From crop insurance to conversational analytics: Inside the winning ideas from the AI for Bharat Hackathon

India’s next wave of AI innovation is not just emerging from research labs or large enterprises. It is being built by students, startups, indie developers, and working professionals trying to solve practical problems across agriculture, healthcare, education, retail, and public services.

That was the larger takeaway from the AI for Bharat Hackathon, launched by Amazon Web Services (AWS) with innovation partner Hack2Skill and media partner YourStory. Over the past few months, the nationwide initiative brought together developers from across the country to build AI-powered applications using enterprise-grade AWS cloud and GenAI tools. 

The hackathon unfolded in two phases. The first focused on learning and skill-building through workshops, hands-on labs, mentorship sessions, and weekly challenges around cloud fundamentals, AI best practices, and India-focused applications. Participants also earned AI for Bharat certifications by completing labs and publishing technical blogs documenting their learning journeys on AWS Builder Center.

The second phase moved from learning to execution. Participants competed across Student and Professional/Startup tracks, building deployable AI solutions for real-world use cases in healthcare, agriculture, education, retail, content creation, public access, and developer productivity.

Each winning team received Rs 3 lakh.

AI for learning, content, and public access

In the Student Track focused on learning and developer productivity, Team Function Override, with Hemant Sight, Mohammad Hassan, Ishan K, and Ishu, won for SocraticDev, a platform designed to address overdependence on AI-generated code among developers.

Instead of simply generating answers, SocraticDev used guided learning techniques inspired by the Socratic method, nudging developers toward solving problems themselves. The platform combined AI chat, GraphRAG-powered codebase understanding, coding challenges, deterministic code visualization, and spaced-repetition learning inside a single workspace.

The platform stood out due to its focus on retention and comprehension rather than passive AI assistance.

The winner in the Students Track for media and digital experiences was Team Content Room, comprising Neil Immanuel, Shantanu V, Amisha Dsouza, and Ancilla Dsouza.

Their platform addressed the fragmented workflows faced by content creators, who often juggle multiple tools for writing, moderation, translation, research, and scheduling. Built on a serverless AWS stack using Lambda, Step Functions, Bedrock, Rekognition, Comprehend, DynamoDB, and CloudWatch, the system brought these workflows together into one AI-powered pipeline.

What tipped the scale in Content Room’s favour was its focus on cultural context and regional adaptation, rather than simple translation.

In the Student Track focused on public impact and accessibility, Team Rayquaza EX, comprising Amar Saifee, Sakshat Sachdeva, Prabal Minotra, and Sarthak Bhudolia, won for BimaSathi, a multilingual crop insurance claims platform built for India’s low-digital-literacy users.

The WhatsApp-first system guided farmers through claims filing in seven Indian languages, validated documents and uploaded images using AI, and generated insurer-ready claim packs. The platform also supported operator-assisted intervention through a dedicated dashboard.

The project tackled a major challenge in rural insurance systems: helping farmers successfully complete claims processes during periods of loss and uncertainty.

AI for retail, healthcare, and agriculture

In the Professional Track for retail innovation, Team MASS, which included Athira PV, Mohit Rautela, Mohammed, and Siddhesh Patil, won for RetailAI, a conversational analytics platform for small retailers.

The idea stemmed from a simple problem: small businesses generate large amounts of operational data but often lack the expertise needed to analyze it. RetailAI allowed users to ask business questions in plain English and receive insights, forecasts, inventory alerts, and recommendations powered by Amazon Bedrock Agents.

Built using AWS Bedrock, Athena, Glue, Lambda, S3, and DynamoDB, the platform aims to simplify business intelligence for India’s SME ecosystem.

The healthcare category was won by Team May The Agents Be With You, comprising Vinayak Mohan, Jaidev K, Lisha Patel, and Athira, for MedhaOS, an agentic AI healthcare platform designed to automate the patient journey from intake to post-discharge care.

The system coordinated multiple specialized AI agents handling triage, diagnostics, ICU forecasting, staff scheduling, supply management, and healthcare insights. The platform functioned across web, mobile, WhatsApp, and voice interfaces while maintaining patient context throughout the treatment journey.

The final winning team, Jay Jawan Jay Kisan, consisting of Rajat Sahu and Vaishali Jha, built Kisan Mitra AI, a Gen AI-powered agricultural assistant for digitally underserved farmers.

Built using Amazon Bedrock, Textract, and Rekognition, the platform translated soil health reports into actionable fertilizer recommendations, enabled AI-powered crop grading, and supported voice-first interactions in regional dialects.

Rather than prioritizing technical complexity alone, the team focused on accessibility, usability, and linguistic inclusion.

Beyond experimentation

Taken together, the winning projects reflected a larger trend emerging across India’s AI ecosystem: developers are increasingly moving beyond experimentation toward applications grounded in operational and social challenges.

Many of the solutions focused not just on automation, but on accessibility, multilingual support, public infrastructure gaps, and simplifying complex systems for everyday users.

AWS says it is already exploring what the next edition of the AI for Bharat Hackathon could look like, with plans to expand participation and deepen the focus on real-world AI adoption across sectors.