Maha Vistaar AI
Agriculture & AgriTech
Real-time agri-intelligence platform using agentic AI and open protocols. Delivers weather, soil health, mandi prices, and government scheme data to 2M+ farmers across Maharashtra.
MahaVISTAAR AI is a voice-first agricultural intelligence platform built for the Government of Maharashtra in partnership with IIIT Bangalore and OpenAgriNet. It delivers real-time crop advisory, weather intelligence, mandi prices, soil health data and government scheme information to farmers across Maharashtra — accessible via a simple phone call in Marathi, regional dialects and tribal languages.
Key capabilities
Voice-first access
Farmers call a phone number and speak in their language. No app download, no smartphone, no internet connection required.
Real-time multi-source advisory
AI agents compose responses from live weather data, soil health records, mandi price feeds and government scheme databases in real time.
Multilingual & tribal language support
Serves queries in Marathi, Hindi, and tribal languages including Bhili — reaching communities no prior digital platform had served.
Agentic AI architecture
Specialised AI agents handle crop advisory, pest identification, weather forecasts, mandi prices and scheme eligibility — each fine-tuned on agricultural domain data.
Built on open protocols and public infrastructure
MahaVISTAAR is built on the Beckn Protocol and OpenAgriNet DPI — India's open digital infrastructure for agriculture. The platform uses a network-of-agents architecture where each agent specialises in a domain (crop, pest, weather, mandi, schemes) and composes responses collaboratively. Open-source language models fine-tuned on Indian agricultural data power the intelligence layer, with the Vistaar Protocol enabling interoperability across state and national systems.
Gallery
Kenpath's role
Kenpath partnered with IIIT Bangalore and collaborated with OpenAgriNet and the Government of Maharashtra to build MahaVISTAAR. As the core technology partner, Kenpath designed the agentic AI architecture, fine-tuned language models on agricultural domain data, built the voice interface pipeline, and integrated with Beckn Protocol and OpenAgriNet DPI. The team worked directly with government officials, agricultural scientists and farming communities to ensure the platform addressed real-world needs, not hypothetical ones.