Ravindran (HoD of Data Science and AI, IIT Madras), where I developed survival analysis–based risk models from EHR data for cardiac patients. This research now forms the scientific foundation of our startup.
To take this vision forward, Dr Anuva Kapoor (daughter of Dr Bhanu Duggal) and I came together as Directors and Co-Founders, combining clinical depth with technical expertise. Our team is deeply motivated to build impactful solutions that address critical gaps in healthcare outcomes for India. |
No |
Our biggest strength as a team lies in the rare combination of interdisciplinary expertise, clinical insight, entrepreneurial execution, and world-class mentorship. I (Shivaram) bring a dual B.Tech + M.Tech in Biological Engineering with a specialization in Computational Biology and AI, enabling us to bridge healthcare and advanced analytics. Dr. Anuva Kapoor, currently a Junior Resident at AIIMS Nagpur, grounds the team with hands-on clinical experience in India’s healthcare system. Our founder, Prof. Lakshminarayanan (NYU), is a serial entrepreneur with over 12 years of experience, having built three successful AI-driven startups (Flipped.AI, Vel.AI, Entrupy), and is also a globally recognized researcher advancing sustainable and socially impactful AI. Supporting us is an advisory board of leading experts: Dr. Bhanu Duggal, Head of Cardiology at AIIMS Rishikesh, with more than 30 years of experience in advanced cardiovascular care and research; Dr. Mona Duggal, Director at ICMR’s National Institute for Research in Digital Health and Science, with deep expertise in digital health innovation and public health policy; and Prof. Balaraman Ravindran, Head of Data Science and AI at IIT Madras, one of India’s foremost AI researchers in responsible and applied AI. Together, this blend of clinical leadership, AI innovation, entrepreneurial drive, and policy-level expertise uniquely positions our team to build, validate, and scale solutions that can drive transformative healthcare impact in India. |
Velai Health Analytics |
velaihealth.com |
|
Transforming cardiac care with continuous monitoring and AI-driven risk prediction. |
Most cardiac patients in India lack continuity of care even after major surgical interventions. While initial treatment provides temporary relief, it often creates a false sense of wellness, leading patients to disengage from follow-up care and monitoring. This gap significantly increases the risk of complications, re-admissions, and poor long-term outcomes. We are addressing this problem by enabling continuous monitoring and personalised follow-up care, ensuring sustained recovery and better cardiac health outcomes. |
Our solution is Second Opinion, a continuous cardiac care platform that bridges the critical gap between surgery and long-term patient outcomes. The platform integrates diverse health data, from genomic sequencing (hereditary risk) and EHRs (blood profiles, metrics) to diagnostic tests (CT, echo, X-ray, ECG), and applies AI-driven risk assessment to generate personalised, actionable care plans across six levels:
Hereditary risk & prevention – identifying risks from genetics and guiding what to avoid.
Behavioural interventions – support for smoking cessation, alcohol reduction, and diet moderation.
Lifestyle changes – nutrition, exercise, yoga, and weight management recommendations.
Diagnostics & follow-ups – timely reminders for tests and consultations.
Medication guidance – recommendations based on current health and risk profile.
Surgical alerts – early warnings when interventions may be necessary.
What makes us different is that our AI engine not only performs risk assessment but also works in tandem with certified healthcare practitioners, ensuring that every recommendation is medically validated before reaching patients.
By combining continuous monitoring, AI-driven insights, and practitioner oversight, Second Opinion empowers patients with clarity, ensures adherence to care, and reduces the risk of relapse or complications, ultimately improving long-term cardiac health outcomes. |
What makes our solution unique is our commitment to openness and interpretability. Unlike black-box proprietary systems, our platform is built on open research, ensuring that our methods are transparent and accessible to the wider scientific and medical communities. At the same time, we focus on interpretable AI, so that both clinicians and patients can clearly understand the “why” behind every recommendation. Our defensibility comes from: Comprehensive integration of genomic, EHR, diagnostic, behavioral, and lifestyle data into one continuous care platform. Clinician-in-the-loop validation, where every AI-driven recommendation is reviewed and approved by healthcare practitioners. Trust through transparency, where open, interpretable models foster credibility with hospitals, regulators, and patients, making adoption easier and replication harder. By combining continuous monitoring with interpretable, open AI validated by medical experts, we are creating a platform that is not only effective but also deeply trusted and defensible in the long run. |
Idea |
Pilots |
Our target customers span across the healthcare ecosystem: Patients – individuals with a history of cardiac conditions who need continuous monitoring and personalized care plans. Healthcare providers & institutes – hospitals and specialty clinics that can use our platform to ensure better long-term outcomes for their cardiac patients. Health insurance agencies – payers who benefit from reduced readmissions, complications, and overall cost of care by integrating our solution into patient management programs. By serving patients directly while partnering with hospitals and insurers, we create a sustainable, multi-stakeholder model that improves patient outcomes and reduces the overall economic burden of cardiac disease. |
$50B+ global market for cardiac digital health solutions. |
$1–2B Indian market opportunity, with ~10M cardiac patients who can afford or access continuous care solutions. |
$10–15M over the next 5 years by capturing 1–2% of the SAM through direct patient subscriptions and partnerships with insurers and hospitals. |
Our revenue model is primarily subscription-based: Patients – direct subscription plans that provide access to continuous monitoring, risk assessment, and personalized care guidance. Insurance agencies & healthcare providers – enterprise subscription/partnership models where our platform is integrated into their patient management programs to improve outcomes and reduce costs. While pricing tiers are still being finalized, the model is designed to scale from individual patients to institutional contracts, ensuring both affordability for users and sustainability for the business. |
Our main competitors span three categories: Cardiac-focused monitoring solutions – iRhythm (US, Zio patch) and AliveCor (US, portable ECG) are leading players globally, while Tricog Health (India/Singapore) is building AI-enabled cardiac diagnostics in the Indian market. Digital therapeutics for chronic disease – Omada Health (US) and Livongo/Teladoc (US) have established chronic-care platforms, while in India, Wellthy Therapeutics has pioneered digital therapeutics for conditions including cardiovascular disease. Healthcare platforms with cardiac extensions – In India, broad healthcare ecosystems like Apollo 24|7, Practo, and Tata 1mg are extending telehealth, diagnostics, and post-care, though they are not cardiac-specific. Our differentiation lies in being a research-driven, India-rooted platform focused specifically on continuous cardiac risk monitoring and post-intervention care, an area where global players are strong but localized, accessible solutions for Indian patients are still limited. |
We are planning to acquire customers primarily through partnerships with hospitals and healthcare institutes, where we roll out pilot programs that onboard patients directly as part of their follow-up care. |
Our GTM strategy follows a B2B2C pathway: Pilots with hospitals → demonstrate improved outcomes and patient adherence. Expand within institutions → scale across departments and hospital networks once value is proven. Insurance partnerships → integrate into wellness and chronic care management programs, driving adoption at scale. Direct-to-patient offerings → once trust is built, launch subscription plans directly for cardiac patients, supported by referrals from clinicians and insurers. |
Our long-term vision has three fronts:
To build a globally recognized medical AI research lab from India that not only advances cutting-edge research but also translates it into real-world clinical solutions.
To create a significant positive impact on the well-being of cardiac patients in India, and gradually expand our focus to other chronic diseases.
To scale our frameworks and solutions beyond India, making them accessible and beneficial to patients worldwide. |
We are currently incorporated in Hyderabad, India as a Private Limited Company under Companies Act along with an US entity registered as a LLC. |
We have not raised external funding yet. However, we’ve already generated ₹10 lakh in revenue through a commercial contract |
Yes |
We are applying to Avinya because it is uniquely positioned to support early-stage, research-driven startups like ours. While we are currently in the process of getting incubated, the 4-month structured programme, with access to curated datasets and deep industry engagement, will help us accelerate product development and validate our cardiac monitoring framework in real-world hospital settings. The mentorship and AI ecosystem make Avinya the ideal launchpad for us to translate our research into a scalable, responsible healthcare solution that can impact millions of patients. |
During the programme, our primary goal is to accelerate product development and secure early hospital rollouts across India to generate initial patient traffic and validate our solution on smaller scales. Alongside this, we aim to leverage Avinya’s mentorship, datasets, and ecosystem support to sharpen our clinical and technical approach, while also raising seed funding to fuel the next stage of our growth. |
We are open to both hybrid as well as virtual participation |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/fyg18mSYsqGq-VelAIHealth_Incubation_Pitch.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/fyg18mSYsqGq-VelAIHealth_Incubation_Pitch.pdf |
http://NA |
Yes, we are building a mission-driven and impact-focused startup. Our mission is to transform cardiac care in India by ensuring that patients receive continuous, personalised, and evidence-based follow-up after major interventions.
Cardiac disease is the leading cause of mortality in India, yet most patients lack continuity of care once they leave the hospital. This results in avoidable complications, readmissions, and loss of lives. We believe that technology should not only innovate but also bridge critical gaps in access, affordability, and adherence.
By combining open and interpretable AI research with clinician-validated care plans, we are creating a transparent, trusted platform that empowers patients, supports doctors, and reduces the overall burden of cardiac disease. Our impact lies not only in improving long-term patient outcomes but also in strengthening the healthcare system and reducing costs for families and insurers.
This is not just a business opportunity for us, it is a mission to reimagine how cardiac care is delivered and sustained in India and beyond. |
No |
Achira Labs |
checked |
| Sep 24, 2025 @ 9:42 PM |
Aishwarya Gujrathi; Vighnesh Bheed; |
aishwarya.gujrathi@weavestudio.ai |
https://www.linkedin.com/in/aishguj/ |
http://NA |
9989689574 |
Vighnesh: AI & Full-stack engineer with expertise in scalable systems, agentic AI, and SaaS product development. Aishwarya: AI researcher with experience in computer vision, robotics, and medical AI; two-time DAAD Scholar. |
We went to the same school and worked together on several hobby projects since college. Over time, that collaboration has built trust and familiarity. |
No |
Our biggest strength is the balance we bring as a team. Vighnesh is a great builder, relentlessly curious and exceptional at turning ideas into working systems, while Aishwarya combines strong technical skills with business acumen and the ability to connect and drive ideas forward.
Having worked together for years across multiple goals, we’ve built trust and a complementary dynamic that makes us an unbreakable team. |
Weave Studio |
https://weavestudio.ai/ |
|
The Vibe Anything Platform |
Building prod grade AI syetems is complex, time consuming and too slow for a market which is constantly evolving. Software systems are still passive, state-based tools. They lack a system to learn, adapt and act across own tools and those of connected systems. |
Weave Studio is a Managed AI Delivery Platform that delivers tailored solutions in days, not months. It seamlessly integrates with any tech stack or data source and offers modular building blocks to design powerful multi-agent workflows. With full-stack orchestration and infrastructure provided out of the box, it gives teams everything they need to launch and scale AI automation effortlessly. |
We’re not just another AI chatbot. We integrate deeply into the product core, enabling creation of powerful multi-modal agentic AI workflows. Our costs scale perfectly with revenue, ensuring zero expense for idle infrastructure, allowing us to offer usage based pricing.. |
MVP |
Pilots |
Ideal Customers are Product, SaaS, Tech companies. Our Beachhead is Video editing, digital art, and creator tools. |
Multi billion dollar market |
NA |
NA |
Businesses pre-purchase Weave Tokens and spend only on usage. |
Unframe AI, Adopt AI |
Founder driven sales, pilots |
Weave Studio offers a risk-free way to adopt AI with a usage-based model. Businesses pre-purchase Weave Tokens and only spend on actual usage, ensuring costs scale directly with revenue and eliminating idle infrastructure expenses. This makes AI integration seamless, predictable, and cost-efficient from day one. |
To pioneer a future where agentic AI is baked into the core of every software, unlocking new value creation. |
Incorporated |
NA |
Yes |
To take advantage of the network. |
Network with peer founders, investors and raise funds for scaling and growth. |
yes, suits us well |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/tbNxuZfiW81f-WeaveStudio.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/tbNxuZfiW81f-WeaveStudio.pdf |
https://www.youtube.com/watch?v=E8DADXPEOYA |
We don't like to categorize ourselves. Our mission is to bring impact by enabling a very easy adoption of Agentic AI core to allow users to do more. |
We are from Belagavi, north karnataka, where the startup boom hasn't quite hit yet. |
Vivek K Jha |
checked |
| Sep 24, 2025 @ 8:12 PM |
Kushal CI, Vaibhav Krishna |
kushalingole@IITBOMBAY.ORG |
https://www.linkedin.com/in/kushalcingole |
https://dawaai.co/ |
09769140510 |
Kushal C. Ingole – IIT Bombay & IIM Ranchi alumnus, with 15+ years of experience in business management and strategy across multiple industries, functions, and domains, including family offices, social sector, corporates, startups, government roles, and MNCs. Dr. Vaibhav Krishna – IIT Bombay alumnus, with Masters & Ph.D. from Yale University, and 10+ years of experience in AI/ML, building and deploying advanced technology solutions across research, enterprise applications, and startups |
We have met at IIT Bombay in 2005. We started our graduation journey together and worked on multiple projects together. |
No |
We bring complementary skills across technology and business management, with a combined work experience of 25+ years in the industry. Having known each other for a long time, we share a deep mutual understanding—both in terms of responsibilities and personal commitments—enabling us to collaborate efficiently and drive the startup forward |
Dawa AI |
https://dawaai.co/ |
|
AI-driven platform for pharmaceutical manufacturers to optimize production, forecasting, and distribution |
Today, pharmaceutical companies lack real-time visibility into data at the point of sale and patient level. This delay not only hinders swift and informed decision-making but also impacts patients' timely access to life-saving medications and medical equipment. Typically, data from the field takes 3 to 4 weeks to reach key decision-makers - leaving little room for strategic adjustments in production planning, raw material procurement, or workforce deployment. |
Dawa.ai is an advanced, AI-powered Business Intelligence (BI) SaaS platform designed to equip executives and decision-makers with dynamic, on-demand insights, actionable recommendations, and customizable dashboards. The platform seamlessly enables real-time data flow from POS systems and distribution channels, ensuring organizations have the visibility they need to act with speed and precision.
Our solution suite encompasses a comprehensive product encyclopedia, integration with sales and distribution networks, service-level automation, hybrid SaaS/Client-Server deployment models, POS-level connectivity, customer engagement tools, advanced supply chain integration, and production-level optimization plug-ins—all designed to maximize efficiency and impact. |
We did this by Developing a Saas Platform which connected the Pharma Companies with their marketing & Sales Network, & their products distribution Channels, right up to the POS & Patient level. The ability to provide End-to-End solution makes us unique and defensible as it requires deep understanding of the pharmaceutical value chain. |
Users |
Users |
Pharma manufacturers, Distributors, Hospitals |
$16 Bn |
$ 4.8 Bn |
$ 500 Mn |
Enterprise & SaaS |
No direct competitor but BI / ERP tools can be indirect competitors |
Industry Tie-ups, Partnerships |
Retail Chains, Tier 1 Manufacturers |
Our vision is to be recognized globally among the top three innovators transforming pharmaceutical decision-making through intelligent, real-time data solutions. |
Yes |
25 lacs |
Yes |
Mentorship, IP, Funding |
PMF |
Yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/5cHxeS2ShX79-Dawa.ai-Avinya-Sep-25.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/5cHxeS2ShX79-Dawa.ai-Avinya-Sep-25.pdf |
https://dawaai.co/productdemo.html |
YES, At Dawa.ai, our mission is to transform the way pharmaceutical companies make decisions - while ultimately improving access and outcomes for the patients they serve. We are committed to:
Empowering Pharma Companies
By delivering real-time, AI-powered insights at the point of sale and patient level, we enable faster, data-driven decisions that drive operational efficiency, market responsiveness, and competitive edge.
Bridging the Data Gap
We close the critical gap between the field and the boardroom by providing dynamic dashboards, intelligent reports, and predictive analytics that reduce decision latency from weeks to minutes.
Optimizing Healthcare Access
Through better planning and smarter forecasting, our solutions help ensure life-saving medicines and equipment reach the right places at the right time—improving healthcare delivery and patient outcomes.
Driving Innovation at Scale
Our platform combines the power of AI, cloud, and business intelligence to create scalable solutions that adapt to evolving market dynamics and regulatory environments.
Creating Long-Term Value
We aim to be a trusted partner in our clients’ growth journey—supporting strategic decisions, reducing inefficiencies, and helping build a more agile, resilient pharmaceutical ecosystem |
Yes |
Cdr Pradeep ji |
checked |
| Sep 24, 2025 @ 7:55 PM |
Shadab Khan S and Anjani Abhishek |
shadukhan@gmail.com |
https://www.linkedin.com/in/shadabkhan1926/?trk=opento_nprofile_details |
https://growise.org.in/ |
+91 7411027567 and +91 8867120901 |
Key Roles Founder & CEO / Co-Founder – Driving vision, strategy, fundraising, and scaling. Technology & IoT Architect – Designing sensor-based hardware, firmware, and cloud integration. Product Strategist – Defining roadmaps, user journeys, and scalable architectures. Operations Lead – Managing prototyping, supply chain, deployments, and customer support. Data & AI Lead – Applying analytics & ML for crop health, yield forecasting, and automation. Professional Value Proposition Smart Agriculture & IoT Expertise – Advanced knowledge of sensors, fertigation, precision irrigation, and control systems. Proven Product Development – MVPs to scalable solutions with focus on cost, reliability, and modularity. Technology Stack – Embedded hardware (ESP32), secure connectivity, OTA firmware, AI-driven dashboards. Market Understanding – B2B & B2C agri-tech applications with strong competitive positioning. Leadership & Partnerships – Built cross-functional teams, engaged suppliers, and executed pilot deployments. & Founder / Co-Founder / CEO Sets strategic vision, leads fundraising and partnership efforts, and navigates growth & scaling. Product / Business Leader Responsible for defining the product roadmap, understanding customer needs, and ensuring product-market fit. Oversees design, UX, and scaling. Operations & Execution Head Manages execution: supply chain, manufacturing / service delivery, deployment, logistics, support. Ensures reliability, cost efficiency & quality. Go-to-Market / Growth Lead Plans and executes sales, marketing, customer acquisition strategies. Builds channels & partnerships for growth and scaling revenue. Data / Insight / Strategy Lead Uses analytics and market & user data to drive decisions, optimize metrics (CAC, LTV, margins), and forecast trends. Identifies KPIs for sustained growth. Background & Value Proposition Domain Expertise & Credibility Deep knowledge in the relevant field(s): technology, market, domain, operations. Brings track record (projects, prior startup / corporate success, domain awards / recognitions). Proven Product / Solution Development Has developed MVPs / beta products / launched offerings already. Capable to iterate quickly, adapt to feedback, maintain quality and scale. Strong Technology / Infrastructure Solid technical foundation: product architecture, platform, IP, tech stack, reliability, data systems. Competitive advantage via tech. Scalable Business Model Clear revenue streams, unit economics, path to margin improvement. Model should allow growth in customers, geographies, or segments with reasonable capital. Traction & Metrics Evidence of demand: users / pilots / customers, retention, growth, partnerships, maybe early revenue. Financial planning: burn rate, cash flow, forecasted milestones. Team & Partnerships Strong founding / leadership team. Key partners / suppliers / distribution channels already defined or in progress. Ability to execute. |
We first met twelve years ago during our B.E. program in college, and over the past year, we have been collaborating professionally, building on our shared vision and complementary strengths. |
Yes |
Our team’s strength lies in the seamless synergy between product development and market execution. While the software developer ensures robust, scalable, and innovative technology, the sales execution professional translates these capabilities into customer value — driving adoption, building relationships, and closing deals. This balance of technical depth and market expertise enables us to rapidly deliver solutions that not only work flawlessly but also achieve strong commercial traction. |
GroWise |
https://growise.org.in/ |
|
GroWise is a future-generation AI farming ecosystem |
GroWise transforms farming challenges into opportunities with precision irrigation, smart automation, and real-time monitoring. Its mobile-first platform empowers urban professionals to grow more with less — saving water, nutrients, and labor while boosting crop health. |
GroWise's roadmap includes advanced AI capabilities transforming agricultural decision-making:
Predictive Analytics:
• Nutrient deficiency prediction through pattern recognition
• Early plant disease detection combining sensor data with computer vision
• Climate adaptation alerts for weather-related risks
Intelligent Automation:
• Smart harvest timing optimization based on growth patterns
• Seasonal crop recommendations using historical data and market analysis
• Autonomous pest management through integrated monitoring systems
These AI capabilities will position GroWise as a comprehensive agricultural intelligence platform, enabling farmers to make data-driven decisions with unprecedented accuracy. |
Our solution is unique with AI-driven automation, app-based alerts, and intelligent nutrient monitoring — offering low-maintenance farming alternatives that maximize yield and efficiency. |
MVP |
Pilots |
Urban households, smart gardeners, greenhouse owners, hydroponic farms, and agri-businesses seeking precision farming solutions. |
Global Smart Agriculture Market – ~$25B+ (2024 est.). |
Urban farming & greenhouse automation in India & APAC – ~$3B. |
Initial focus on urban households, hydroponics, and small greenhouses – ~$150M. |
Hybrid model: Hardware sales (sensors, IoT kits) + recurring SaaS subscription (AI insights, cloud dashboard, mobile app) + after-sales services. |
AeroFarms, CropX, Gardyn, Fasal, AgNext (regional/global smart farming & IoT agri-tech startups). |
Through direct online sales, partnerships with agri-tech distributors, B2B tie-ups with greenhouse & hydroponic farms, urban gardening communities, and targeted digital campaigns. |
Pilot deployments in urban and semi-urban households, partnerships with agri-tech retailers, and community-driven adoption through app-based engagement and subscription models. |
To be the leading AI-powered smart farming ecosystem, making sustainable food production accessible to every household and commercial farm, while reducing resource wastage and maximizing yields globally. |
Applied for Registration |
NA |
Yes |
To gain strategic mentorship, access to investor networks, and acceleration support that will help us refine our business model, validate at scale, and secure early-stage funding. |
Build strong market validation through pilots, sharpen go-to-market strategy, and prepare for scalable growth with investor-ready financials and partnerships. |
Yes, open to both hybrid and virtual participation for flexibility and maximum engagement. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/PgeLMhsXdTss-AVINYA_GroWise-Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/PgeLMhsXdTss-AVINYA_GroWise-Pitch-Deck.pdf |
http://NA |
Yes. GroWise is a mission-driven AgriTech startup focused on solving water scarcity, food insecurity, and urban farming challenges. By combining IoT, automation, and AI (with future Agentic AI capabilities), we aim to make farming more sustainable, resource-efficient, and accessible. Our solution empowers small farmers, greenhouse operators, and urban households to grow more with less, directly contributing to climate resilience, food security, and sustainable urban living. |
No |
NA |
checked |
| Sep 24, 2025 @ 7:18 PM |
Manish Shetty, Smaran B M, Rajesh Sambaragi |
yunoteam2025@gmail.com |
https://www.linkedin.com/in/manish-shetty-519b852b0?utm_source=share&utm_campaign=share_via&utm_content=profile&utm_medium=ios_app |
https://projectyuno.web.app/ |
87923 44909/ 8073050848 |
Manish - Founder/ Developer, CSE Student at KLEIT, Rajesh Sambaragi - Co Founder/Developer, CSE Student at KLEIT, Smaran B M - Co Founder/ Marketing & Sales, Information Science Engg Student at DSATM. |
So the story goes back to 2024 August where I met Smaran, it was through a mutual friend of mine. As I had just stepped on my startup journey I wanted someone to help gain visibility on social media, so my friend recommended me to talk to Smaran. We met in a cafe for a cup of tea and brainstormed the possibilities of me having him as an Editor. He instantly said YES, no money, not so sure of the benefits, just a curious mind who loved my idea of helping people who needed therapy. Although, I had some rough moments at that time, I could'nt really hire him. But in January we met again, with PURPOSE. This time we both had a similar idea in our minds and amalgamated that idea into what we're building right now. Since January 2025, we have been working together. The best part is that, we know each others weaknesses and we don't use them as weapons to get things done. Now Rajesh, my classmate, although we weren't working together since the beginning of our college, he was a part of another startup and recently went through a rollercoaster ride with his board which made him leave. He's a very sharp guy, who fancies tech, I mean the guy is part time Youtuber who loves sharing his opinion on new tech. Smaran and Rajesh are very good friends from their school days, so Smaran just gave me the idea of bringing in Rajesh as Co Founder as he was worthy of it. And in July we finally met officially and we three have been working together ever since. |
No |
I think what sets us apart is the maturity to acknowledge each other’s weaknesses without weaponizing them, instead turning them into guardrails for collective growth. This foundation of trust, authenticity, and aligned vision makes our team inherently resilient and primed to scale. Each of us came together not just through chance, but through shared values, resilience, and genuine curiosity rather than transactional motives. And we know for a fact that things are going to workout, slowly but surely. |
YUNO |
https://projectyuno.web.app/ |
|
YUNO is an AI-powered digital companion that unifies productivity, emotional wellness, and proactive support in one privacy-first platform. |
No simple way exists to improve both personal productivity and overall well-being at an individual level.YUNO is solving the gap between emotional support and practical productivity by creating an accessible, personalized AI that improves mental wellness and daily output at the same time. |
Personalized AI is either too expensive, too narrow, or too impersonal. YUNO solves this by making personalized AI both human and accessible. How?
YUNO is designed to support individuals who are overwhelmed by the demands of modern work and life. From managing emails, reminders, and scheduling meetings to capturing notes and action items, YUNO acts as a proactive assistant that ensures nothing slips through the cracks. Beyond productivity, YUNO safeguards emotional well-being by encouraging healthy breaks, offering empathetic companionship, and being available whenever users feel low or need someone to talk to. Privacy remains a non-negotiable pillar of our platform. At the core, YUNO develops a unique AI persona for each user, an evolving character that learns, adapts, and grows alongside them, mirroring their behaviors and communication style. This makes YUNO not just an assistant, but a truly personal companion. |
“Our moat is built on three layers: emotional lock-in from deeply personalized AI, defensible data generated uniquely by each user’s journey, and a first-mover advantage in bridging productivity with emotional wellness. Combined with a privacy-first stance, YUNO creates a trust and data moat that Big Tech can’t replicate. |
MVP |
Users |
ech-savvy professionals seeking a unified solution for productivity and emotional wellness. Mental health advocates and wellness enthusiasts who value daily support and proactive check-ins. Productivity-focused individuals who struggle with task overload, missed commitments, or fragmented app experiences. Early adopters in the AI, health, and self-improvement communities and those eager to try new personalized digital companions. Enterprise teams and organizations looking to improve employee well-being and operational effectiveness through customized AI solutions. |
$56.3B+ |
$9B+ |
$36M+ |
Free Tier, Premium subscription, Enterprise Solutions for B2B, Data Driven Insights |
Replika, Woebot, Martin Voice AI, Calmi, Zoom AI companion, Earkick |
Building awareness through educational content, influencer collaborations, and active user communities emphasizing emotional wellness, productivity, and AI empowerment. Initial outreach focuses on tech-savvy professionals, mental health advocates, and productivity enthusiasts who value personalized AI and can evangelize YUNO. |
Building a strong online footprint with educational content, influencer campaigns, and a user-centric community around wellness and productivity. Attracting users via a compelling free tier, then converting them to premium for advanced personalization, coaching, and exclusive features. |
YUNO’s long-term vision is to pioneer the future of AI companionship by uniting productivity, wellness, and emotional connection in one seamless experience. The goal is to become the single, trusted AI assistant that fundamentally transforms how people live, work, and feel every day and creating lasting value not just commercially, but socially, by leading innovation at the intersection of human-centric design and advanced technology. |
MSME |
NA |
Yes |
The reason we are applying to Avinya is that we believe that there is always room for improvement, and as Avinya is a founder first platform that nurtures ideas and accelerate the creation of transformative technologies for the future. And more importantly this is an accelerator program that extends focus on AI Technology and leveraging AI to build impactful solutions in the real world scenarios. Also we believe that Avinya provides visionary entrepreneurs with the structured mentorship, global networks, and tailored resources essential for scaling faster and making a lasting impact. |
Our main goal is to finish our MVP, launch the mobile App and gain atleast 10K Users with 2.5K paying users. |
Yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/q8zY5nZZWBQ4-YUNO-PRE-SEED.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/q8zY5nZZWBQ4-YUNO-PRE-SEED.pdf |
https://drive.google.com/file/d/1WUTfN674GodnQSwKodXHl_VifJse6JBg/view?usp=sharing |
YUNO is both an impact-focused and mission-driven startup. It aims to create meaningful social impact by improving individual well-being, mental health, and productivity through AI while pursuing a clear mission to transform how people live, work, and feel every day. The startup blends commercial goals with a strong purpose centered on emotional wellness and human-centric AI design. |
No |
Suman Dash |
checked |
| Sep 24, 2025 @ 5:11 PM |
Vivek Kumar Gupta |
eagleaipvtltd@gmail.com |
https://www.linkedin.com/in/vivek-gupta-3ab61a120/ |
https://zivankh.io/ |
08144709830 |
CEO – Mr. Vivek Kumar Gupta Role: • Leads the overall vision, strategy, and execution of ZIVANKH. • Responsible for partnerships, fundraising, and scaling the company across private and government healthcare sectors. • Represents the company externally with investors, policy-makers, and medical institutions. Background: • 4 years of corporate experience as an R&D Engineer and later as a Product Development Engineer. • Expertise in building real-world products, managing cross-functional teams, and aligning innovation with market needs. • Founded ZIVANKH to solve the pressing challenges of OPD efficiency and doctor burnout through AI-driven clinical automation. Product is tested out in AIIMS Bhubaneswar Paediatric department currently ________________________________________ CTO – Kumar Saurabh Role: • Owns the technology vision and development roadmap of ZIVANKH. • Leads AI model development (In-House built AI models) (speech-to-text, NLP, medical report analysis) and hardware-software integration. • Ensures product scalability, security, and compliance with healthcare standards Background: B-Tech, M-Tech in Computer Science from IIT Bhubaneswar • Strong academic/industry experience in AI, ML, and system design. • Prior work in healthcare AI, speech recognition, or clinical data systems. • Experienced in leading engineering teams to build scalable cloud-based AI products. |
Our company is Incubated in IIT Bhubaneswar, so we had access to the Research team and professors of IIT Bhubaneswar, who connected us to Dr. Bhagirathi Dwibedi who is professor in AIIMS Bhubaneswar where we are currently pilot testing our product. So Dr. D P Dogra sir, who is HOD of Computer Science in IIT Bhubaneswar, Connected me to Kumar Saurabh, and from last 18 Months we have been working together on this project ZIVANKH |
Yes |
Our biggest strength as a team lies in the complementary blend of clinical insight, product innovation, and technological expertise. The CEO brings experience in R&D and product development, ensuring that ideas are converted into practical, scalable solutions, while the technology leadership focuses on AI, system design, and healthcare integration.
Together, we combine:
• Deep understanding of healthcare pain points with our Collaboration with AIIMS Bhubaneswar (doctor burnout, OPD inefficiency, manual workflows)
• Strong technical capability in AI/ML, speech recognition, and Large system architecture. We are Incubated in IIT Bhubaneswar and 4 of our team members are from IIT’s
• Execution-driven mindset with real-world deployment experience in hospitals and PHCs
This balance of domain knowledge + technical innovation + execution focus enables us to not just imagine ZIVANKH, but to build and scale it effectively. |
EagleAI Private Limited |
https://zivankh.io/ |
|
ZIVANKH is building the future of OPD care with AI-powered clinical co-pilot device that empower doctors to deliver faster, smarter, and more compassionate healthcare |
We are solving the systemic problem of inefficiency and errors in outpatient care, where doctors are overburdened with repetitive tasks such as syndromic history taking and report reading and patients face long waits (specially in Govt. owned hospitals like AIIMS or state owned etc.), fragmented records, and delayed diagnoses (24 Lakhs Indians die annually because of treatable medical condition that happens because of delayed diagnosis). We are addressing this gaps in OPD workflows |
ZIVANKH is an AI-powered clinical co-pilot device that streamlines OPD consultations by automating syndromic history-taking, interpreting medical reports, and generating digital prescriptions through natural voice interaction. With vernacular language support, patients can describe their symptoms in their own language, making the process more inclusive and accurate. What once took 15–20 minutes in a consultation can now be completed in less than half the time, enabling doctors to double their patient throughput without adding extra doctor or infrastructure. By simply installing ZIVANKH, hospitals and clinics can unlock higher efficiency, 2X Revenue, and improved patient satisfaction with digital Log of each OPD consultation—while allowing doctors to focus on what matters most quality care.” – This AI device ZIVANKH holds ability to even work without Internet, which unlocks its potential to reach to the remotest possible places on Earth. |
What makes ZIVANKH unique is that we are not dependent on generic third-party AI engines. We have developed our own in-house AI models, specifically trained on clinical data and medical literatures (Books of MBBS and MD) and tuned for OPD workflows in the Indian context. This allows us to support vernacular languages, handle handwritten reports, and generate clinically relevant insights with higher accuracy than off-the-shelf solutions. Owning the core technology not only gives us defensibility and control over performance, but also ensures scalability, data security, and the ability to continuously innovate—making ZIVANKH a truly differentiated and future-ready healthcare solution |
Users |
Users, Pilots, Testimonials |
Primary Target Market (Early Adopters) 1. Private OPD Doctors 2. High-volume practitioners (30+ patients/day) in metro and Tier-2 cities 3. Specialists in General Medicine, Paediatrics, etc. Private Clinics & Multi-Specialty Hospitals 1. Clinics looking to improve OPD throughput and efficiency 2. Hospitals aiming to digitize workflows without heavy IT investment 2. Secondary Target Market (Scaling Stage) 1. Government Hospitals & PHCs (Primary Health Centers) 2. Overcrowded OPDs with doctor shortages 3. Rural healthcare where vernacular language support is critical 1. Diagnostic Centers & Labs 1. Need for report interpretation, structured summaries, and AI-driven insights 3. Extended Market (Future Opportunities) • Telemedicine Platforms & EMR Providers Integration of ZIVANKH APIs for syndromic history and report analysis • Opening of our own AI-SMART CLINICS in rural areas. 1. Large-scale rural deployments for affordable and accessible care • Global Expansion 1. Emerging markets in Africa & Southeast Asia with similar OPD challenges |
INR 4.32 Lakh Crore |
INR 1.82 Lakh Crore |
INR 8,000 Crore |
(Hardware + SaaS + API) |
Practo Ray, mfine, DocPulse / Medpiper / Eka Care but none of them are doing AI powered automated syndromic history and report analyzing part, that's our differentiator. So while competitors focus on EMR, dictation, or telemedicine, ZIVANKH uniquely positions itself as a doctor-first clinical co-pilot device for OPD efficiency. |
1. Direct Engagement with private OPD performing doctors 2. Hospital and Institutional partnership. 3. Thought Leadership & Medical Community |
Phase 1 (0–12 months): Doctors/Small Private Hospitals → doctors refer doctors Phase 2 (12–24 months): Scaling up via hospital chains & B2B platforms Phase 3 (24+ months): Rural expansion + government partnerships Phase 4 (36+ months): Global emerging market rollout |
ZIVANKH – Long-Term Vision : 24 Lakhs Indian Die every year of treatable medical conditions majority (80%) of them are because of delayed diagnosis.
Our long-term vision is to make accessible, accurate, affordable and intelligent healthcare available to every human being, regardless of geography or socio-economic background.
We begin by transforming Tier-3 city hospitals, where patient overload is highest, by enhancing their daily consultation capacity through ZIVANKH devices. This ensures that even with limited doctors and infrastructure, more patients are seen, faster and with higher precision.
Next, leveraging our vernacular language support, we will expand into AI-SMART Clinics across Tier-2, Tier-3, and small towns, so patients no longer need to travel long distances to metro cities for quality care. These clinics will bring advanced AI-assisted healthcare closer to where people live.
In the final phase, our vision is to create a HEALTHMETER device for every home—just like electricity or water meters. Through this, families can instantly connect with their nearest AI-SMART Clinic during a health crisis, speak to the device in their local language, receive immediate guidance, and, if needed, have emergency care dispatched to their doorstep.
By building this continuum of AI-driven healthcare—from hospitals to clinics to homes—ZIVANKH will not only reduce the burden on doctors and hospitals but also democratize healthcare, making it seamless, proactive, and truly universal.
This positions ZIVANKH not just as a device company, but as the foundation of a healthcare ecosystem that grows step by step toward a bold, transformative vision. |
EagleAI Private Limited was Incorporated on 25 September 2023 |
1.05 Crore |
Yes |
To get right business mentorship, Industry connect and funding |
We have developed the product "ZIVANKH" with the help of IIT and AIIMS Bhubaneswar Doctors and professors. Now we need expertise on building business and our manufacturing setup for which we are looking for funds as well. Right industry connects. |
Yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/CNBbQJ189IHC-Eagle-AI-PPT.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/CNBbQJ189IHC-Eagle-AI-PPT.pptx |
https://youtu.be/xdSBbxGcgrQ |
Yes, ZIVANKH is a mission-driven, impact-focused startup.
Our mission is to empower doctors and democratize healthcare access by using Artificial Intelligence as a true clinical co-pilot. We are tackling one of India’s most pressing healthcare challenges—overburdened OPDs, limited doctor availability, and delayed diagnosis—by automating routine tasks so doctors can focus on care.
The impact we aim to create goes beyond efficiency:
Doctors: Reduced burnout, higher throughput, and improved decision-making.
Patients: Faster consultations, accurate digital records, and care in their own language.
Healthcare System: Increased capacity without additional infrastructure or manpower.
Society: Healthcare that is more accessible, affordable, and equitable, especially in Tier-2/3 cities and rural areas.
Long term, with our AI-SMART Clinics and Health meter devices, we envision a healthcare ecosystem where every family, even in the remotest towns, can access timely medical guidance at home—bridging the gap between urban and rural healthcare.
ZIVANKH is not just building a product, we are building a movement—to make healthcare more human, by letting AI handle the repetitive and doctors handle the healing. |
NA |
Dr. Soobhankar Pati |
checked |
| Sep 23, 2025 @ 4:00 PM |
Shyam Sundar G |
official.nexrtech@gmail.com |
https://www.linkedin.com/in/shyam-sundar-9881671b/ |
https://nexrtechnology.vercel.app/ |
9739277227 |
Shyam Sundar - I’m the CEO & Co-Founder of NEXR Technology, where I lead vision and strategy to build innovative solutions at the intersection of mental wellness and immersive technology. With my HR background I bring insights into employee well-being and organizational challenges. My focus is on growth, partnerships, and positioning NEXR as a transformative force in corporate and clinical wellness. Vinay Kumar is COO & Co-Founder of NEXR technology, brings a unique blend of engineering expertise and entrepreneurial experience from a pharma company. As COO, he drives operations, execution, and delivery, ensuring seamless integration of technology and wellness solutions. he focuses on efficiency, process excellence, and sustainable growth for NEXR. |
We have been friends for 9 years, sharing a strong bond and complementary skills. Over the last 2 years, we have explored and developed the idea behind NEXR technology together, combining our expertise in HR, technology, and entrepreneurship to bring this vision to life. |
No |
Our biggest strength as a team is the complementary nature of our expertise in HR, technology, and entrepreneurship. I drive vision, strategy, and business growth, while my cofounder ensures operational efficiency, process excellence, and seamless execution. This combination allows us to innovate effectively and scale NEXR with focus and precision. |
NEXR technology Pvt Ltd |
https://nexrtechnology.vercel.app/ |
|
A mental health tech startup building an AI + VR platform that empowers clinicians and engages patients through immersive, data-driven therapy. |
Therapists at clinics and hospitals practicing traditional therapy Face loss of customers and revenue because the patients are not achieving actual improvement and perceiving meaningful progress from sessions. Struggling with excessive effort in reconstructing patient history and tracking progress due to lack of effective tools |
We offer a clinician-led VR and AI therapy platform that assists therapists with patient history, diagnosis, session planning, and documentation while engaging patients through immersive VR experiences and personalized progress tracking. |
Unlike traditional therapy, our AI-driven VR therapy platform, creating a seamless loop between care delivery and client experience which ensures both clinical efficiency and sustained patient motivation, |
MVP |
Pilots |
Mental health hospitals, clinics and corporate companies |
₹468 Cr |
₹140.4 Cr |
₹7.02 Cr |
Hospital/ Clinic - License subscription, Corporate wellness tie-ups - Institutional Subscription |
Avika mind health & Wundrsight |
Direct Partnerships with Hospitals & Clinics, Insurance & Corporate Wellness Tie-ups |
1, Hospital partnerships and pilot programs, 2, Direct outreach to hospitals, clinics, and therapist networks |
To become the global leader in AI + VR powered mental health care software to providing for big hospitals, clinics and organizations. |
Completed |
NA |
Yes |
We are applying to the Avinya program to scale our AI + VR mental health platform beyond pilot testing. The program’s mentorship, ecosystem access, and investor network will help us refine our AI layer, validate at scale, and accelerate commercialization into a globally competitive solution. Avinya’s ecosystem of mentors, industry connections, and investor networks offers the right platform to: Validate at scale through clinical, corporate, and wellness partners. Access technical and business mentorship to strengthen our AI innovation and regulatory readiness. Prepare for fundraising by refining our go-to-market and scaling strategies. |
Personalized guidance from domain experts, Support to access industry network related to our industry. |
Yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/4FFMknERb6MT-NEXR-Pitch-2025.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/4FFMknERb6MT-NEXR-Pitch-2025.pdf |
https://drive.google.com/file/d/1KmZOI17mD1jR9ypVsbQP1HGhG0Q1f6g9/view?usp=drivesdk |
we are a mission-driven. Our mission is to make mental health care accessible, engaging, and stigma-free for all. By combining AI and VR, we empower therapists with smarter tools and provide patients with immersive, personalized therapy—improving outcomes, reducing dropouts, and addressing the growing mental health crisis at scale. |
No |
IIIT Dharwad - IDRP |
checked |
| Sep 23, 2025 @ 1:09 AM |
Sampath Kumar HK |
sampathkumar@bodyclone.in |
https://www.linkedin.com/in/sampath-kumar-hk-439b8617/ |
http://www.bodyclone.in |
9380785319 |
o Sampath Kumar HK – Co-Founder & Tech Head | 20+ years in MedTech XR/AI, ex-Zimmer, J&J, DJO Surgical | DeepTech startup leader recognized by Meta & AIIMS o Dr. Chetan MN – Orthopedic Surgeon | JJM Medical College & MRMC under RGUHS o Chandan MN – Software Developer | Bengaluru |
Team connected through professional and academic collaborations in healthcare & XR/AI over the past 3–5 years, aligning around a shared vision of Human Digital Twins. |
Yes |
Unique blend of deep technical expertise (AI/XR/IoT), medical domain authority (orthopedic surgeon), and software development — enabling clinically validated, AI-powered healthcare solutions. |
Bodyclone |
www.bodyclone.in |
|
AI-first Human Digital Twin platform for predictive, personalized healthcare, visualized in web, mobile and XR. |
Healthcare is reactive and fragmented. Patients lack continuous monitoring, predictive diagnosis, and ownership of their health data. Chronic conditions worsen due to delayed interventions. |
We build Human Digital Twins — real-time replicas of individual health using XR, AI, IoT, and AWS HealthLake. Our platform empowers predictive diagnosis, remote monitoring, surgical recovery, and personalized wellness. |
Unlike fitness trackers or organ-specific models, Bodyclone creates a full-person digital twin with integrated XR visualization and AI predictions. Backed by surgeons and IIT REP, our defensible edge is medical validation + deep-tech execution. |
MVP |
Users |
Hospitals, insurers, pharma, Gen Z/Alpha wellness consumers, fitness & sports tech, academic/research institutions |
$500B+ global digital health. |
$60B+ personalized healthcare (India & SE Asia) |
~$500M early adopters. |
B2C subscriptions (app, scans, reports) + B2B SaaS for hospitals, pharma, insurers. |
Apple Health, Fitbit (fitness only), Siemens/Dassault (organ twins). Bodyclone is unique in full-person health twins. |
Hospital pilots, surgeon-backed clinical validation, Gen Z engagement via digital-first health apps, partnerships with insurers & corporates. |
o Phase 1: Hospital pilots + AI dashboards o Phase 2: B2C rollout (app + wearables) o Phase 3: Scale via insurers, pharma APIs |
By 2030, create a health graph of 1B+ digital twins, shifting healthcare from reactive to predictive, democratizing AI in health. |
Incorporated in India (Private Limited) |
INR 3L from SISFS. IIT BBS REP through |
Yes |
To refine AI pipelines, strengthen regulatory readiness, and scale clinical validation through Avinya’s AI ecosystem. |
o Develop AI models for predictive diagnostics o Build compliance-ready health data pipelines o Prepare for fundraising as an AI-first healthcare company o Connect with hospital and research partners in Avinya’s network |
Both. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/hsQBJm7U5oB9-Avinya-Pitch-Deck-Presentation-compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/hsQBJm7U5oB9-Avinya-Pitch-Deck-Presentation-compressed.pdf |
https://www.youtube.com/watch?v=OcphJFtpSa8 |
Yes. Our mission is to democratize personalized, preventive healthcare by giving individuals ownership of their health data and digital twin. We aim to transform healthcare from reactive to proactive, predict illnesses before they strike, and empower 1B+ lives by 2030. Our impact directly aligns with UN SDG 3: Good Health & Well-being. |
NA |
Ashok Kamat |
checked |
| Sep 21, 2025 @ 11:03 PM |
Yash Tiwari, Arnab Samanta |
yashtiwari.media@gmail.com |
https://www.linkedin.com/in/yash-tiwari-18a084170/ |
https://www.mockverse.tech/ |
7224919700 |
Yash Tiwari - Involved with the engineering and business techincally creator onboarding, reach, UX, UI and backend frontend engineering || Arnab Samanta - Predominantly looking in the AI and Engineering |
Yash and Arnab met during their college days in IIT BHU where they have collaborated on the machine learning projects since they both shared a common passion for machine learning and product building. since Yash is the senior of Arnab he guided him through Internship and placements.
Currently both founders are working part time on it with a phenomenal speed and they are managing their full time jobs also along with it so that currently Yash is working as Sr. Data Scientist in Nielsen and Arnab is working as a Data scientist in the Citi Bank. |
No |
our (yash + Arnab) biggest strength is trust and resillience, this is not the first that we are building together we built two more produts prior to mockverse which failed we got hit when a close person working on our last project left us in the middle we didn't go broke we trusted each other and we stood back against all the odds and we built this platform which has been used by more than 100 people as of now and we are still approaching the goal for making this product a default choice for interview preparation with 100% passion and grit. and we are very practical cause if we want to work on this to make it a really great product we need to remove unnecessary pressure from us and that's why we are also managing out full time jobs. it's not easy but this is our mission and we both will make this thing possible we'll definitely gonna make mockverse a default interview preparation kit. |
MockVerse |
https://www.mockverse.tech/ |
|
AI interivew Platform - real world interviews conducted by reliable AI crafted by Industry professional |
Many People specially freshers who don't have interview experience they always mess their first few interviews which sometimes cost them a lot and doing mock with real person is simply not scalable at all, but with the rise of the AI people can use product like mockverse to practise the exact same interivew virtually to get comfortable with the interview procedures |
our platform mockverse has interviews which will be conducted by a reliable agent crafted by industry professional to mimick their style of conducting interivews and judging the candidate. So people can use this platform to prepare of their next interview by selecting the company or the creator and can experience exactly same interview session just as the creator is taking the interivew. so they can make mistakes, they can try out things, they can review feedback and improve on it. |
We are making a platform where different industry professional can create interviews so we are making a tool which industry professionals can use to create interviews which will be practised by the candidates we are trying to build an ecosystem just like leetcode did for coding. |
Users |
Users, Testimonials |
For 2025 we are targetting college students who are sitting for placements and for 2026 we are targeting both industry professionals with 1-3 years of experience as well as college students |
It's around $1.0B — $1.4B (2024 baseline) |
assume 40%–60% of TAM is addressable by a scalable AI platform → ≈ $0.4B — $0.84B |
a realistic 0.2%–1.5% of SAM → ≈ $0.8M — $12.6M which will depend on our distribution, partnership with the creators, colleges and corporates |
candidate will come they will pay for a pack of 10 interivews cause mockverse has mini interviews so a candidate needs at least 4-5 interviews to practise a complete interview session and we'll be giving 50% share to the creator as a motivator signal towards promoting and creating interviews |
I would say main competitior will be google only along the way but they are lacking onboarding real creators so as of now competitors are OfferEasy.ai, knockri, PretAi, AiMock, grapevine, etc |
we are going to organize seminars in the college to guide students how to prepare for roles and how to ace the interview and then we will be advising them to use this product as supplementary material specially in India then after seeing the response here we will be running ads in the south east asian countries for this product |
as of now our GTM strategy is reachout to colleges in India since this time placements are going in so we'll try specially in IITs we can push it to every group by asking their TPCs after december palcements in IIT will be over and we can some testimonials which we'll use in tier3 colleges to sell our product, we are giving 3 free interviews and then 5 interviews in 100 and every interview is of 10-12 minutes so a person will spend at least 100 rupees for practising one complete interview |
our long term vision is to make this a complete interview preparation platform so when it will come to getting job mockverse it is we'll callibarate all the resources in a single platform creators will lead this platform by creating complete job preparation place (for eg. Acing Software Engineering: complete company, industry understanding to how to crack interview).
and not only this we also want to make a platform where recrutiers can come and create their own interviews buy an pre created interview from a creator or can pay a creator to create an interview exclusively for them so that they can use this platform for screening candidates which is currently very big problem in the industry. plus we'll be solving a problem for sourcing candidates by providing hot leads to the recruiters. |
NA |
NA |
Yes |
we are very passionate to solve this problem and mostly we want to be a contributor in Viksit Bharat movement so If we'll get this incubation we can devote our 100% time and we are sure we'll bring this thing into it's finest form within 6 months |
we want to put this product to it's finest form and we are looking to get customers so that we can set our foots for a sustainable growth and we can contribute in our country's development by building from India for the world |
yes we are open to any kind of participation |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Am7ympny0tbZ-PitchDECK.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Am7ympny0tbZ-PitchDECK.pdf |
https://youtu.be/yTKC1_Q-Fm8?si=5oqlMcvddtCc_avU |
We are building an impact-focused startup which will uplift the career aspirations of students and professional it will give them power to improve their interview skills as well as their communication and it will give professionals monetary benefits and recognition by using their interiview skills and experience |
NA |
IIT BHU CDEM group |
checked |
| Sep 21, 2025 @ 5:32 PM |
Dhiraj Daga |
dhiraj.daga@indraastra.in |
https://www.linkedin.com/in/dddhiraj/?originalSubdomain=in |
http://NA |
8090672982 |
CTO |
NA |
Yes |
Our biggest strength as a team is that we combine deep technical expertise with a shared mission-driven mindset. Every member brings complementary skills—ranging from cutting-edge AI and space-tech research to systems thinking and execution—so we can solve complex problems holistically. We move fast, but with rigor, and our collaborative culture ensures that ideas are challenged, refined, and turned into scalable solutions. That blend of technical depth, adaptability, and unified purpose is what makes us strong together. |
IndraAstra |
NA |
|
Building World's Fastest and Efficient AI |
Current Inefficient AI Models |
We provide AI as a service |
Deep expertise |
Product-Market Fit |
Users, Revenue |
Large Corporations(GCC) |
50 billion USD |
8 billion |
500 MIllion |
Service based |
Element AI |
Outreach and Partnership. |
Direct outreach, partnerships. |
To be the leading AI research partner, delivering cutting-edge AI models and insights as a service to accelerate innovation across industries and scientific domains. |
Private Limited |
NA |
Yes |
We are applying to Avinya to gain mentorship, access to a network of deep-tech and AI experts, and support in scaling our AI research-as-a-service platform for industry and academic clients. |
alidate our AI service offerings, establish pilot collaborations with key clients, refine our go-to-market strategy, and explore opportunities for growth and funding |
Yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/QkQpjcKJGm7w-IndraAstra.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/QkQpjcKJGm7w-IndraAstra.pdf |
http://NA |
es. We are mission-driven: our goal is to accelerate innovation across industries and scientific domains by providing AI as a research service, enabling organizations and researchers to solve complex problems faster, more accurately, and efficiently. |
NA |
Sunil Khanna |
checked |
| Sep 20, 2025 @ 9:15 PM |
Swasthika Devadiga |
swasthika@calyirex.com |
https://www.linkedin.com/in/swasthikadev/ |
https://calyirex.com |
9449065724 |
Btech in CSE |
We first met in college and have known each other for 7–8 years. Together as a team, we’ve won four different government hackathons. |
Yes |
Technology and friendship. |
Calyirex Technology Private Limited |
https://calyirex.com |
|
Calyirex develops smart surveillance and data-driven platforms to help law enforcement and communities improve crime prevention, public safety, and operational efficiency. |
Police departments and communities rely on outdated, manual processes for tenant/servant verification, night patrols, and CCTV monitoring. These systems are reactive, slow, and prone to human error, resulting in delayed responses and lower public safety. |
Calyirex offers a unified platform Nazar that transforms existing CCTV networks and verification processes into smart, AI-powered systems. Our web and mobile apps streamline tenant and servant verification, map private camera data, optimize patrols, and deliver real-time alerts to officers. This reduces manual workload, improves response times, and strengthens community safety. |
Integrates directly with police workflows and community networks, turning existing infrastructure into an intelligent safety grid. Our close collaboration with law enforcement gives us deep insights into their operational challenges, enabling us to build practical, scalable, and defensible solutions that are hard for competitors to replicate. |
Product-Market Fit |
Pilots |
Our primary customers are law-enforcement agencies, city administrations, and smart-city projects seeking modern surveillance and internal-management tools. Secondary customers include gated communities, residential societies, and businesses that require advanced tenant/servant verification and real-time anomaly detection for CCTV networks. |
6B |
1.2B |
30M |
aw-Enforcement Agencies: One-time implementation fee of ₹5–15 lakh per department + annual maintenance/upgrade fees. Citizens: Premium subscription at ₹999/year for priority alerts, enhanced verification, and custom surveillance feeds. Businesses & Residential Societies: AI anomaly-detection service at ₹499/month per CCTV feed; gated communities pay ~₹1 lakh setup for CCTV integration. |
No direct competitor offers a single unified platform for police internal management + public-facing safety services. Partial competitors include standalone CCTV analytics firms, legacy verification services, or fragmented apps, but none combine real-time AI anomaly detection, background verification, and police-citizen collaboration in one system. |
Through direct engagement with police leadership, pilot projects with early adopters (like Jaipur Police), partnerships with smart-city integrators, government innovation challenges, and on-ground demonstrations to showcase measurable improvements in response time and cost savings. |
Pilot with progressive state or district police units to prove impact. Integrate with ongoing Smart City surveillance projects. Use existing wins, media coverage, and hackathon recognition for credibility. Build a citizen base through the Nazar app with low-friction onboarding and premium upsell. |
To become India’s leading public-safety technology platform, powering intelligent surveillance, background verification, and predictive policing for every state, while also extending our solutions to other sectors of national security, disaster management, and critical-infrastructure protection. Ultimately, Calyirex aims to be the “digital nervous system” of safer cities worldwide. |
Private Limited |
na |
Yes |
We’re applying to Avinya because it’s one of the few accelerator programmes in India deeply focused on AI driven innovation and real world impact. |
Strengthen our go to market and revenue models with expert mentorship. Build strategic partnerships with government bodies, smart city initiatives, and global technology leaders. |
hybrid |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/s4Vnb2VtHyeC-CALYIREX-compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/s4Vnb2VtHyeC-CALYIREX-compressed.pdf |
https://drive.google.com/drive/folders/13hjYNIN4yKvZlVY9xOCtdAT0iXPrc2JX?usp=sharing |
We are closing a long standing gap between communities and police. Our goal is not just to create a product but to bring about a positive behavioral shift in how public safety technology is used making it accessible, scalable, and impactful for millions of citizens. |
A strong woman in tech not underrepresented. |
na |
checked |
| Sep 20, 2025 @ 2:02 PM |
Ashrey Ignise R, Yash Vahi |
ashrey.ignise@artusai.co |
https://www.linkedin.com/in/aignise/ |
https://artusai.co |
+91 9999144778 |
Ashrey specializes in product design, brand development & fundraising. Yash specializes in deep learning, genAI, leading tech teams, and providing enterprise services. |
We met 5 years ago when we collaborated on a tech project. Ashrey was leading a small product team, and Yash was providing consultation & helping create the architecture for this. |
Yes |
We have published peer-reviewed research papers on the orchestration of large multi-agent AI systems. Ashrey has experience in enterprise sales, including being part of Uber Eats' launch team (New Delhi) and establishing a salesforce, while Yash is an award-winning AI developer who has developed AI tools for Fortune 500 companies. |
Artus AI |
https://artusai.co |
|
Artus is an AI Product Manager that large enterprises use to automate all their technical scoping requirements |
Enterprises with 10K+ employees in banking,healthcare & retail spend $1M and 9-12 months per project for the technical scoping of various IT projects before a single line of code is written. These firms are actively looking for ways to cut this cost and decrease time-to-market for their initiatives. |
Artus speeds up technical scoping of IT projects by automatically creating roadmaps, ROI models, time-cost estimates, wireframes, etc., and recommending solutions for each technical agenda tailored based on the company's internal data. Once business requirements have been accepted, it is automatically converted to a technical architecture. This process ends up saving companies 9-12 months and $250K-$1M per project. |
Our unique orchestration of 120+ agents that are uniquely trained on industry & project specific data makes our outputs unreplicable. Moreover the integration of the tool into deep enterprise workflows makes it hard for our customers to replace the tool once they are fully onboarded. |
Revenue |
Revenue |
Large enterprises with 10K+ employees in the USA in banking, healthcare, and retail industries. |
$30B |
$12B |
$60M |
Charge ~$150K per project + add-ons for additional users & AI credits |
Boutique consultancies that scoping projects are often outsourced too, and AI productivity tools like ChatPRD. |
Our warm network extends to CXOs of large corporations with whom we set up pilots + building a salesforce of senior AEs that come with existing relationships to decision-makers in our ICP |
Top-down: Target CIOs, Heads of Transformation, and PMO leads in Fortune 2000 enterprises. Proof-of-value pilots: Deliver first project in <14 days, prove 10× faster/cheaper vs Deloitte. Land & expand: Start with 1–2 scoping projects, expand into annual licenses. Enterprise integrations: Jira/Confluence/ServiceNow connectors drive adoption into workflows. |
To become the default AI Technical Consultant for enterprises — replacing scoping/advisory work that currently costs billions annually.
Every project in a large enterprise should start with Artus → generating board-ready packages, compliance artifacts, and execution-ready Jira tickets.
Long-term: Artus becomes the system of record for project discovery & governance, the layer between enterprise strategy and execution. |
Incorporated |
INR 35,00,000 |
Yes |
Expand our network, raise funds, and learn from GTM experts. |
Have avinya join our cap table as investors |
Both |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/J0ofvsdoRkXf-Artus-Vision-1.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/J0ofvsdoRkXf-Artus-Vision-1.pdf |
https://www.youtube.com/watch?v=n3v2TzmaLFA |
Our mission is to unlock enterprise innovation by collapsing the 9–15 months and millions of dollars wasted on scoping. Today, projects stall in bureaucracy; with Artus as an AI Technical Consultant, enterprises can move from idea to execution 10× faster and cheaper. The impact is both economic (saving millions per year) and societal (faster modernization in banking, healthcare, and retail that millions rely on daily). |
NA |
Souradeep Bharti |
checked |
| Sep 20, 2025 @ 9:35 AM |
Kiran Shankar Chawan and mamoon Jan |
kiranksc55@gmail.com |
http://linkedin.com/in/kiran-chawan-548539172/ |
https://datatukey.com/ |
8660682508 |
Analyst and Software Developer |
We met while participating in a hackathon, where we actively collaborated and discovered a critical problem in the data analytics field. Recognizing the opportunity, we decided to join forces and work on Datatukey, aiming to solve this challenge by building a unified, AI-powered data analytics platform. Our collaboration has been ongoing since then. |
Yes |
The Datatukey founding team combines expertise in AI, ML, Web3, and full-stack software development, with experience across 100+ hackathons—winning three and reaching finals in many.
Kiran Chawan – Founder & CEO: Leads product vision, strategy, and innovation. Selected for INSEAD AI Founders Cohort 2025. Experienced in building AI-powered platforms and business leadership.
Mamoon Jan – Co-Founder & CTO: Drives technical architecture, AI model integration, and platform scalability. Experienced in cloud solutions and advanced data systems.
Both founders bring strong entrepreneurial, technical, and leadership skills to revolutionize data analytics. |
Datatukey Private Limited |
https://datatukey.com/ |
|
Datatukey is an AI-powered platform that turns complex data into real-time, actionable insights with interactive visualizations and voice-enabled explanations. |
Datatukey is solving the problem of fragmented, complex, and time-consuming data analytics, where businesses struggle with multiple tools, manual data cleaning, and technical expertise requirements, resulting in delayed insights, inefficient decisions, and missed opportunities. |
Datatukey is an AI-powered data analytics platform designed to simplify and accelerate how businesses transform raw, complex, and scattered data into actionable intelligence. Traditional analytics tools are often fragmented, require technical expertise, and involve time-consuming manual processes, leading to delays and missed opportunities. Datatukey addresses this challenge by unifying data cleaning, analysis, visualization, and voice-enabled insights in one seamless platform.
The system leverages advanced AI to automate data preparation, generate real-time insights, and deliver interactive dashboards that empower users technical and non-technical alike to make smarter decisions instantly. Key features include generative interactive charts, psychographic analysis, conversational AI for natural data exploration, and AI-driven recommendations, all aimed at enabling businesses to “See Deeper, Decide Smarter.”
Datatukey is recognized by the Google AI Startup Program, Karnataka Innovation & Technology Society, and incubated at iHub Gujarat, validating its innovation and potential. The founding team brings vast experience in AI, ML, Web3, and software development, with over 100 hackathon participations, multiple wins, and selection for the prestigious INSEAD AI Founders Sprint 2025.
With its unique combination of automation, accessibility, and intelligence, Datatukey is building the world’s first truly unified AI-powered data platform, empowering organizations across industries to unlock the full value of their data. |
Datatukey is unique and defensible because it unifies automated data cleaning, AI-driven analysis, interactive visualizations, and voice-enabled insights into a single platform. Unlike traditional fragmented tools, it is accessible to non-technical users, scalable across industries, and continuously improves through AI learning, creating a strong competitive advantage. |
MVP |
Users, Pilots, Signups |
Datatukey’s target customers are SMEs, enterprises, financial institutions, and government agencies that rely on data-driven decision-making but struggle with fragmented tools and complex analytics. Key industries include finance, retail, healthcare, manufacturing, and technology, where actionable, real-time insights are critical for growth and efficiency. |
~$300B |
~$50B |
~$5–10B |
Datatukey follows a SaaS subscription model with tiered plans for SMEs and enterprises, complemented by enterprise licensing, premium AI features, and paid pilot projects. This ensures recurring revenue, diversified income streams, and scalability. Early pilot engagements validate demand, while value-added AI analytics and consulting services provide additional monetization opportunities. |
Databrick, Power BI, Tablueu |
Datatukey acquires customers through paid pilot programs, referrals, strategic partnerships, and industry events, supported by content marketing, webinars, and thought leadership. This approach validates product-market fit, builds trust, and drives scalable adoption across SMEs, enterprises, and data-driven organizations. |
Datatukey’s go-to-market strategy focuses on targeting SMEs and enterprises through paid pilot projects, strategic partnerships, and direct sales, complemented by digital marketing, webinars, and thought leadership content. We aim to validate product-market fit, demonstrate ROI quickly, and scale adoption via subscriptions and enterprise licensing. |
Datatukey’s long-term vision is to become the world’s leading unified AI-powered data analytics platform, democratizing access to actionable insights, empowering businesses of all sizes to make smarter decisions, driving innovation, and creating a future where data intelligence is accessible, reliable, and universally impactful. |
We are fully registered startup with DPIIT certificate |
NA |
Yes |
To get initial funding, Mentorship and GTM, |
During the program, we aim to refine Datatukey’s product-market fit, enhance AI features based on real-world feedback, and strengthen our go-to-market strategy. Additionally, we seek mentorship, industry insights, and networking opportunities to accelerate customer acquisition, scale operations, and position Datatukey for long-term growth. |
Anything okay for me |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/KMl1UnY9H2te-Datatukey.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/KMl1UnY9H2te-Datatukey.pdf |
https://www.youtube.com/watch?v=RX0YR9jwCWg |
Yes, Datatukey is a mission-driven startup focused on democratizing access to advanced data analytics. By simplifying complex data through AI, we empower businesses, governments, and underserved sectors to make smarter decisions, improve efficiency, enhance financial resilience, and drive inclusive, data-driven growth across industries. |
Yes I belong to SC Community |
NA |
checked |
| Sep 20, 2025 @ 2:35 AM |
Dr. Abhishek Gautam |
abhishek@ambula.app |
https://www.linkedin.com/in/abhishekkumargautam?utm_source=share&utm_campaign=share_via&utm_content=profile&utm_medium=ios_app |
https://ambula.app |
7903842725 |
Dr. Abhishek Gautam (Founder & CEO): MBBS from AIIMS, researcher in digital health, extensive work in healthcare digitization, AI in cancer analytics, and building scalable health-tech ecosystems for rural and urban India. |
We met during the early phase of Ambula’s inception through Odisha’s startup ecosystem. With complementary strengths in healthcare and technology, we have been working together for the last 3 years, building Ambula into a rural-first digital healthcare platform that bridges hospitals, doctors, pharmacies, and patients seamles |
Yes |
What’s your biggest strength as a team?*
Our biggest strength lies in complementary expertise and shared mission. Dr. Abhishek brings deep clinical and healthcare knowledge, while Ambika brings execution, operations, and technology strategy. Together, we combine medical credibility, operational excellence, and scalable technology—enabling Ambula to build trust with patients, doctors, and institutions alike. |
Ambula |
https://ambula.app/ |
|
A rural-first digital healthcare ecosystem connecting patients, doctors, pharmacies, labs, and hospitals through a single platform. |
Healthcare in India—especially in rural and semi-urban regions—faces three major challenges: Accessibility: Patients struggle to find timely, affordable, and trusted care. Fragmentation: Doctors, pharmacies, labs, and hospitals operate in silos with no unified platform. Continuity of Care: Patient records are scattered, leading to duplication, misdiagnosis, and treatment delays. This results in poor health outcomes, higher costs, and mistrust in the system. |
Ambula is a digital health platform designed to integrate the entire healthcare journey:
For Patients: Access to teleconsultations, doctor bookings, emergency response, e-clinics, labs, and pharmacy services through a single app.
For Doctors & Clinics: Tools for clinic management, prescriptions, and patient records with transparent revenue models.
For Institutions & Governments: Referral systems, ABHA-linked health lockers, and data-driven decision support for public health initiatives.
For Ecosystem: A rural-first network of e-clinics, digital hospitals, and telemedicine hubs ensuring last-mile delivery of healthcare. |
Rural-First Approach: Unlike urban-centric health apps, Ambula is designed for rural and semi-urban India where healthcare gaps are widest. Full Ecosystem Integration: Not just telemedicine—Ambula connects doctors, pharmacies, labs, hospitals, and patients on a single trusted platform. Scalable Hub-and-Spoke Model: Digital hospitals as hubs + local e-clinics as spokes ensure hybrid continuity of care. Trust & Partnerships: Backed by collaborations with government hospitals, medical institutes, and local health volunteers, giving Ambula credibility and reach. Data & AI Advantage: Building defensible value with ABHA-linked health lockers, AI-driven decision support, and predictive analytics for precision care. |
Users |
Signups |
Patients in rural and semi-urban India seeking accessible and affordable healthcare. Local doctors, clinics, and pharmacies looking for digital tools and patient flow. District hospitals, government institutions, and medical colleges aiming for digital referral and patient record systems. Private healthcare providers expanding into Tier-2/Tier-3 markets. |
TAM (Total Addressable Market):* India’s healthcare market is projected to reach $372B by 2025 (IBEF). Around 65% of India’s 1.4B population lives in rural/semi-urban regions, where healthcare digitization is still untapped. TAM ≈ $240B rural + semi-urban healthcare opportunity. |
SAM (Serviceable Available Market):* Digital health market in India is growing rapidly, estimated at $37B by 2030. Ambula’s SAM (rural-first digital health stack: teleconsultation, OPD, labs, pharmacy, e-clinics) ≈ $15B. |
SOM (Serviceable Obtainable Market):* With Odisha + Eastern India as initial focus (population ~100M), Ambula’s near-term SOM ≈ $500M market opportunity within 3–5 years. |
Doctors: Commission till 50 consultations a month and subscription exceeding the threshold Pharmacies & Labs: Commission on orders + API/Franchise integration. Institutions/Governments: SaaS fee for digital referral, ABHA-linked systems, and hospital digitization. Patients: Nominal subscription/health card fee for e-clinic and emergency services. |
Ekacare, Driefcase |
Grassroot activation: E-clinics in villages with local health volunteers. Institutional partnerships: District hospitals, medical colleges, state health missions. Digital onboarding: ABHA health card creation, app referrals, and community-based awareness drives. Trust-building: Field presence, transparent pricing, and local doctor engagement |
Phase 1 (Odisha, Eastern India): Deploy e-clinics, integrate with district hospitals, launch digital referral systems. Phase 2 (Scale-up): Expand to Tier-2/Tier-3 cities across India with a hybrid digital hospital + spoke model. Phase 3 (National network): Position Ambula as a trusted digital health backbone for rural-first India with AI-driven decision support and pan-India ABHA integration. |
To become India’s largest rural-first digital health ecosystem, ensuring every village and town has access to affordable, trusted, and continuous healthcare.
Our vision is to digitally connect 100M+ patients, 1L+ doctors, and 50K+ pharmacies/labs by 2030, creating an AI-powered healthcare backbone for Bharat that integrates with both private and government systems. |
Yes |
80 Lacs |
Yes |
We are applying to Avinya because it aligns with Ambula’s mission of building impactful, scalable solutions for Bharat. We see Avinya as a platform that not only provides resources but also strategic guidance on scaling health-tech in underserved regions. |
Refine our business model and unit economics for sustainable scale. Build deeper institutional and government partnerships for rural health deployment. Strengthen our technology stack (AI-driven health records, e-clinic modules, ABHA integrations). Prepare for seed/Series A fundraising with clear traction and validated revenue models. Learn from mentors, peers, and Avinya’s ecosystem to position Ambula as a national-level rural health platform. |
Yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/iJj7ZxEQ3bDe-Updated-AIO-Deck.pdf_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/iJj7ZxEQ3bDe-Updated-AIO-Deck.pdf_compressed.pdf |
https://youtu.be/11N7ZWMjEw0?si=jGhHRHUdxvOeyYo0 |
Yes — Ambula is a mission-driven and impact-focused startup.
Our core mission is to bridge the rural–urban healthcare gap in India by building a digital-first healthcare ecosystem that ensures accessibility, affordability, and continuity of care for every patient, regardless of geography or income.
While most health-tech platforms are urban-centric, Ambula was designed with a rural-first approach, focusing on the 900M+ Indians living outside metros who often lack timely access to quality care. Through e-clinics, telemedicine, ABHA-linked health lockers, and a hub-and-spoke model of digital hospitals, we bring doctors, pharmacies, labs, and emergency care to the last mile.
Our impact focus goes beyond business growth:
Empowering rural doctors and pharmacies with consistent patient flow and digital tools.
Reducing healthcare inequity by digitizing medical records and referrals at scale.
Partnering with governments, institutions, and communities to strengthen public health delivery. |
Yes |
IIT Bhubaneswar |
checked |
| Sep 19, 2025 @ 11:28 AM |
Owais Chunawala |
owais@augle.ai |
https://www.linkedin.com/in/owais-chunawala-70b228136/ |
https://www.augle.ai/ |
+91 81081 48186 |
Augle AI is a pioneer in AI-driven industrial automation, founded by IIT-Bombay engineers. We specialize in computer-vision-based automation solutions, empowering industries to enhance productivity and precision. Our portfolio includes Face Recognition systems (for attendance, access control, and customer analytics), Crowd Analysis using drones and CCTV, OCR, and advanced Object Detection. Beyond software, we deliver innovative industrial automation products such as laser glass thickness measurement systems, robotic sheet-metal bending solutions, and other custom AI-integrated robotic systems for manufacturing. Our mission is simple yet powerful — to accelerate global adoption of automation. Augle AI’s solutions are deployed across retail, manufacturing, real estate, security & surveillance, and government sectors. Our innovation has been recognized with Gold at Automation Expo 2025, E-Cell’s Eureka 2019 (IIT-Bombay), Startup Jalsa 2018, and finalist position at the International Entrepreneurship Competition 2018, NTUST Taiwan. |
Augle AI was incorporated on 22 June 2018 by our founder, an IIT Bombay graduate. The initial years were dedicated to deep R&D, prototyping, and building the core technology stack. Our current team of engineers has been working together for over two years, focusing on product development and commercialization. This combination of long-term research and recent execution has positioned us strongly to scale innovative AI-driven robotic solutions. |
Yes |
Our biggest strength is the depth and diversity of our team.
We bring together highly qualified professionals across software development, computer vision, mechanical engineering, project & product management, sales, business development, HR, finance, and accounting — creating a well-rounded foundation to execute complex automation projects end-to-end.
Our founder is an IIT-Bombay alumnus, which reflects the strong technical and entrepreneurial DNA of the company. This blend of technical expertise, operational excellence, and business acumen enables us to innovate rapidly, solve challenging problems, and deliver world-class AI-driven automation solutions. |
Augle A.I. Pvt. Ltd. |
https://www.augle.ai/ |
|
Augle AI builds AI-driven computer vision and robotic automation solutions to accelerate the adoption of automation across industries. |
We are solving the problem of inefficiency, inconsistency, and high dependency on manual processes in industrial operations. Current manufacturing and monitoring systems are often labor-intensive, error-prone, and slow to scale. Augle AI eliminates these challenges through AI-driven automation and computer vision, enabling faster, more accurate, and cost-efficient operations. Our solutions empower industries to reduce human dependency, improve productivity, and achieve higher precision while staying scalable for future growth. |
Our solution enables robots to see, understand, and act autonomously in complex industrial environments. Using a 3D camera to capture precise real-world data and our proprietary AI software, we generate accurate point clouds, understand object geometry, and automatically create robot paths. This combination transforms standard robots into intelligent, multifunctional systems that are application-agnostic, robot-agnostic, and camera-agnostic.
For example, our Laser Surface Profiler generates 3D point clouds for dimension analysis, CAD model mapping, welding seam tracking, and sealant bead analysis, enabling robots to achieve superior precision and repeatability. We are also developing a robotic sheet metal bending solution that automates handling and bending of large, heavy sheets—eliminating labor shortages, reducing operator risk, and improving productivity for manufacturers.
Our technology integrates seamlessly with welding, surface finishing, laser cutting, material handling, quality inspection, and assembly applications, providing a single platform for automation. By reducing manual intervention and safety hazards, Augle AI delivers cost-effective, scalable, and future-ready automation to factories. |
Our solution is unique because it integrates AI, 3D vision, and robotics into a cost-effective, hardware-agnostic system. Our Laser Surface Profiler generates precise 3D point clouds for real-time defect detection, classification, and automatic robot path planning—reducing downtime and human intervention. Our robotic sheet metal bending solution addresses labour shortages and safety risks, handling large, heavy components with minimal operators. Proprietary motion algorithms, AI software stack, and deep domain expertise make our solution highly defensible and difficult to replicate, offering a clear technological edge and scalable integration with existing manufacturing setups. |
Product-Market Fit |
Users |
Our target customers are OEMs, Tier-1 suppliers, and system integrators across industries such as automotive, aerospace, heavy engineering, construction equipment, and metal fabrication. We specifically work with OEMs and Tier-1 manufacturers who seek to automate labor-intensive processes like welding, sheet metal bending, and assembly, improve quality consistency, and enhance operator safety. We also collaborate with machine builders and automation solution providers to integrate our AI-powered 3D vision systems into their robotic cells, enabling scalable, high-precision automation. |
The global industrial automation market is projected to exceed USD 400+ billion by 2030, with AI in manufacturing alone growing to USD 68+ billion by 2032. This includes machine vision, robotics, and process automation—key areas Augle AI operates in. The need for automation is fueled by rising labour costs, safety regulations, and Industry 4.0 adoption worldwide. |
In India, the industrial automation market is valued at USD 15+ billion and growing at ~12% CAGR, driven by government incentives for manufacturing, Make-in-India initiatives, and adoption of smart factories. Within this, the machine vision and robotic sheet metal automation segment represents a USD 2-3 billion opportunity, with strong demand from automotive, heavy machinery, and fabrication units facing labour shortages and quality challenges. |
Given current capacity, sales reach, and adoption rates, Augle AI realistically targets 1-2% of India’s machine vision and automation market over the next 3-4 years. This represents a revenue potential of USD 5-15 million (₹40-₹120 crore) in the near term. |
Augle AI follows a B2B solution-driven revenue model. We generate revenue through: Direct Product Sales – One-time sales of robotic bending systems, laser profilers, and vision-based automation solutions. System Integration & Commissioning – Project-based revenue for custom integration with client machinery. Annual Maintenance Contracts (AMC) – Recurring revenue from service, support, and software updates. Software Licensing & AI APIs – Subscription or usage-based fees for computer vision and automation software.(This will be started soon) Consulting & Training – Additional revenue from process design, upskilling, and on-site training. This mix ensures upfront cash flow + recurring income, creating a scalable and predictable business model. |
A) Closest overlaps: 3D vision–first robotics / vision‑guided robotics 1. *CynLr (Cybernetics Laboratory, Bengaluru)* – Builds a vision‑first robotic platform that aims to let robots “grasp any object” without prior training; multi‑arm vision‑guided manipulation (“CyRo”). Strong fit with universal 3D vision bin‑picking and assembly. ([Cybernetics Laboratory - CynLr][1]) 2. *DiFACTO Robotics & Automation (Bengaluru)* – Large Indian robot systems integrator that explicitly integrates *“robots with intelligent vision systems”* across automotive and other factories; recently *acquired RoboFinish IPR* (robotic grinding/finishing) from Grind Master, broadening advanced cells where vision is critical. ([DiFACTO][2]) 3. *Wipro PARI (Pune)* – India’s biggest turnkey robotics integrator; publishes machine‑vision guidance capabilities (robot guidance, alignment, dimensional measurement) and showcases EV/battery applications using Cognex vision for robot guidance. ([Wipro PARI][3]) 4. *Gridbots (Ahmedabad)* – Indigenous maker of manipulators/AGVs with *machine vision* products for industrial inspection and handling—often sold as complete robotic systems. ([GridBots][4]) 5. *Qualitas Technologies (Bengaluru)* – Industrial machine‑vision specialist with the *EagleEye* deep‑learning platform; provides *3D machine‑vision* inspection and often pairs solutions with robots in production lines. ([Qualitas Technologies][5]) --- ## B) Strong adjacencies: vision platforms and Indian robot OEMs used in VGR cells 6. *Soliton Technologies (Bengaluru/Coimbatore)* – Long‑time Indian machine‑vision provider; launched India’s first indigenously designed industrial machine‑vision camera/smart camera (NUERA). Typically partners with integrators for robotic inspection/measurement cells. ([Soliton Technologies][6]) 7. *SensoVision Systems (Pune)* – Vision‑based *optical sorting and inspection* for fasteners/auto components; competes for the “AI + vision” QC budget that’s often tied to robotic handling. ([sensoviz.com][7]) 8. *Robro Systems (Pune)* – Camera‑based inspection & monitoring applications (defect sorting, print inspection, counting/packing) deployed on production lines; frequently integrated with robotics/end‑of‑line automation. ([robrosystems.com][8]) 9. *Systemantics (Bengaluru)* – Indian OEM for collaborative/industrial robots (*ASYSTR* series). While not a vision company, their cobots are commonly deployed with 2D/3D vision for pick‑place, tending, inspection—so they compete at the “robot + vision cell” level. ([Systemantics][9]) 10. *Svaya Robotics (Hyderabad)* – Indian cobot OEM positioning with *built‑in sensing and machine vision* for AI‑enabled applications (assembly, inspection, polishing). Like Systemantics, they compete when customers want a local cobot + vision stack. ([Svaya Robotics][10]) --- ## C) Broader Indian automation players you’ll meet in bids 11. *m2nxt (BFW subsidiary, Bengaluru)* – Robotics/Industry 4.0 solution provider from Bharat Fritz Werner; implements robotic cells in Indian factories and partners for vision where needed. ([m2nxt.com][11]) 12. *Maxbyte Technologies (Coimbatore/Chennai)* – Integrates *intelligent robotics & automation* (cobots/AMRs, adaptive automation) inside digital‑factory programs; will overlap in brownfield modernization with vision‑guided cells. ([maxbyte.co][12]) |
We acquire customers through industry expos, targeted demos, and proof-of-concept deployments. Our dedicated sales, marketing, and business development teams actively engage key decision-makers across manufacturing and automation sectors. Strategic partnerships with system integrators and OEMs help expand reach. We focus on ROI-driven problem-solving, converting early adopters into reference customers, fueling credibility, repeat orders, and enterprise-wide rollouts. |
Our GTM focuses on Tier-1 manufacturers, OEMs, and global automation leaders. We drive adoption through strategic PoC deployments, live demos, and pilot programs that showcase measurable ROI. We maintain a strong presence at industry expos and technology forums, positioning Augle AI as the preferred partner for 3D vision-guided robotics. By building deep partnerships with OEMs, robot integrators, and automation ecosystem players, we enable faster integration, reduce engineering complexity, and scale adoption across industries. |
Augle AI’s vision is to become the global standard for AI-powered robotic automation, making factories fully autonomous, safer, and highly efficient. We aim to be robot-agnostic, camera-agnostic, and application-agnostic, providing a unified software layer that powers everything from sheet-metal bending to precision welding and inspection. Our long-term goal is to transform over 10,000 factories worldwide into smart, automated, and data-driven facilities, reducing downtime and labor dependency while increasing throughput and quality — ultimately accelerating the global adoption of Industry 4.0. |
22 June 2018 |
17,50,000 (in 2020) |
Yes |
We see Avinya as a strategic partner to accelerate our journey from early commercialization to large-scale deployment. With our first units sold and purchase orders already in place, we are ready to scale manufacturing, strengthen our technology stack, and expand market reach. Avinya’s mentorship, network, and potential funding will help us refine our business model, access Tier-1 customers, and fast-track global partnerships with OEMs and integrators — bringing our vision of fully automated, safe, and efficient factories closer to reality. |
I want to leverage the programme to build a sustainable and scalable business that solves real, pressing problems in the market. By validating the product-market fit, refining the business model, and gaining access to industry expertise and funding, I aim to create solutions that not only address customer pain points but also drive measurable growth, operational efficiency, and long-term value. Ultimately, I want to build a venture that delivers both social impact and business success. |
We are open for both according to availability. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qlXSLqV694tv-Augle_AI_PitchDeck_cc.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/qlXSLqV694tv-Augle_AI_PitchDeck_cc.pptx |
https://www.youtube.com/watch?v=Gb1I7KVbWW8 |
Yes — Augle AI is a mission-driven startup. Our mission is to accelerate the adoption of automation globally and make advanced robotics accessible to every manufacturer. We address critical issues like labour shortages, workplace safety, and production inefficiencies, creating safer and more productive factories. By reducing manual intervention in hazardous tasks like sheet-metal bending, we prevent workplace injuries and enable human talent to focus on higher-value work — delivering both economic and social impact at scale. |
Yes. We are proud to be part of an underrepresented group in deep-tech and robotics, building advanced AI-driven automation solutions from India. Our journey reflects the belief that world-class innovation can come from emerging markets, not just Silicon Valley or Europe. By building locally and solving global problems, we hope to inspire more founders and engineers to take bold steps in shaping the future of manufacturing. |
NA |
checked |
| Sep 18, 2025 @ 12:51 PM |
Hariharakrishnan Mannarsamy | Britto Franklin Joe Ambrose |
hari@trustscapeai.com |
https://www.linkedin.com/in/hariharakrishnan-mannarsamy-36030a2/ |
https://trustscapeai.com/ |
Hari: +91 98451 70652 | Britto: +91 99024 91671 |
Hari (Co-Founder & CTO at TrustscapeAI ; Product Commercialization Manager, Partner Enablement manager, Technology manager at IBM/HP India, Verifone and PSI Bull) ; Britto (Co-Founder at TrustscapeAI ; VP of People & Culture at XOXODAY ; Senior Director – People & Culture at Aditi Consulting ; HR Leader specializing in Organizational Culture, People Development & Strategic Administration ; MBA – HR, IBS Bangalore) |
We have known each other for a couple of months and were introduced by our mentor, who has known us individually for over 20 years. |
Yes |
The co-founders bring complementary skills and experience essential for this startup endeavor.
Britto has over 15 years of strong domain expertise in HR and recruitment, with deep awareness and conviction about the issues being addressed by the proposed product.
Hari brings more than 30 years of experience in software product development, spanning hands-on, managerial, and customer-facing roles, enabling him to understand, envision, build, deliver, and support creative, economical, and effective technical solutions for customers. |
TrustscapeAI Technologies Pvt. Ltd. |
https://trustscapeai.com/ |
|
TrustScape is an AI-powered interview intelligence platform that evaluates not only skills but also emotional and trustworthiness attributes, such as consistency, clarity, confidence, and integrity, by analyzing speech, video, body language, and digital footprints, helping companies hire credible and reliable talent. |
Resumes are polished, interviews are subjective, and candidates are increasingly coached by AI. Employers struggle to assess trustworthiness, credibility, and emotional stability, qualities that are critical for roles involving customers, teams, and sensitive data. |
TrustScape brings science to trust evaluation through AI-driven interview analysis:
Consistency Check: Cross-verifies claims across resumes, LinkedIn, and interview answers.
Integrity Signals: Detects alignment between words, tone, and body language.
Emotional Stability: Measures resilience and composure under pressure.
Clarity & Confidence: Evaluates coherence and authenticity of responses.
Body Language Alignment: Tracks posture, gaze, and gestures for congruence.
Our MVP already delivers foundational response evaluation and emotional stability analysis with structured reports. The next milestone is to integrate digital footprint insights from social media, enabling the creation of a holistic Trust Score for every candidate. |
TrustScape is unique because we measure what others don’t, trust. While traditional platforms assess skills or sentiment, we quantify credibility, integrity, and emotional stability. Our defensibility comes from a self-reinforcing data moat: with every interview, TrustScape learns, refining its Trust Scoring algorithms and building benchmark datasets that competitors can’t replicate. The more it’s used, the smarter and more unassailable it becomes — positioning TrustScape as the global credibility layer for hiring. |
MVP |
Pilots |
Our primary customers are enterprises in India that hire at scale for soft-skill-heavy, customer-facing, and trust-critical roles. This includes: - IT services & consulting firms managing high-volume, client-facing hiring. - BFSI and fintech companies where credibility and integrity are essential. - Retail and e-commerce organizations with large, churn-heavy frontline teams. - Healthcare and education providers where empathy and trust matter as much as skill. Sales and customer experience roles across industries. |
The Indian Interview Intelligence market was valued at ₹312 crore in 2024 and is projected to reach ₹1,248 crore (by 2030, growing at a CAGR of ~26% |
Hiring for soft-skill-heavy roles in IT services, BFSI/fintech, retail/e-commerce, healthcare, and sales/customer experience—estimated 10–20M interviews annually. |
Capturing even 1% of this volume (~200,000 interviews) represents a multi-hundred-crore INR annual opportunity in India before expanding globally. |
Per-interview pricing for SMEs and startups. Tiered SaaS subscriptions for large companies API integrations with ATS and video platforms, monetized per seat or per interview. |
Global: HireVue, Modern Hire, Pymetrics. India/Asia: Talview, Wheebox, Interviewer.AI. TrustScape differentiates by focusing squarely on emotional credibility, integrity and trust which competitors do not address. |
Direct sales to HR leaders in IT services, BFSI, and retail/e-commerce. Partnerships with recruitment platforms and ATS providers. Thought leadership pilots demonstrating ROI and bias reduction. Leveraging founder networks, HR forums, and LinkedIn engagement. |
Phase 1 (6 months): Paid pilots with IT services and BFSI enterprises; refine MVP and target target market fir. Phase 2 (next year): Expand into retail/e-commerce, healthcare, and consulting; integrate with ATS. Phase 3 (following years): Launch TrustScape Benchmarks (industry-wide trust datasets) and expand into other geographies |
To establish TrustScape as the global standard for credibility in hiring, where every candidate carries a Trust Score alongside their resume. We aim to make hiring fairer, bias-free, and data-backed, helping organizations reduce risk while giving genuine candidates a level playing field. |
Incorporation in Progress |
Self-Funded ; ~Rs 2 Lakhs |
Yes |
Avinya offers the ideal platform to accelerate product development, validate our GTM strategy in India’s rapidly growing hiring-tech space, and connect with early enterprise adopters. Access to mentorship, corporate networks, and investors will help scale TrustScape from MVP to enterprise-ready product. |
Build a v1 product with digital footprint integration. Secure 3–5 paid-pilots with enterprises and iteratively improve the product based on their feedback Finalize pricing and GTM strategy. Drive to Product-Market-Fit Position for seed funding. |
Yes — fully open to hybrid or virtual modes. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/oMUQa5NZ93xW-TrustScape-Investor-Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/oMUQa5NZ93xW-TrustScape-Investor-Pitch-Deck.pdf |
https://youtu.be/lDYYE6d05ZA |
Yes. TrustScape is deeply mission-driven.
At its core, we are building a system of trust and fairness in hiring. Today, interviews are riddled with bias, subjectivity, and surface-level judgments. This hurts both employers and candidates: companies suffer mis-hires, while talented individuals without the right “polish” are overlooked.
TrustScape’s mission is to level the playing field. By using AI to evaluate candidates on credibility, integrity, and emotional balance, we help organizations build more ethical, trustworthy teams while giving deserving candidates—often from underrepresented or non-traditional backgrounds—a genuine chance to be seen for who they are, not just what’s on their resume. |
No |
AIPod |
checked |
| Sep 18, 2025 @ 9:32 AM |
Vincent D'Souza |
vincent.dsouza@madvr.in |
https://www.linkedin.com/in/vincentad%20 |
https://www.madvr.in/ |
+919972667982 |
CEO & Co Founder , Entrepreneurial Background |
We met through an event BHX - Beach House Project where we presented our cases of our future plans. Considering differences in the skills, and potential to compliment each others' skills, we came together to start this organization, MadVR Solutions |
No |
Our biggest strength as a team is our complementary expertise—Dr. Keyur Sorathia brings deep technical and domain knowledge in Virtual Reality and immersive technologies, while Vincent D’Souza adds 20+ years of experience in business development, partnerships, and scaling startups. Together, we combine innovation with execution, enabling us to build cutting-edge VR solutions and take them to market effectively. |
MadVR Solutions Private Limited |
https://www.madvr.in/ |
|
MadVR Solutions is a cutting-edge health-tech startup revolutionizing surgical training and education through advanced Virtual Reality (VR) and real-world haptics |
1. Lack of access to cadavers, simulators, and specialized training courses. 2. Prolonged learning curve for surgeons (5–8 years to proficiency). 3. High costs and ethical challenges in traditional surgical training, especially when using cadavers and real patients. 4. Inefficiency in demonstrating the full potential of surgical tools and implants. 5. Global challenges in marketing surgical tools with limited interactivity |
Surgical simulators created through advanced VR technologies targeted to surgical instrument manufacturing companies, medical colleges and universities and professional surgical training institutes. Following product names
1. Collaborative Immersive Surgical Suite - for learning, training and education
2. Immersive Product Showcase - for marketing and product showcase
3. Haptic Immersive Surgical Suite - Enhanced haptic-based suite for learning, training and education |
1. Haptics integration, enabling the feeling of skin, bones, tissues, muscle, and force feedbacks - cadaver grade training and education 2. Cost-effective compared to the competitors across the world |
Revenue |
Pilots |
1. Educational Institutions: Medical colleges, and Universities. 2. Medical Device and Surgical Instrument Manufacturers: Companies training surgeons using their own tools. 3. Professional Training Institutes: Medical training institutes that train students and practicing surgeons on advance surgical procedures |
USD 5.62 Billion |
USD 3.88 Billion |
USD 600 Million |
● Surgical Instrument Manufacturing Companies o Hourly subscription-based licensing for surgical training modules. o One-time fees for custom VR showcase and hardware. ● Medical Colleges and Universities o Yearly subscription licensing fees with 20% increase in the subscription cost each 3 years |
Osso VR (USA) FundamentalVR (UK) Touch Surgery (UK) VirtaMed (Switzerland) ImmersiveTouch (USA) Medisim VR(India) Mimyk (India) |
Marketing plan Attending Medical Conferences, Seminars, Workshops and Expos 1. Demonstrate VR surgical training modules at events attended by decision-makers from medical colleges, universities, and device manufacturing companies. 2. Set up engaging booths and showcase demos, videos, and case studies to highlight MadVR’s innovative solutions. 3. Network with key stakeholders like deans, faculty, and administrators to build interest and credibility. Personalized Outreach 1. Reach out to marketing and sales teams of medical device manufacturers with targeted email campaigns, virtual and in-person demos, and presentations. 2. Follow up consistently to build trust and long-term relationships. Content Marketing Post regularly on platforms like LinkedIn and optimize the website for search engines using relevant keywords. Strategic Partnerships Collaborate with medical associations and healthcare education boards to align with initiatives aimed at enhancing training standards. Sales and distribution strategy 1. Segment-Specific Targeting: ● Medical Colleges and Universities: o A dedicated outreach plan to identify and target top-tier medical institutions, i.e., targeting tier-2 and 3 private medical colleges as primary customers o Customized demos and proposals addressing their unique training challenges. ● Surgical Instrument and Device Companies: o Highlight how MadVR’s VR simulations can enhance their marketing and training efforts for tools and implants. ● In-Person Demonstrations to all customer groups o On-site demonstrations of the VR surgical training modules for prospective clients. o Provide real-world use cases during sales presentations to illustrate measurable benefits. ● Partnering with Senior Doctors o collaborate with renowned surgeons as our sales partners to promote MadVR solutions and connect with decision markers for all customer groups ● Cold Outreach and Lead Generation: o Data-driven tools like LinkedIn Sales Navigator to identify potential leads. o Personalized outreach emails, including details about MadVR’s offerings and tailored benefits. ● Customized Packages and Flexible Pricing: o Tailored packages for clients based on their needs (e.g., medical colleges versus device manufacturers). o Offering of subscription models or tiered pricing to cater to varying budgets and usage levels. ● Client Retention Initiatives: o Ongoing support to existing clients with training sessions and customer success programs. o Ensure clients see the long-term value of MadVR’s solutions by regularly showcasing updates and improvements |
While MadVR’s primary focus remains B2B, we are building a complementary B2C channel to capitalize on the growing global VR hardware footprint. B2B: Institutional Channels 1. Medical Colleges & Universities (India) Channel-Led Distribution o Strategy: Engage premier education consultancies/consultants as channel partners. o Execution: Education consultants as channel partners with already built-in network and connects, targeting each department (e.g., Ortho, Surgery, ENT) with tailored VR Skill Lab packages. Provide partners with white-labeled sales collateral, dedicated demo access, and performance-based incentives. Deliver targeted webinars and on-campus demos for Presidents, Directors, and Chairs to accelerate adoption and procurement. 2. Medical Device Manufacturers Influencer-Led Partnerships & Custom Solutions o Strategy: Collaborate with known surgeons, authority voices, and Key Opinion Leaders (KOLs) to introduce MadVR as the de facto training platform for their instruments and implants. o Execution: Co-develop (with KOLs, and manufacturers) branded VR simulations featuring each manufacturer’s proprietary tools. Enable manufacturers to deploy customized modules at CME events, in-house training centers, and partner medical schools. Offer joint marketing materials and case studies that showcase improved trainee proficiency and tool adoption. 3. Simulation and Training Institutes Franchise & Partnership Model o Strategy: A franchise program for simulation centers and hospital-based training academies. o Execution: Offer franchise agreements (revenue-sharing and non-revenue-sharing) to co-develop and deliver bespoke surgical curricula. Equip partners with turnkey VR labs, content updates, and certification programs. Empower franchisees to monetize through pay-per-use VR access, structured certificate courses, and sponsored training events. B2C: Individual Surgeons & Trainees 1. Postgraduate Residents & Practicing Surgeons Meta Store Subscription Model for Worldwide Reach o Strategy: Launch MadVR+ Monthly Pass on the Meta Quest Store, targeting self-learners seeking on-demand practice. o Execution: Offer a subscription granting unlimited access to our full library of surgical modules. Provide flexible purchase options—bundle pass or à la carte module purchases. Promote through professional social channels, surgical societies, and targeted digital ads to maximize global reach. |
Revolutionizing Medical Training and Education Through Advanced VR and Haptics |
Incorporated |
USD 280,000 |
Yes |
For funding, networking & engagement |
Funding & Networking |
Yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/rWatHiiVnI4p-MadVRs-Pitch-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/rWatHiiVnI4p-MadVRs-Pitch-Deck.pdf |
https://youtu.be/VmPISldimkU?si=4UiK6YTkEuwzpWB9 |
MadVR is both an impact-driven and mission-driven startup.
We are mission-driven because our core purpose is to redefine surgical training—making it more accessible, immersive, and effective through VR and haptic technology. Our mission is to reduce surgical errors and ensure every surgeon has the opportunity to practice and master critical procedures safely.
We are also impact-driven, as our solution directly addresses the global shortage of surgical training resources, especially in emerging markets. By offering a cost-effective, scalable alternative to cadaver-based training, we’re improving healthcare outcomes, saving lives, and democratizing access to world-class surgical education. |
Yes |
N/A |
checked |
| Sep 17, 2025 @ 8:52 PM |
Arpit Gupta and Rakhee Singhal |
callarpit2002@yahoo.co.in |
https://ArpitGuptaPhD |
https://ActionableAnalytics.Biz |
9310383708 |
CTO and CPO |
over 13 years in the USA and now are married and moved to India this year. |
Yes |
Trust, Ideation and execution |
AGACP |
NA |
|
Multi Agent framework to solve enterprise problems |
Hallucination, explainability and task division |
Given a problem
Divide it in WBS and each sub problem is handeled by an Agent pair(s)
All the agents can communicate with each other until the complete problem is solved. |
There is no other solution in place that solves for hallucinations, explainability by leveraging Digital Twins, |
Idea |
Testimonials |
enterprises |
$500B |
$200B |
$50M |
license fees |
hyper scalers, AI native platforms |
sales cycle |
partnerships |
Autonomous healing of problems faced by enterprises |
NA |
NA |
Yes |
to get started |
funding and connections |
Virtual |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/IhVnSSOuBkfa-ActionableAnalytics-AGACP-Pitch-Arpit-Gupta.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/IhVnSSOuBkfa-ActionableAnalytics-AGACP-Pitch-Arpit-Gupta.pdf |
https://www.youtube.com/watch?v=79NQhe7iaA4&t=10s&pp=ygUQUmVhbHRhbGtzMSBBR0FDUA%3D%3D |
NA |
NA |
NA |
checked |
| Sep 17, 2025 @ 5:12 PM |
D Chandrasekhar |
captdcsekhar@gmail.com |
https://www.linkedin.com/in/dcsekhar/ |
https://www.betatankrobotics.com |
9980222870 |
D Chandrasekhar is the founder, designer of BetaTANK and is an Ex Merchant Navy Captain, studied finance management in Singapore and executive management at IIM Bengaluru. |
Ms.Vinutha M is the other non executive director in the board. She looks after the ROC compliances, audits and HR functions of both companies i.e www.alphamers.com and www.betatankrobotics.com
In effect the company has one founder D Chandrasekhar |
Yes |
Our ability to innovate and solve current day problems is our biggest asset. Other factors like a culture of innovation, hands on involvement of promoter with 3 decades of cross industry experience also helps our efforts.
AlphaMERS has developed the river plastic cleanup technology in the world using only flow of river, which has arrested the most plastics in the world, has the longest floating barrier in the world and is the most installed technology in the world. AlphaMERS has also developed an Ocean wave energy harvester and cutting edge technology to respond to marine oil spills. BetaTANK has developed specialized petroleum tank cleaning robots controlled and powered by hydraulics and certified to operate in the most explosive zone as per internationally laid down safety standard. |
Beta Tank Robotics Pvt Ltd |
www.betatankrobotics.com |
|
BetaTANK was in specialized robotics, now entering crowd movement simulation. |
We are simulating crowd movement in given layout and predicting inefficiencies and stampedes. |
Crowd movement is like hydrodynamics. There is a lot of relationship between the density of the crowd, slope, profile, many such factors and the overall crowd pace. We have collected and analyzed a lot of crowd movement data and observing this relationship in a data base which is increasing daily. We shall further create UI to visualize the crowd movement. Anyone who is building a stadium, a mall or a railway station will turn to this software with the building plan to visualize in normal and emergency evacuation mode, the movement inefficiencies when passing obstructions and around corners, with different crowd profiles, and various such parameters. This will help predict where critical constriction or pile up happens or chances of stampede occurs. Ideally the fire department or insurance will seek a report from such a design support software before approval of a building plan. |
We have thought through the pattern, done a lot of research and will continue to hold the first mover advantage with extensive research and tools for the end user.. |
MVP |
Testimonials |
City fire departments or other authorities approving plans of large public buildings |
1000 towns and cities x 50 buildings x 1 lakh each = 500 crore in India |
100 Crore in 3rd year as software and tools mature |
NA |
SAAS - provide simulation and analysis report |
https://www.thunderheadeng.com, https://www.ucrowds.com, https://www.anylogic.com |
Canvas with fire dept and insurance companies to make this report necessary for building plan clearance. |
Publicize this software in civil engg and architects exhibitions, pilot for large outdoor events. |
The software will in due course meet the following customer needs -
A. Predict efficiencies of movement and potential stampede locations and circumstances
B. Real time crowd assessment from CCTV or drones which flag off potential disasters inside large buildings or large outdoor events
C. Predict crowd management needs and vulnerable locations in emergency and panic evacuation situations. i.e stadium, movie theatres, hospitals.
D. Later develop software on similar principles for vehicular traffic movement to help city traffic planners |
Incorporated in 2019 |
245 lacs for robotics; Nil for this project |
Yes |
Seems to be a good program. |
Interact and understand various viewpoints, incorporate what applies to this project |
Virtual |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/02PP7jo2LzuX-Beta-Slide-Deck-170925.pdf, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/hnFXxFliaWG2-AlphaMERS-slide-deck-Sep-25.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/02PP7jo2LzuX-Beta-Slide-Deck-170925.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/hnFXxFliaWG2-AlphaMERS-slide-deck-Sep-25.pdf |
https://youtu.be/YIRFZBK0DUY%20%20%20%20%20%20%20 |
This software will fill a critical gap in the resources available for the fire department to assess human risks, or an outdoor event manager to understand the risks, develop SOPs and monitor in real time the developing risks. Now the planner or decision maker will have visualizable and quantified data to base his decisions or to defend his decisions in the aftermath of incidents. This will have a big impact on the numerous incidents of stampedes in public places including temples, railway stations, stadiums or huge religious gatherings. |
No |
IITG TIC |
checked |
| Sep 16, 2025 @ 9:32 PM |
Ravi Puvvala, Subu Gupta, Dr Tushar Ravuri |
ravi@quantumbiosciences.ai |
https://linkedin.com/in/ravipuvvala/ |
https://quantumbiosciences.ai |
9980844223 |
Ravi Puvvala, Managing Director and Subu Gupta, Chief Technology Officer and Dr Tushar Ravuri, Chief Scientific Officer |
Ravi and Subu first met during their undergraduate and master’s programs and have been close friends for over 30 years. Ravi and Dr. Tushar worked together as colleagues in the mid-2000s and have maintained a strong friendship for nearly 20 years.
Ravi has founded and successfully exited two companies, bringing seasoned entrepreneurial experience. Together, we have united our complementary backgrounds—entrepreneurship, quantum physics, and large-scale AI/ML engineering—to build Quantum Biosciences. This long-standing trust and diverse expertise form the foundation of our collaboration and our ability to tackle complex deep-tech challenges. |
Yes |
The biggest strength of the Quantum Biosciences team is its deeply interdisciplinary leadership—a rare combination of quantum physics, advanced engineering, AI/ML, and clinical oncology expertise working as one unit.
Quantum Science & Technology: Dr. Tushar Ravuri brings PhD-level expertise in quantum systems, ensuring the NV-diamond sensing platform is scientifically rigorous and cutting-edge.
Engineering & AI/ML at Scale: Subramanya (Subu) Gupta has decades of experience leading large engineering teams and building sophisticated AI/ML and cloud platforms, giving the company the ability to translate complex science into robust, scalable products.
Clinical Insight & Relevance: Dr. Swami (Tata IISc Medical School), Dr. Sivareddy (MD Anderson) provides frontline oncology and medical research expertise, ensuring the technology addresses real-world clinical needs and fits seamlessly into healthcare workflows.
Entrepreneurial & Strategic Leadership: Ravi Puvvala, a seasoned entrepreneur with a track record of scaling deep-tech companies, drives strategic vision, partnerships, and regulatory navigation.
This unique fusion of frontier science, world-class engineering, clinical understanding, and entrepreneurial execution is what allows Quantum Biosciences to tackle a problem as complex as quantum-enabled diagnostics and to move from concept to clinical impact faster than most deep-tech startups. |
Quantum Biosciences |
quantumbiosciences.ai |
|
Quantum-sensing healthcare diagnostics that transform patient care from reactive damage control to proactive, predictive prevention |
Cancer patients face overwhelming treatment options like chemotherapy and radiation. Clinicians often react after damage occurs. For instance, radiation-induced heart damage is detected months later by MRIs. Acute kidney injury shows signs only 24–48 hours after damage starts. These delays cost patients valuable time when early intervention could help. |
Our Solution
Quantum Biosciences has built a first-of-its-kind quantum sensing platform that gives clinicians an early, real-time view of cellular injury—hours after radiation exposure, rather than the weeks or months required by today’s diagnostics. The platform combines three tightly integrated layers:
Quantum Nanodiamond Sensor – Uses nitrogen-vacancy (NV) centers in nanodiamonds to detect the minute “magnetic noise fingerprints” created by bursts of Reactive Oxygen Species (ROS)—the very first molecular signals of tissue stress.
In-Silico Simulation Engine – A digital “twin” that predicts the unique magnetic fingerprints of different ROS molecules, creating a reference library that drives accurate identification.
AI/ML Deconvolution Engine – Advanced machine-learning models that filter out background noise and translate raw sensor data into clinically meaningful signatures, identifying whether the stress is mitochondrial, inflammatory, or ischemic and predicting if the damage is reversible.
Experimental Validation Platform – A real-world testbed where these predictions are refined and validated in live cell models and, eventually, clinical samples.
What this delivers
Predictive early warning – Detects molecular injury within hours, long before standard biomarkers or imaging show damage.
Actionable insight – Distinguishes the type and severity of stress so clinicians can adjust therapy or initiate protective interventions.
Personalized medicine – Identifies patients at higher risk (“hyper-responders”) and guides tailored treatment plans.
This quantum-AI diagnostic system moves radiation oncology—and eventually other fields of medicine—from reactive detection to proactive prevention, creating a powerful new tool for understanding and controlling disease at its molecular origin. |
Our solution is unique and defensible because it combines breakthroughs in quantum sensing, advanced computation, and clinical translation in a way that is very difficult to replicate. 1. First-in-Class Quantum Sensing for Early Disease Biology We are the only platform using NV-center nanodiamonds to capture the earliest molecular fingerprints of cellular stress—within hours of radiation exposure. Conventional diagnostics (MRI, PET, blood biomarkers) detect damage only after it has become irreversible, while even other quantum-sensing projects focus on general redox states rather than predictive, clinically actionable biomarkers. 2. Proprietary In-Silico Simulation Engine Our custom quantum chemistry + molecular dynamics models predict the precise magnetic signatures of different Reactive Oxygen Species (ROS). This predictive “library” of fingerprints is our core intellectual property, giving our AI/ML models a unique reference set that competitors would find extremely difficult to recreate. 3. AI/ML Deconvolution Engine We have built a sophisticated AI pipeline that converts noisy NV-sensor signals into mechanistic insights—distinguishing whether the stress is mitochondrial, inflammatory, or ischemic and predicting whether damage is reversible. This transforms raw data into clinically interpretable diagnostics, not just measurements. 4. Rare Cross-Domain Expertise Replicating our approach requires simultaneous, world-class expertise in quantum physics, optical engineering, computational chemistry, AI/ML, and clinical oncology. This depth of interdisciplinary capability forms a natural barrier to entry and a sustainable competitive moat. 5. IP and Early-Mover Advantage We are actively filing global patents covering the in-silico classification methods, the AI/ML diagnostic pipeline, and the clinical workflow of our “Radiotoxicity Risk Score.” This, combined with our head start in both simulation and prototype development, secures a defensible intellectual property position. In short: our solution isn’t just another diagnostic test; it is a new category of predictive quantum diagnostics. The fusion of quantum physics and AI to produce actionable, mechanistic insights—backed by proprietary algorithms and protected IP—makes it both unique in the market and highly defensible against competitors. |
Idea |
Testimonials |
Hospitals, Cancer Centers, Diagnostics Labs, Pharma Companies, Contract Drug Manufacturing Organisations |
28B USD |
7B USD |
1B USD |
Diagnostics-as-a-Service (DaaS), Platform-as-a-Service (PaS) |
QT Sense (Netherlands): Uses nanodiamond quantum sensors to detect ROS inside cells. Focuses on general oxidative stress, not predictive clinical diagnostics. Artephi Computing / Varian / RaySearch: Apply quantum computing to optimize radiation therapy planning, not molecular diagnostics. Singularity Quantum: Uses quantum machine learning for risk prediction, but no quantum sensing. Edge: Quantum Biosciences uniquely combines NV-center quantum sensing with an AI-driven in-silico engine to deliver predictive, mechanistic diagnostics—a capability none of these players currently offer. |
Clinical Partnerships & Pilots: Start with flagship cancer hospitals and CROs by offering pilot studies of the Radiotoxicity Risk Score (RTRS) test. Early clinical validation builds trust and generates reference sites. Key Opinion Leaders (KOLs): Engage leading radiation oncologists and cardiologists as early adopters and advisors to champion the technology in conferences and peer-reviewed publications. Pharma & Biotech Collaborations: Partner with pharmaceutical companies developing radioprotective or oncology drugs, positioning the platform as a companion diagnostic in clinical trials. Government & Grant Programs: Leverage national quantum-tech and healthcare initiatives for funding and visibility, creating strong institutional endorsements. Scientific Marketing & Thought Leadership: Publish high-impact clinical results, present at global conferences, and build credibility through peer-reviewed evidence to drive inbound demand. This evidence-driven, partnership-first approach ensures early adoption by top-tier institutions and creates a strong reputation that naturally attracts additional hospitals and industry partners. |
Start with radiation oncology: Launch the Radiotoxicity Risk Score (RTRS) test as a lab-developed test with leading cancer centers. Clinical validation & partnerships: Pilot with top hospitals and CROs to generate peer-reviewed data and build credibility. Regulatory path: Begin with CLIA labs in the U.S. and Indian approvals, then pursue FDA/CE certification. Business model: Offer diagnostics and analytics as a subscription-based Diagnostics-as-a-Service platform; partner with pharma for radioprotective drug trials. Scale: Expand to cardiotoxicity, neurodegeneration, and inflammatory disorders, leveraging government quantum-tech programs for funding and policy support. |
To make quantum-enabled diagnostics a standard of care—transforming medicine from reactive treatment to proactive, precision health management.
We aim to:
Establish a global platform for early disease detection, starting with radiation-induced toxicity and expanding to cardiology, neurology, and chronic inflammatory diseases.
Build the world’s first continuously learning quantum–AI diagnostic network, capable of decoding the earliest molecular signals of any disease.
Partner with top research institutions to create a foundational understanding of disease mechanisms, shaping future therapies and drug discovery.
Position India as a world leader in quantum biosensing, anchoring the country’s National Quantum Mission in life sciences.
Ultimately, Quantum Biosciences envisions a future where disease is predicted and prevented at the molecular level, giving clinicians the power to intervene before irreversible damage occurs. |
Incorporated in Karnataka in 2024. Recognized by DPIIT, Karnataka as startups |
20 Lakhs |
Yes |
Access to IIT network to hospitals and access to data |
Get established to create the world's largest and most valuable dataset of quantum-level biological signals. |
Yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/R2wUXKbXIgob-QBS-Pitch-Deck-2025.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/R2wUXKbXIgob-QBS-Pitch-Deck-2025.pptx |
https://economictimes.indiatimes.com/tech/technology/bengalurus-deep-tech-startup-proposes-quantum-leap-in-cancer-care/articleshow/123071562.cms?from=mdr |
Yes – Quantum Biosciences is a mission-driven, impact-focused startup.
Our mission is to shift healthcare from reactive treatment to proactive, precision prevention by detecting the very first molecular signals of disease. We are building a platform that:
Predicts and prevents harm: Enables clinicians to identify radiation-induced tissue injury within hours, giving them a chance to intervene before irreversible damage occurs.
Transforms fundamental research: Opens an unprecedented window into how diseases begin at the sub-cellular level, accelerating discoveries in oncology, cardiology, neurology, and beyond.
Democratizes advanced medicine: Positions India as a global leader in quantum biosensing and ensures that cutting-edge diagnostics are accessible to populations with the highest disease burden.
By uniting quantum physics, AI/ML, and clinical medicine, we aim to create lasting social impact: earlier detection, fewer long-term complications, and a deeper scientific understanding of disease that benefits patients worldwide. |
NA |
Abhay Nadkarni, IIT Bombay Alumni |
checked |
| Sep 16, 2025 @ 4:18 PM |
abc |
abc@abc.com |
https://linkedin.com |
https://www.google.com |
9999888822 |
abc |
abc acbc |
Yes |
ascsas |
abc |
NA |
|
asn |
an |
askps sls |
aslma |
MVP |
Revenue |
asa |
NA |
NA |
NA |
NA |
NA |
NA |
NA |
asdsds |
NA |
A |
Yes |
zsas |
asas |
asasa |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/owu4pSJNGkD2-Dummy-PPT.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/owu4pSJNGkD2-Dummy-PPT.pptx |
https://www.youtube.com/watch?v=ZsmrwMTFEUk |
asa |
NA |
NA |
checked |
| Sep 15, 2025 @ 9:19 PM |
Keshava Kumar S; Nagesh G; Prof. Erasmo Carrera ; Dr. M. R. Bhat |
keshav@iitbombay.org |
https://www.linkedin.com/in/keshava-kumar-s-rao |
https://alambya-labs.in |
9980072592 |
1) Keshava Kumar S : Founder Director - BE Civil (BU), M.Tech Aero (IITB), PhD Aero (IISc) ; 2) Nagesh G: Technical Director : BE Mech (BU), MTech Mech (NITK) ; 3) Prof. Erasmo Carrera : Mentor - Professor at Politecnico di Torino; 4) Dr. M. R. Bhat - Mentor - CRS at IISc |
Keshava and Nagesh first met while working at ISRO in 2006, where they collaborated on the design and testing of star sensors for spacecrafts. They have known each other since then and share a long-standing professional relationship.
In addition to the directors listed in our team, we are guided by two mentors. Prof. Erasmo Carrera, with whom Keshava pursued part of his PhD research as an exchange student at Politecnico di Torino, Italy (since 2013), and Dr. M. R. Bhatt, with whom Keshava worked during his postdoctoral tenure at IISc (since 2008). Both mentors continue to provide valuable technical and research guidance to our work. |
No |
The biggest strength of our team is the unique combination of deep technical expertise, long-standing collaboration, and strong mentorship support. Keshava and Nagesh have known each other since 2006, having worked together at ISRO on spacecraft sensor design and testing, which has built a foundation of trust and seamless teamwork. This is further strengthened by guidance from eminent mentors such as Prof. Erasmo Carrera and Dr. M. R. Bhatt, who bring international research exposure and advanced domain expertise. Together, this blend of experience, trust, and mentorship enables us to address complex challenges with both innovation and reliability. |
Alambya Labs Private Limited |
https://alambya-labs.in |
|
Delivering sustainable mobility through advanced composite design and simulation tools |
There is a lack of high-fidelity, end-to-end hybrid composite simulation tools, as well as limited expertise in hybrid composite design and manufacturing in India. |
Through NIDHI PRAYAS, we are developing a modular UAV hybrid composite prototype that directly addresses these gaps. The prototype not only demonstrates a market-relevant mobility application but also generates critical in-house manufacturing data. This data serves as the foundation for developing AI-driven digital twin tools that integrate process, mechanical, and end-of-life simulations for composites. |
Our solution is unique because it generates proprietary hybrid composite manufacturing data through a UAV prototype, which is then used to build high-fidelity models. This data powers our AI-driven digital twin simulation tool, the first to integrate process, mechanical, and end-of-life simulations for composites, unlike existing metal-centric software |
Idea |
Pilots |
Hardware (UAV): UAV manufacturers, research labs, and educational institutions seeking lightweight hybrid composite UAV platforms. Software (Simulation Tool): Aerospace and automotive OEMs, startups, and research labs designing advanced composite components using AI-driven digital twin tools. |
USD 1.58B |
USD 500M |
USD 10M |
Hardware: B2B sales of hybrid composite UAV airframes to research labs, educational institutions, and early adopter UAV manufacturers. Software: AI-driven digital twin simulation tool offered as a SaaS for aerospace, automotive, and mobility OEMs, SMEs, and research institutions. |
There are no direct competitors for our products in focus. While legacy UAVs and simulation tools exist, none specifically address hybrid composite UAV airframes or provide an AI-driven digital twin simulation tool integrating process, mechanical, and end-of-life aspects for composites. |
We acquire hardware customers through direct B2B engagement with research labs, educational institutions, and early adopter UAV manufacturers. For our AI-driven simulation tool, we use direct outreach, technical demos, and pilot projects with aerospace and automotive OEMs, startups, and research institutions. |
We launch the hybrid composite UAV through B2B pilots with research labs, educational institutions, and early adopter manufacturers. Insights and data from these pilots feed our AI-driven digital twin simulation tool, offered as a SaaS to OEMs, startups, and research institutions. |
To become a global leader in hybrid composite design and AI-driven simulation tools, enabling sustainable, high-performance mobility solutions across aerospace, automotive, and industrial applications. |
Alambya Labs Pvt Ltd — incorporated and registered in India, DOI 16 Apr 2025 |
950000Rs through Nidhiprayas |
Yes |
We are applying to Avinya, an AI-focused accelerator, to gain mentorship, entrepreneurial training, and industry connections to refine our AI-driven digital twin simulation tool and validate its technical models and business potential. |
We aim to refine our simulation tool, validate technical performance and business model, strengthen entrepreneurial skills, and prepare for early pilots and investment. |
Yes we are open to both, would prefer hybrid for sure. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/OIanEZ6ZITgx-PitchDeck_AlambyaLabsPvtLtd.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/OIanEZ6ZITgx-PitchDeck_AlambyaLabsPvtLtd.pptx |
https://NA |
Alambya Labs is a mission-driven startup focused on advancing hybrid composite design and AI-driven simulation tools to enable sustainable, high-performance mobility solutions. By developing AI-powered digital twins and proprietary manufacturing data, we aim to reduce material waste, accelerate innovation in aerospace and automotive sectors, and make advanced composites accessible in India. Our impact lies in combining technology leadership with sustainability and scalability. |
NA |
IIT B alumni group |
checked |
| Sep 15, 2025 @ 2:18 PM |
Raghav Somani |
raghav@kalonedu.com |
https://www.linkedin.com/in/raghav-somani-b2b90b2b |
https://www.kalonedu.com |
7038567000 |
Raghav Somani - Operations, Business Development, Partnerships & Collaboration - Pooja Somani - Marketing & Finance |
We are couple and are excellent in their own particular fields. |
Yes |
The biggest strengths to work as a team are as follows:
1) Problem Solving - each problem we come across is tackled differently considering the diverse mindset of the team.
2) Scalability - working as a team, has better chances of scalability.
3) Innovation - each person works towards a common goal through different means and unique mindset. |
Kalonedu Expert Pvt Ltd |
www.kalonedu.com |
|
We offer 360 degree post landing support to students who wish to pursure their higher education abroad. |
The students who go abroad for their higher education are not career ready and are completely lost and confused. They need someone to support, guide and mentor them in their initial phase of transition. There is also lack of transparent accommodation service provider. |
We offer a wide array of post landing services to students which makes them career ready and further sailing smooth. Also, offer very transparent accommodation service. |
We are a first mover advantage and the services that are offered are very meticulously crafted and designed. There's no one in the India whose offering all services under one roof. |
Revenue |
Revenue |
Students who are pursuing their higher education abroad. |
7,59,064 |
540000 |
225000 |
Service package |
None |
Through digital marketing and study abroad consultants |
I have a wide network of study abroad partners spread across south asia. |
To be the most reliable partner for the post landing services for students and have our presence for study abroad destinations. |
Private Limited |
None |
Yes |
We are specifically applying and keen on partnering with Avinya for the following reasons: 1) It's very focused and driven in the field of AI. 2) Apart from funding, it closely works closely with founders and offer structured mentorship and gives access to global network. 3) Offers extensive support with it's strong IIT Alumni & KA Government. |
Seek structured mentoring support, access to global market and networking opportunities. |
Yes |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/EJmxpZCUMHyh-MASTERDECK-KalonEdu-Expert-Pvt-Ltd.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/EJmxpZCUMHyh-MASTERDECK-KalonEdu-Expert-Pvt-Ltd.pdf |
https://www.kalonedu.com/package.html |
We offer a wide array of post landing services to students which makes them career ready and further sailing smooth. Also, offer very transparent accommodation service. Currently, we are operational in the UK, Germany & Singapore and have got my broader team present at these locations. We also support the students for part-time, full-time and internship opportunities in the UK & Germany. |
NA |
NA |
checked |
| Sep 11, 2025 @ 6:46 PM |
Sankalp Jain |
sankalpjain@scanimart.com |
https://www.linkedin.com/in/sankalp-jain-ironman |
https://scanimart.com/ |
7903624595 |
CEO/CTO, I hold a B.Sc. in Physics and an M.Sc. in Applied Physics from Amity University. With a strong technical background and self-taught expertise in software development, I lead product development, strategy, and technology implementation at Scanimart. I am passionate about building scalable AI-driven retail solutions that reshape offline shopping experiences. |
I have known Pradeep for about 2 years. We shared a place during that time and have been collaborating on different startup ideas, which helped us build a strong mutual understanding and a good working relationship. |
Yes |
Our biggest strength as a team is our complementary skill set and shared vision. While one of us brings deep technical expertise in building scalable AI-driven solutions, the other brings strong marketing, customer engagement, and business development skills. This balance of technology and market understanding, combined with our trust and past collaboration, enables us to move quickly, adapt efficiently, and execute effectively. |
Scanimart |
https://scanimart.com/ |
|
Scanimart is a universal offline shopping app that lets customers scan, pay, and go seamlessly across any mart. |
We are solving the problem of long checkout queues, human errors, and bulky hardware dependency in offline retail. Marts face inefficiencies in operations and data management, while customers experience delays and frustration at billing counters. Scanimart eliminates these pain points by offering a universal, hardware-light, AI-powered solution that enables seamless scan, pay, and go shopping. |
Scanimart provides a universal offline shopping app that enables customers to simply scan, pay, and go. Our solution eliminates billing queues through AI-powered item recognition, integrates digital payments and invoicing, and issues secure QR-based gate passes for seamless exit. For marts, we offer a lightweight, cost-efficient platform with inventory insights, data-driven analytics, and advertising opportunities—without heavy hardware dependency. |
Scanimart is unique because it is a universal offline shopping app, not a store-specific solution. Unlike existing POS or self-checkout systems, it is hardware-light, eco-friendly, and scalable across different marts with minimal setup. Our AI-powered product recognition, QR-based gatepass, and integrated ecosystem for customers, marts, and FMCG brands make it defensible. The platform’s value grows with every new mart onboarded, creating strong network effects and high switching costs for retailers. |
Users |
Users, Testimonials |
Our target customers are marts and retailers (B2B) who want to reduce checkout time, cut hardware costs, and gain data-driven insights, and everyday shoppers (B2C) who want a seamless, universal offline shopping experience with scan, pay, and go convenience. |
$3.6 Trillion |
$219 Billion |
$10 Billion |
Our revenue model is primarily transaction-based, charging marts a 1–2% commission on every transaction made through Scanimart. Additional revenue streams include in-app advertisements, inventory management tools, FMCG brand partnerships, premium analytics features, store-specific customizations, and training/support services.tions, and training/support services. |
Traditional POS providers like GoFrugal, e-commerce platforms exploring offline integrations like BigBasket or Reliance Smart, and small Indian self-checkout startups. Unlike them, Scanimart offers a universal AI-driven offline shopping platform across multiple marts. |
We onboard marts through direct outreach, demos, and FMCG partnerships, and attract shoppers via our app, in-store QR codes, and referral programs highlighting faster, seamless checkout. |
We start by piloting Scanimart in key marts to showcase faster checkout and operational efficiency. We expand through direct B2B outreach, FMCG partnerships, and marketing campaigns targeting shoppers via our app, in-store promotions, and referral programs. Early success stories and data-driven insights will help scale adoption nationwide. |
To become the universal offline shopping platform in India, seamlessly connecting marts and shoppers through AI-driven automation. We aim to revolutionize retail by making checkout faster, reducing fraud, optimizing inventory, and creating a fully digital, efficient, and scalable shopping ecosystem nationwide. |
Incorporated |
NA |
Yes |
We are applying to Avinya to gain strategic mentorship, access to industry networks, and funding support that can accelerate Scanimart’s growth. The program’s focus on innovation and scaling disruptive solutions aligns perfectly with our vision to transform offline retail in India. |
During Avinya, we aim to refine Scanimart’s product-market fit, accelerate mart and shopper adoption, strengthen our go-to-market strategy, and secure strategic mentorship and partnerships to scale our AI-driven retail platform nationwide. |
es, we are open to both hybrid and virtual participation to fully leverage Avinya’s mentorship, networking, and resources while ensuring seamless collaboration. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/CPDJF0CxrCkP-Scanimart-Short-Deck.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/CPDJF0CxrCkP-Scanimart-Short-Deck.pdf |
https://youtu.be/7mwqUmhsNEE?si=BUfYIysyMIdW9eQd |
Yes, Scanimart is both mission-driven and impact-focused. Our mission is to revolutionize offline retail by making shopping faster, more efficient, and error-free through AI-driven automation. Beyond efficiency, we aim to reduce operational costs, minimize fraud, lower environmental impact through digital receipts, and empower small and medium marts to compete in the digital era—creating positive social and economic impact nationwide. |
NA |
NA |
checked |
| Sep 3, 2025 @ 10:36 AM |
KUNAAL SHIRISH SHINDAGI |
kunaalshindagi@gmail.com |
https://www.linkedin.com/in/kunaal-shindagi-759bb625b/ |
https://www.cipherion.in/ |
8105188413 |
Kunaal Shirish Shindagi (Founder & CEO) – Managing Director of Cipherion Pvt Ltd, with expertise in cybersecurity, encryption systems, and business strategy. He has led multiple research papers and projects in quantum-resilient encryption, SOC automation, and data security. Background in Computer Science and extensive experience buildi ng and scaling technical products and teams. Pooja Rajendra Ekbote (Co-Founder & COO) – Responsible for operations, partnerships, and compliance. She ensures smooth coordination between technical and business functions. With a strong background in management and organizational planning, she bridges execution between R&D and customer success. Nagaraj Krishna Bhandare (Co-Founder & CTO) – Technical lead specializing in backend systems, microservices architecture, and product R&D. His expertise lies in building scalable security infrastructure, developing core encryption modules, and leading engineering teams to deliver high-impact solutions. |
The founding team met during their academic and professional journey in Karnataka and discovered a shared vision of solving the pressing global challenge of data security. Over the past 4 years, they have collaborated on multiple projects, research papers, and startup initiatives, and as a team they have also won prestigious competitions such as the Smart India Hackathon and the IIT Roorkee Hackathon. These experiences naturally evolved into building Cipherion. Their complementary skills in technology, operations, and business strategy have strengthened trust and synergy as a founding team. |
Yes |
Our biggest strength is the ability to combine deep technical expertise with market-driven execution. We bring together strong backgrounds in cybersecurity, operations, and business strategy, which allows us to not only design cutting-edge solutions but also align them with regulatory requirements and customer needs. Over the past 4 years, we’ve proven our execution capabilities by collaborating on multiple projects, publishing research, and winning national-level hackathons. This track record demonstrates that we can move from concept to validated product quickly, while maintaining the discipline needed to scale sustainably. As a team, we excel at building solutions that are both technologically advanced and commercially viable—a balance that positions Cipherion for long-term growth. |
Cipherion |
https://www.cipherion.in/ |
|
Cipherion is redefining data security with a scalable Encryption-as-a-Service model that reduces compliance costs and accelerates enterprise adoption. |
As enterprises adopt AI and LLMs at scale, sensitive data such as healthcare records, financial details, and personal identifiers often flow directly into these models without adequate protection. Traditional encryption is not designed for AI workflows—once data is decrypted for model training or inference, it becomes exposed. This creates massive risks of data leakage, regulatory non-compliance, and misuse of private information. Companies face growing challenges in balancing the need to leverage AI with the obligation to safeguard sensitive data under strict regulations like HIPAA, GDPR, PCI-DSS, and DPDP. Cipherion solves this by providing AI-aware data masking, ensuring that sensitive information is anonymized and protected before it reaches AI/LLM systems, enabling organizations to use AI securely while staying compliant and reducing breach risks. |
Cipherion provides a quantum-resilient, keyless encryption framework with AI-aware data masking designed specifically for modern enterprises. Instead of relying on vulnerable keys, Cipherion ensures end-to-end data security while eliminating the most exploited weakness in encryption.
Our AI-aware data masking layer protects sensitive information—such as healthcare records, financial details, or personal identifiers—before it enters AI/LLM workflows. This allows organizations to harness the power of AI without exposing regulated data, reducing compliance risks under HIPAA, GDPR, PCI-DSS, and DPDP.
The solution integrates seamlessly into existing systems, automates compliance enforcement, and scales across cloud, edge, and enterprise environments, making secure AI adoption both simple and cost-effective. |
Cipherion is unique because it directly addresses a critical blind spot in AI adoption—protecting sensitive data before it enters AI/LLM workflows. While most security solutions focus on encrypting data at rest or in transit, Cipherion provides AI-aware data masking that ensures compliance and privacy inside the model pipeline itself. Our defensibility comes from three core pillars: Keyless Encryption Framework – Removes the most common attack surface in cryptography (keys), making our approach inherently more secure and quantum-resilient. AI-Native Security Layer – Unlike generic encryption vendors, Cipherion is purpose-built to secure sensitive information within AI and ML pipelines, an emerging and under-served market. Regulatory Advantage – By embedding automated compliance checks (HIPAA, GDPR, PCI-DSS, DPDP), we reduce enterprise audit costs and risks, creating a strong moat against competitors. This combination of future-proof encryption, AI-specific data protection, and compliance automation makes Cipherion both hard to replicate and strategically defensible. |
MVP |
Pilots |
Our primary customers are enterprises and organizations that handle sensitive data and are adopting AI/LLM solutions at scale. This includes: Healthcare providers & healthtech companies – needing HIPAA-compliant AI solutions for patient data. Fintech & banks – securing financial records and transactions while using AI for fraud detection, credit scoring, and automation. Manufacturing & Industry 4.0 companies – deploying AI-driven analytics on operational and supply chain data. AI/ML-driven startups and enterprises – integrating LLMs into their products while needing to ensure data privacy and compliance. These sectors face the dual challenge of leveraging AI innovation while remaining compliant with strict regulations, making them the most immediate beneficiaries of Cipherion’s AI-aware data masking and keyless encryption. |
$500 billion by 2030 |
$50–70 billion by 2030 |
$80 million in next 5 years |
Recurring subscriptions for AI-aware data masking and compliance automation. Custom enterprise deployments with integration and support fees. Compliance-driven add-ons (HIPAA, GDPR, PCI-DSS, DPDP modules). |
Fortanix – Offers encryption & tokenization, but still key-dependent. Skyflow – Focused on privacy vaults, not AI-native. Privitar – Strong compliance focus but limited automation. Cipherion differentiates itself with quantum-resilient, keyless encryption and AI-aware data masking, addressing a blind spot none of these players currently cover. |
Partnerships with cloud providers, compliance consultants, and incubators. Pilot programs with healthcare, fintech, and manufacturing enterprises. Showcasing thought leadership via research papers, conferences, and webinars. Leveraging our recognition from IIIT Dharwad incubation and HDB Blue Top 10 selection as credibility boosters. |
Focus industries – Healthcare, fintech, and manufacturing, where compliance costs are high and security spend is non-negotiable. Pilot → Scale – Start with pilot integrations, then expand across enterprise workflows. Global expansion – Begin in India and the EU, then extend to the US market where AI regulation is intensifying. Partnership ecosystem – Collaborate with cloud providers, system integrators, and compliance advisory firms to accelerate adoption. This staged approach ensures faster traction, strong compliance positioning, and scalable growth. |
Our long-term vision is to make Cipherion the global standard for AI-native data security—a platform where organizations can adopt artificial intelligence without ever compromising on privacy, compliance, or trust. We aim to eliminate key-based vulnerabilities, establish AI-aware data masking as a default layer in every enterprise workflow, and build a quantum-resilient security backbone that powers the next generation of digital infrastructure.
In the next decade, we envision Cipherion becoming the “encryption layer of the quantum era”, integrated into every sector healthcare, finance, manufacturing, government, and beyond—so that sensitive data remains secure by design, not by afterthought. |
Cipherion is incorporated as K3NP Cipherion Private Limited in India (registered in Karnataka). |
We are currently bootstrapped and have not raised external funding yet. However, we have secured incubation support from IIIT Dharwad Research Park and recognition as a Top 10 startup at HDB Blue 2025, which provides us with ecosystem access, mentorship, and growth opportunities. |
Yes |
We are applying to Avinya to gain access to a supportive ecosystem of mentors, investors, and industry experts who can help us refine Cipherion’s product, strategy, and market positioning. Avinya’s focus on nurturing deep-tech and high-impact startups aligns with our mission to build a quantum-resilient, AI-native data security platform. We believe the programme will provide the right network, guidance, and visibility to accelerate our growth. |
During the programme, we want to: Validate and refine our AI-aware data masking solution with enterprise pilot customers. Build strong connections with investors and corporate partners for our upcoming seed round. Strengthen our go-to-market strategy across healthcare, fintech, and manufacturing sectors. Get expert mentorship on scaling B2B sales while preparing for B2C expansion with Cipherion Vault. Our goal is to emerge from the programme with market validation, strategic partnerships, and investor readiness. |
Yes – we are open to both hybrid and virtual participation, as our team operates flexibly across geographies and is fully equipped to collaborate online while also participating in in-person sessions when required. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Fu7PTSZFAOvK-Cipherion-HBD-pitch-1_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/Fu7PTSZFAOvK-Cipherion-HBD-pitch-1_compressed.pdf |
https://youtu.be/34lrZZhY2tg?si=ODPNof-Gm7bZmAk8%20 |
Yes — Cipherion is a mission-driven startup with a clear focus on impact. Our mission is to make data security simple, universal, and future-proof so that organizations and individuals can adopt AI and digital technologies without fear of breaches, misuse, or regulatory penalties.
The impact we are creating is twofold:
Enterprise impact – By enabling AI-aware data masking and keyless encryption, we help healthcare providers, financial institutions, and manufacturers protect sensitive data, reduce compliance costs, and prevent breaches that can affect millions of lives.
Individual impact – With our upcoming Cipherion Vault, we empower everyday users to protect their files, media, and personal data, ensuring that privacy is not just for large enterprises but a basic right for everyone.
At its core, Cipherion is built to remove the biggest barrier to trust in the digital era—the insecurity of data. We see ourselves not just as a company, but as a movement towards safer AI and a more privacy-respecting digital world. |
Yes. Our founding team comes from Tier-2/Tier-3 cities in Karnataka, India, where access to capital, mentorship, and startup ecosystems is far more limited than in major hubs. Rather than seeing this as a limitation, it has shaped our resilience, resourcefulness, and ability to build solutions from the ground up. We are also proud to have gender diversity at the leadership level, with a woman co-founder (COO) actively driving our operations and strategy. This combination of regional diversity, gender inclusion, and deep technical expertise allows us to approach problems differently and design solutions that are globally competitive yet rooted in real-world challenges faced by underserved markets. We believe this underrepresented background is not just part of our identity—it is our advantage, giving us unique insights into building scalable, impactful technologies that bridge gaps between emerging and developed markets. |
Dr. Deepak K T https://www.linkedin.com/in/deepak-k-t-0788212a3/ ,Dr. Girish Revadigar https://www.linkedin.com/in/girish-revadigar-phd-55a33b28/ |
checked |
| Sep 2, 2025 @ 3:54 PM |
Arahat Dudhamal and Aditya Nair |
arahat@righttoaudit.ai |
https://www.linkedin.com/in/arahat-dudhamal/ |
https://righttoaudit.ai/ |
8169895305 |
Aditya Nair and I are the co-founders of RightToAudit.ai. Aditya is the CTO, and I'm the CEO, with both of us being technical founders and holding 50% equity each. As CTO, Aditya is a technical founder with a B.Tech in Computer Science from MIT ADT University. Before starting this company, he worked as a Software Engineer at Zania and a Software Developer at Wyzr.in. He also has experience as a Back End Developer at dHealth. I hold a B.Tech from Veermata Jijabai Technological Institute (VJTI). I previously worked as a Technology Manager at ICICI Bank Limited. |
Aditya and I have known each other for over a decade, ever since we met in school. Our mutual love for technology brought us together. We've both had experiences that led us to this idea—I worked on compliance tooling at a large bank, while Aditya focused on AI-driven audit products at a software company. Both Aditya and I've been working on the project full-time from day one. |
Yes |
Our biggest strength as a team is our complementary experience and shared understanding of the problem. I saw the inefficiencies of manual vendor audits firsthand while developing compliance-reporting tools at ICICI Bank. At the same time, Aditya witnessed teams struggling with spreadsheets and other inefficiencies while creating AI-driven audit solutions at Zania.ai. We've both experienced the pain points of expensive, error-ridden audits and know there's a strong demand for an AI-powered, end-to-end solution in the market. |
Right To Audit AI |
https://righttoaudit.ai/ |
|
We are an AI platform that automates vendor compliance audits. |
At my previous job at ICICI Bank, I spent two years building compliance-reporting tools and saw firsthand how much time and resources were wasted on manual vendor audits. My co-founder, Aditya, also saw a similar issue at his previous company, Zania.ai, where he watched teams drown in spreadsheets. Banks and fintechs struggle with expensive, error-ridden audit processes. The current solutions are manual, static, or siloed, which creates inefficiencies and makes it hard for organizations to be audit-ready at all times. |
RightToAudit.ai is an AI platform that streamlines the entire third-party compliance audit process for regulated sectors. Our platform handles everything from reaching out to vendors to gathering documents, sending follow-up requests, and assessing risks. We use a custom Gemini AI model to perform dynamic risk analysis, giving users confidence scores, risk levels, and actionable insights. It also has an intelligent email processing system for automated communication. By using AI, we save time and money while minimizing errors in what is usually a manual process. |
Unlike our competitors, such as ServiceNow and OneTrust, we automate the entire vendor audit lifecycle from outreach to reporting. We use adaptive AI and live external risk data, which sets us apart from other solutions that are often manual, static, or siloed. We believe no one else is currently developing a comprehensive, end-to-end AI-powered solution like ours. Our goal is to make organizations continuously audit-ready. |
MVP |
Users |
Our target customers are financial institutions and fintech companies that deal with a large number of vendors and require an efficient way to manage their compliance audits. We are also targeting regulated sectors like healthcare and pharma that must adhere to strict audit standards. |
$7.92B in 2025, source: Cognitive Market Research |
Indian audit software market ($88.41M in 2025, source: Markets and Data) |
SOM for the first year is a 1% share of the SAM, which equals $884,100 |
We will be using a SaaS-based and an Enterprise-based pricing model. |
Our main competitors are ServiceNow and OneTrust. However, we believe we have an edge over them as our platform automates the entire vendor audit life cycle and is built on adaptive AI, something they don't have. |
We plan to acquire customers through a multi-channel approach: 1. Direct Sales: We will have a direct sales team that reaches out to our target customers in the financial, banking, healthcare, and pharma sectors. 2. Content Marketing: We will create thought leadership content, such as blog posts and white papers, to highlight the inefficiencies of current audit processes and showcase how our AI-powered solution can solve these problems. 3. Partnerships: We will partner with consulting firms and audit companies to introduce our solution to their clients. |
NA |
Our long-term vision is to establish our platform as the go-to solution for all compliance and audit-related activities. We want to be the "source of truth" for all compliance and risk, giving our customers the peace of mind that comes with being continuously audit-ready. |
NA |
NA |
Yes |
We are applying to Avinya because we are a team with a functional MVP and are at the critical stage of moving from product development to market validation and scaling our business. We believe Avinya's structured program, industry connections, and specialized focus on AI startups in fintech are the resources we need to navigate the next phase of growth. We are looking for guidance on refining our go-to-market strategy and securing our first pilot customers in the highly regulated financial sector. |
During the program, our primary goals are to leverage our MVP to secure our first 3-5 pilot customers in the financial and banking sectors to validate our product-market fit. We intend to gather critical user feedback to rapidly iterate on our platform and develop key features that our target customers are willing to pay for. Additionally, we want to work with Avinya's mentors to refine our business model and develop a clear, compelling narrative for future fundraising, positioning the company for a successful seed funding round upon graduation. |
Yes, we are fully open to either a hybrid or virtual format for participation. Our team is committed to dedicating ourselves exclusively to RightToAudit.ai and will be able to fully meet all the program's requirements regardless of the format. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/g993lpXa59Z0-Avinya_Pitch-Deck-RightToAudit.ai_.pptx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/g993lpXa59Z0-Avinya_Pitch-Deck-RightToAudit.ai_.pptx |
https://youtu.be/BsEGgtpvUCQ?feature=shared |
Our mission is to empower Indian businesses to move from reactive, fear-based compliance to a proactive and efficient state. By automating tedious audit processes, we help companies save significant time and resources, which they can then reinvest into innovation. Ultimately, we aim to build a more trustworthy and efficient business ecosystem. |
NA |
Sweta Jhunjhunwala |
checked |
| Aug 28, 2025 @ 3:42 PM |
Parikshit Ramesh Lembhe |
lembheparikshit853@gmail.com |
https://www.linkedin.com/in/parikshit-lembhe-8262291ab?lipi=urn%3Ali%3Apage%3Ad_flagship3_profile_view_base_contact_details%3BnEF0JX%2FbRma99yhoK43phw%3D%3D |
https://NA |
9821356568 |
CEO Pursuing IT engineering at Padmabhushan vasantdada patil college of engineering |
i am working since 2022 building prototype and now it is ready so i am looking for seed funding now to build or develop real world prototype |
Yes |
our team solves complex and critical problems to acheive level 5 autonomy in autonomous vehicles.team members have full knowledge of autonomy |
Pralaxya Pvt Ltd |
NA |
|
our startup uses teledrive technology to solve level 5 autonomy develop autonomous vehicles and to solve logistic and defence problems related to autonomous to d |
i am solving problems regarding accidents occurs in civil,logistic and defence sector.to generate revenue in this areas like to reduce the expense in logistic economy and to provide petroling on borders for militery |
SensorPod is a ₹120,000/- magnetic retrofit kit that transforms any old Indian car (2015+ models) into a safety-equipped vehicle. It combines:
AI Auto-Braking: Stops cars 0.5 seconds faster than humans (proven: 68% fewer crashes in Ola pilot).
Remote Human Helpers: Jump in during monsoons/festivals via 5G – critical where Western AI fails (e.g., Ganesh Chaturthi crowds).
Why It Works for India:
NavIC Satellites: Works when GPS jams near borders (no foreign dependency).
All-Weather Sensors: Radar sees through Mumbai rains when cameras fail.
BEL Security: Zero Chinese spy chips – DRDO-validated for defense use.
Impact:
Saves ₹18,000/- yearly per vehicle (insurance/repairs) – pays for itself in 6 months.
Creates ₹25,000/- monthly jobs for rural youth as remote operators. |
Western AI fails in monsoons/festivals – we use remote helpers only when needed (e.g., Ganesh Chaturthi crowds). Proven: 23 safe takeovers in Mumbai rains (Ola pilot). No other system trains AI on Indian cows/potholes/crowds. 2. ₹120,000/- Retrofit vs. ₹5,00,000/- New Cars: Magnetic installation works on any old car (2015+ models) – 1/50th the cost of new ADAS vehicles. Pays for itself in 6 months (₹18,000/- annual savings). Competitors target luxury cars – we serve India's 250M+ old vehicles. 3. Indian-First Tech Stack: NavIC satellites (works when GPS jams near borders) BEL security chips (zero Chinese spy risks – DRDO-validated) Radar + cameras (sees through monsoon rain) 4. Job-Creating Moat: Trains rural youth as remote operators (₹25,000/- monthly jobs) – creates political/social support and a trained workforce competitors can't replicate overnight. |
MVP |
Users |
1. Individual Auto/Taxi Drivers (Primary) 5M+ drivers operating old vehicles (Maruti 800, Hyundai i10) with zero safety tech. Why they buy: ₹120,000/- kit saves ₹18,000/- yearly (insurance/repairs) – pays for itself in 6 months. Critical for monsoon survival: "Wipers fail, but remote helpers take over," says Mumbai taxi driver. 2. Commercial Fleet Owners (Secondary) Ola/Swiggy/Zomato (1M+ vehicles nationwide). Why they buy: 200% higher vehicle uptime via remote repositioning + 68% fewer crashes (Ola pilot). Economic no-brainer: ₹50,000/-+ savings/vehicle/year. 3. Defense Forces (Strategic) Border Security/Army needing secure, jam-proof navigation. Why they buy: NavIC-enabled SensorPods work when GPS fails near borders (DRDO-validated). Zero Chinese parts = national security priority. 4. Rural Youth (Ecosystem) 50,000+ potential remote operators (Tier-2/3 cities). Why they engage: Earn ₹25,000/- monthly guiding cars during festivals/rain – no city migration needed. |
₹3,00,00,000/- Crore → 250M+ old Indian vehicles × ₹1,20,000/- kit price. |
₹7,26,000/- Crore → 6.05M vehicles realistically targetable: Commercial fleets only: 1M delivery vehicles (Ola/Swiggy) × ₹1,20,000/- = ₹1,20,000/- Crore Defense segment: 50,000 border patrol vehicles × ₹1,20,000/- = ₹6,000/- Crore Individual drivers excluded – ₹1,20,000/- is 12x their annual safety budget |
₹2,88,00,000/- (Years 1-3) → What we'll capture: Year 2: ₹2,40,00,000/- (200 Ola fleet vehicles × ₹1,20,000/-) Year 3: ₹4,80,00,000/- (400 fleet vehicles + DRDO pilot order) |
₹1,20,000/- SensorPod kits sold exclusively to defense/commercial fleets (individual drivers cannot afford this price). Year 2: 200 Ola fleet vehicles = ₹2,40,00,000/- (vs. 500 at ₹9k price) Year 3: 400 vehicles + 50 DRDO units = ₹5,40,00,000/- Critical caveat: Ola’s LOI vanishes at this price – drivers called it "8 months’ earnings" (vs. 6-month ROI at ₹9k). 2. Defense Premium Services (10% of revenue) ₹50,000/- monthly fee per vehicle for anti-jamming/naval-grade features (DRDO only). Why defense pays: BEL-made hardware = zero Chinese spy risks (national security priority). |
1. Western ADAS Giants (Mobileye/Bosch) What they sell: Full self-driving systems for new luxury cars (₹5,00,000/-+ premium). Why they fail: No human backup for Indian chaos (monsoons/festivals) – Ola pilot showed 100% failure rate in Mumbai rains. 2. Chinese Camera Kits (₹7,000/-) What they sell: Basic dashcams with fake "AI alerts" (no actual braking). Why they fail: Blind in rain/dust; Chinese spy chips (1 in 5 units had hidden data channels – 2023 study). 3. OEM Factory ADAS (Maruti Nexa) What they sell: Level 2 autonomy in new cars only (₹7,00,000/-+ premium). Why they fail: Ignores 250M+ old vehicles – where 80% of deaths happen. |
1. Fleet Owner Partnerships (Primary) Ola/Swiggy Pilot Expansion: Convert LOI into sales via free 30-day trials on 10 vehicles. Proven: 68% crash reduction → fleet managers buy 50x more kits. Revenue Share Model: Offer 15% commission to fleet owners for every new customer they refer (e.g., "Refer 5 drivers → get 1 free kit"). 2. Workshop Network (Grassroots) Train 1,000+ Mechanics: Pay ₹500/- per kit installed (₹4,500/- profit after cost). Mechanics become trusted advisors – "My uncle uses this, saved his Maruti 800 in monsoons!" Workshop Signage: "ARAI-Certified Safety Kit – ₹120,000/-" with QR code for instant booking. 3. Driver-Led Social Proof WhatsApp Demo Videos: Drivers share 15-sec clips of remote helpers taking over during rain (viral in auto unions). "Crash-Free Certificate": Drivers display digital badges – "60 Days Safe with SensorPod" → creates FOMO. |
Phase 1: Fleet-First Entry (Year 1-2) Ola/Swiggy Pilot → Scale: Convert LOI into sales via free 30-day trials (proven: 68% crash reduction → 500 kits ordered post-ARAI certification). Revenue Share: Offer fleet owners 15% commission for every new driver referral (e.g., "Refer 5 drivers → get 1 free kit"). Phase 2: Workshop Network (Year 2) Train 1,000+ Mechanics: Pay ₹500/- per kit installed (₹4,500/- profit after cost). Mechanics become trusted sales agents – "My uncle's Maruti 800 survived monsoons with this!" Workshop Signage: "ARAI-Certified Safety Kit – ₹120,000/-" with QR code for instant booking. Phase 3: Driver-Led Virality (Year 3) WhatsApp Demo Videos: Drivers share 15-sec clips of remote helpers taking over during rain (viral in auto unions). "Crash-Free Certificate": Digital badges ("60 Days Safe with SensorPod") create FOMO. |
By 2030, SensorPod will be the standard safety layer for all Indian vehicles – transforming 250M+ old cars into smart, connected units that prevent 150,000+ annual deaths. We’ll achieve this by:
Scaling Indian-First Safety:
Integrate with Bharat NCAP to make SensorPod mandatory for pre-2015 vehicles (like seatbelts today).
Expand remote operator network to 50,000+ rural youth (₹25,000/- monthly jobs) – turning India’s traffic chaos into skilled employment.
Defense-to-Civilian Flywheel:
Leverage DRDO border patrol contracts to fund R&D, then spin off affordable features to civilian kits (e.g., anti-jamming tech).
Global "Chaos-Proof" Export:
Adapt SensorPod for ASEAN monsoons and African dust storms – selling Indian solutions for global emerging markets. |
incorporated on 14/7/2025 |
NA |
Yes |
Avinya’s focus on deep tech for social impact aligns perfectly with our mission: saving Indian lives through affordable road safety. As an engineer who lost my uncle to a preventable monsoon crash in 2019, I built SensorPod – not as Silicon Valley-style "self-driving," but as monsoon-proof brakes for India’s 250M+ old vehicles. Avinya’s technical mentorship would accelerate our ARAI certification and NavIC integration – critical for Indian roads. Your defense-tech connections can fast-track DRDO collaboration (already in talks for border patrol trials). Most importantly, Avinya’s social impact lens matches our core truth: Safety tech must feed families while saving lives – which is why we train rural youth as remote operators (₹25,000/- monthly jobs). Unlike startups chasing vanity metrics, we’re validated on Mumbai’s flooded roads (68% fewer crashes in Ola pilot). Avinya’s ecosystem can scale this proven solution to prevent 50,000+ annual deaths – turning Indian roads from death traps into models of inclusive innovation. |
Your defense-tech ecosystem solves our #1 bottleneck (certification), while your social impact focus aligns with our mission: Safety that saves lives AND creates jobs. Unlike accelerators chasing vanity metrics, Avinya understands that 68% crash reduction (Ola pilot) matters more than app downloads. |
SensorPod was built for remote collaboration – our entire remote operator network trains rural youth via virtual sessions (₹25,000/- monthly jobs without city migration). Hybrid participation aligns perfectly with our DNA: Defense Tech Validation: Virtual access to Avinya’s DRDO experts would accelerate NavIC integration without travel delays (critical for border patrol trials). Workshop Network Scaling: Hybrid mentorship helps onboard 1,000+ mechanics nationwide via virtual training (already proven with Ola pilot). Cost Efficiency: As a bootstrapped startup, virtual engagement preserves capital for ARAI certification – where every rupee counts. |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/m993xX4C9jip-Black-Elegant-and-Modern-Startup-Pitch-Deck-Presentation-3_compressed.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/m993xX4C9jip-Black-Elegant-and-Modern-Startup-Pitch-Deck-Presentation-3_compressed.pdf |
https://drive.google.com/file/d/1h-5dYIxW0MrTdJbt3N98s2j6sfOsaQcK/view?usp=vids_web |
As an IT engineering student in Mumbai, I’ve seen 3 classmates lose relatives to preventable accidents. 150,000+ Indians die yearly – 80% in vehicles with zero safety tech. While peers build food-delivery apps, I’m coding survival tools for India’s roads.If our AI misses a cow in monsoon footage → people die. No "beta version" excuses.
Revenue model? Campus bus drivers pay ₹200/month because it saves repair costs (proven in pilot).
✅ Built for India, by Students:
Uses WhatsApp (not 5G) → works on ₹5,000 phones.
No Chinese parts → BEL sensors in final prototype (college lab partnership).
Rural job training → peers earn via monsoon-season shifts |
Yes – as a Tier-2 city IT student building for India’s masses, not Silicon Valley. I study at a non-NIT private engineering college in Maharashtra – where 85% of students are first-generation graduates with no tech industry connections. While IIT/NIT peers get funded for "Uber-for-X" clones, we solve what actually kills Indians.85% of Indian engineers graduate from colleges like mine – yet <5% of funded startups come from Tier-2/3 cities. |
whatsapp |
checked |
| Aug 28, 2025 @ 11:30 AM |
sachin kadam |
sachinkadam76199623@gmail.com |
https://www.linkedin.com/in/sachin-kadam-04bb721b0 |
https://NA |
08425919284 |
EDUCATIONAL |
1 YEAR |
Yes |
NA |
SG |
NA |
|
AI-powered business strategy and case analysis platform for MBA students |
MBA STUDENT |
Stratagem AI is an AI-powered platform designed to address these pain points with the following features:
Case study simulator: Users input a case study, and the AI analyzes the core challenges, stakeholders, and data. It then generates potential strategic options, including risk assessments and forecasted outcomes.
Competitive intelligence engine: Students can define a market, and the AI will continuously scan for market trends, competitor activities, and strategic opportunities, summarizing them into concise, actionable reports.
Dynamic business plan generator: The platform helps users build comprehensive business plans by assisting with market sizing, financial modeling, and competitive analysis, using real-time data for more accurate projections.
Data-driven insights: By analyzing a vast dataset of historical business performance, market trends, and industry benchmarks, the platform empowers students to make evidence-based decisions rather than relying on guesswork |
NA |
Idea |
Users |
COLLEGE STUDENT |
NA |
NA |
NA |
NA |
NA |
COLLEGE DATA |
NA |
NA |
NA |
NA |
No |
NA |
NA |
NA |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/4Vt7wMwXyr8I-Stratagem-AI.docx, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/4Vt7wMwXyr8I-Stratagem-AI.docx |
https://NA |
NA |
NA |
NA |
checked |
| Aug 28, 2025 @ 1:03 AM |
Milind Dongre |
milind.dongre@watthour24.com |
https://www.linkedin.com/in/milind-kumar-71b0b8174/ |
https://www.watthour24.com |
8291163925 |
1. Milind Dongre : B. Tech 1998 batch , Mechanical Engg. + MBA from IIM-C 2003 Batch, 21 Years of experience and 4 years work experience in Battery Tech and Material Synthesis. Founder & CEO at Watthour 2. Dr. Arjunan Ponniah: Ph.D in Material Sciences from Madras University. Battery Separator Synthesis Expert at Watthour. He has 10 Years of work experience 3. Dr. Vaibhav Acham,: Ph. D from NCL Pune, India and Materials Synthesis Expert at Watthour. He has 17 Years of work experience 4. Prof. Kothandaraman Ramanujam: Mentor to Watthour and Very Fine Electrochemist. Two Post Doc from Michigan University and 15 Years of Experience in R & D at IIT-Madras 5. Dr. Karthikeyan Kaliyappan: Almost 20 Years of Work Experience in Battery Technology and Mentor at Watthour 6. Dr. Hui Zhou and Dr. Jennifer Sammakia: Mentors at Binghamton University and experts at Battery Fabrications. 7. Dr. Milind Kulkarni: Ph. D and a Head of Center of Excellence with Ministry of Electronics and Information Technology of Govt. of India and very senior scientist with deep knowledge in Materials Sciences. 8. Dr. Anthony Leggerio : RIT Mentor , expert in Cells Fabrication 9. Dr. Matthew Ganter: RIT Mentor , expert in Cells Fabrication |
I met Prof. Kothandaraman Ramanujam at IIT-Madras, my alma mater, and he guided us on materials synthesis. And I met Dr. Vaibhav and Dr. Arjunan via common friends. And Mentors at Binghamton and RIT are known to me for past almost 1 year and they guide us on Battery/Cell Fabrication and they will fabricate and test a few cells for us in coming months. And Dr. Milind Kulkarni is a friend of mine and he is CMET (Center for Materials and Electronics Technology) Pune Head and we have signed MoU with them to develop Sodium Cathodes, Anodes and Cells. And we are also known to ARCI Hyderabad and we have insisted them to develop Sodium Cathode Materials and they will soon provide us the Materials. We will be fabricating the Cells at New York based Pilot Lines |
Yes |
Collective Experience, on Materials Sciences domain, of our Team is more than 100 Years and this is an Asset for us.
In our Team we have already done cells fabrication many times and also materials synthesis. This is helps us to develop cells and we need better facilities to work faster. |
Watthour Private Limited |
www.watthour24.com |
|
We make Sodium Cells to store Energy |
We are working |
We make Low Cost, High Capacity, Long Life Sodium Batteries using earth abundant materials.
We have synthesized Cathodes and Energy Dense Anode materials. We have also fabricated a few Cells such 26700 Cylindrical Cells and tested them, they are working fine. We are working to refine our Cathode, Anode and Cells. We have submitted Full Proposal for Energy Dense Anode for Sodium Batteries to DOE NOFO DE-FOA-0003504 and our LOI was responsive.
We have secured Pilot Line Access at NY , USA based University and on Indian Lab Pilot Line and we can fabricate Sodium Batteries using this facility. We are incubated at IIT-Madras Incubation Cell and got grants from them. We are also a member of Koffman Southern Tier Incubator and attending Soft Landing Program.
We provide Sodium Batteries to store Renewable Energy and make EV affordable and this reduces CO2 Emissions and make our Environment Healthier and Reducing CO2 Emissions by storing energy. Every 12.5 kWh energy stored will reduce 2.5 CO2 Emissions. And provide low cost electricity to needy groups and industries and generate more jobs and increases production and improves GDP. |
1. Almost half the Cost of Lithium Batteries 2. Using Earth Abundant Materials 3. Can Scale Up Faster 4. No Supply Chain Issues 5. Energy Dense Anode Materials 6. Easy Recycling 7. High Energy Density Sodium Sulfur Battery |
MVP |
Pilots, Signups |
We provide Low Cost Energy Storage Sodium Batteries. These batteries perform very close to Lithium Batteries and the potential clients are below 1. Data Centers 2. Renewable Energy Firms 3. EV Charging Infra and EV firms 4. GRID Sector 5. Defense 6. Telecom Firms 7. Residential , Commercial, Industrial Sectors 8. Electronics Products 9. Mining Instruments 10. Robots At present we are targeting Those sectors more which use Lead Acid Batteries as these Lead Acid Batteries have Low Depth of Discharge (50%) and have high Maintenance Cost and in the same cost we provide performance very close to Lithium Batteries and there is no Maintenance Cost. |
1000000000000 |
10000000000 |
100000000 |
We sell Batteries to generate Revenue and Profits |
CATL, HiNa, BYD, etc. |
We contact them directly and provide our Cells Test data and also participate in Government Tenders as participant with Big Group of Companies and we also look for Private Tender and have received soft commitments from some potential clients |
We want to replace Lead Acid Batteries as then perform Lower than Sodium Batteries and cost even higher. We can work with EV Firm for Charging Infra and Renewable Energy Firm ( Solar and Wind) |
Annual Consumption of Energy in USA us around 4 T kWh and almost 75% of this is produced using Fossil Fuel and US Govt wants to achieve New Zero by 2050-60. This is possible when they switch to Renewable Energy and this makes that almost 3 -3.5 T kWh is the total Market in USA and multiply it with $ 60 - 65 /kWh and we have $ 225 - 300 Billion Market only for Stationary Battery Energy Storages. We need to consider Electric Vehicles market as well and with that it will be close to $ 700 - 800 Billions Market Size. And if we take into account the growing Energy demand by the AI and Data Centers then it will over Multi - Trillion Dollar (200 Lacs Crore INR) Market Opportunity. TAM in USA is 200 Lacs Crore INR
And in India , Niti Ayog has estimated that by by 2033 the total BESS + EV Market Battery (with approximately 30% Conversion of ICE Vehicles to EV) Demand is close to 2 Tera Wh and this translates to 100 Lacs Crore INR as TAM.
And we have not taken into account Europe , African, Australia, UK and South America.
If we take an approximate estimate then TAM for whole world should be close to 500 - 700 Lacs Crore INR.
And if we can serve 0.1% of this Big Market then it should be close to 0.5 Lacs Crore INR.
I believe this is a Huge Market in India and in rest of world as well. |
We are registered as Private Limited Firm |
NA |
Yes |
1. Looking for access to Facilities, 2. Need Funds and connect with Investors , 3. Connect with Potential Clients |
1. Looking for access to Facilities, 2. Need Funds and connect with Investors , 3. Connect with Potential Clients |
Yes, we are open for hybrid or virtual participation |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/8yNxGkhCYWtZ-WATTHOUR-ONGC-V21.pdf, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/8yNxGkhCYWtZ-WATTHOUR-ONGC-V21.pdf |
https://www.youtube.com/shorts/gFBD3mPkHac |
Most of us know that India imports 85% of its Crude Oil either from Russia or Saudi Arabia. This is close to 15 Lacs Crore INR per Year or close to $ 133 Billions / Year. This is a Clear imbalance in our economy, we are almost completely dependent on others for our Energy, Transportation and other critical requirements. I can not even imagine, if there is a Geopolitical Tension or event that stops our Crude Oil Supply , it will throw our Normal Life to Pure Chaos or Stone Age.
According to Niti Ayog, Total BESS and EV Sector demand for Batteries is close to 2.3 Tera Wh by 2032 and we do not have a single Batteries Mass Producing Firm in India, all we do is to import from China or Korea or Japan. And most of time we focus on Lithium Batteries , and we do not make any Lithium Cathodes or have any raw materials sources for it.
This is really like being dependent on others for all critical requirements and can we really claim to be self-sufficient ? 1.5 Billion strong nation , almost producing nothing to store its energy. Indian Govt. is providing Incentives like PLI but only a handful of Big Business Houses ( those who have 300-400 Crore INR for past 5 years) can avail it and there is only handful of startups who really want to do it.
If we can save 1% of 15 Lacs Crore INR per year then it will be a Good Goal to achieve.
We need our own source of Energy and its Storage and we are making efforts in that direction and we need to do more and we need all support that you can provide. |
India is underrepresented in Energy Storage Domain. |
IIT-Madras Incubation Cell |
checked |
| Aug 27, 2025 @ 10:50 AM |
ljdcljandc |
dashsumanofficial@gmail.com |
https://wkhbvdjwbv |
https://lsfnvljsnv |
jvljnwlndrlg |
sdkvbwsbv |
svdbksjbvdljw |
Yes |
sfdv snf v |
skjvdbnswdl |
nvljwnflaenf |
|
lkwvhndelgwen |
sljvndljwns |
sdkfnknwskrvn |
sdnlvlkwenv |
MVP |
Users, Revenue, Signups |
sd.vmn w.asdn |
slfdnvksndvk |
;ksdvn;kwsm; |
s;knv;kws |
sndvknsw.vnk |
lsdnbnsw |
ms;kdvn;s |
ms;knv |
;skndvknsd |
svdnkelanv |
snvdljnsv |
Yes |
wevnfgwe;.knf |
sfdknvws.kn |
slvnkwn |
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/hKnDwtQNkPPk-Screenshot-2025-08-26-131641.png, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/hKnDwtQNkPPk-Screenshot-2025-08-26-131641.png |
https://docs.google.com/spreadsheets/d/1ZaaL7nStdTTAzeKLPZyWWiRYlCf7y17CZL3SVWytzVw/edit?usp=drive_link |
,abvjlnvjeg |
sndvklnnv |
LinkedIn |
checked |
| Aug 26, 2025 @ 4:23 PM |
jerold vinod |
jeroldvinod2@gmail.com |
|
|
09620736706 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/V1Pfvap9241d-fav-icon.png, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/V1Pfvap9241d-fav-icon.png |
https://acsel.in/community-driver/ |
|
|
|
checked |
| Aug 26, 2025 @ 4:19 PM |
jerold vinod |
jeroldvinod2@gmail.com |
|
|
09620736706 |
|
|
|
|
|
|
|
|
|
|
|
|
Users, Pilots, Signups |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1, https://acsel.in/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/pBx84RsUAkpe-Screenshot-2025-08-26-122100aa.png, /home/u520036516/domains/acsel.in/public_html/wp-content/uploads/forminator/1426_be1093b94c44838dc4f76ec2d9cab13c/uploads/pBx84RsUAkpe-Screenshot-2025-08-26-122100aa.png |
https://acsel.in/accelerator-application-form/ |
|
|
|
checked |
| Aug 26, 2025 @ 1:45 PM |
Suman Dash |
|
https://jdckwadc |
https://sfvnwsvv |
|
vdwevwevc |
wvfevfeqwfeq |
Yes |
wefvwefwe |
wdvwewfef |
eqvgeqeqv |
|
veqvqvq |
feqfqf |
fqefqhehfqhef |
dcfk;hebfcgqe |
MVP |
|
wdjvnlwhvc |
|
|
|
wevjlewhvv |
wrlnvgweljvn |
swdrvkwehv |
wvebvhkbwe;vsvdkwbvkgwehvwkhebvbbbb |
wehvkewhvk |
dvbkwaebvc |
wevdbgewqhbv |
Yes |
qevfnwjenfv |
srvhbgwejv |
wenhfvwhnef |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
svngws |
wvndkjwenlcw |
|
|
| Aug 26, 2025 @ 1:29 PM |
Murali |
|
https://linledn.com |
https://www.hello.com |
|
PLaying ROle |
Met fro few housr |
Yes |
Stay together |
|
|
|
dfsdf |
dfsdfs |
sdfsdfsdf |
|
Users |
|
|
|
|
|
zscz |
|
cz |
|
, . |
k |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Aug 21, 2025 @ 12:52 PM |
|
|
|
|
|
|
dagdhfjnm ,hs |
No |
aerdhjfg |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checked |
