KissanAI and UNDP: a CoPilot for farmers

On 5 January 2024, KissanAI announced a partnership with the United Nations Development Programme to develop a voice-based, vernacular CoPilot for Climate Resilient Agriculture. The aim is to make regional agricultural guidance easier to reach for smallholder farmers affected by climate volatility, with particular attention to women farmers.

What the CoPilot is designed to do

The CoPilot is designed around Dhenu, our domain model for Indian agriculture, and a voice interface for Indic languages. A farmer should be able to ask about crop scheduling, a pest symptom, mandi prices, scheme eligibility, or irrigation timing in a familiar language and receive an answer in that language.

The climate-resilience work combines agricultural knowledge with current local context. For example, a sowing-window answer may need the short-range forecast, soil moisture, and the varieties recommended for that crop and district. Those inputs have to come from maintained sources rather than from the language model’s memory alone.

The system is not a substitute for an extension officer or a soil test. It is intended to help with the gap between a farmer’s question and access to qualified, local guidance.

What the partnership covers

UNDP brings experience with climate-resilient agriculture, local programmes, and the organisations that work directly with smallholder farmers. KissanAI brings the model, voice and language pipeline, agricultural knowledge base, and deployment engineering.

The partnership focuses on climate-vulnerable smallholders, including women farmers who are often excluded from technology programmes. For a rainfed cotton or millet farmer, a change in the sowing window or water plan can affect the whole season. The advisory therefore has to be local, current, and clear about uncertainty.

Why a development agency and a startup are working on the same problem

“The smallholder farmer is not a venture-priced customer,” says KissanAI founder Pratik Desai. “They are also exactly the population that climate-resilient advisory can serve at the lowest unit cost.” The partnership combines a development mandate with a technical team that can build and test the service.

Deployment results should be published only when a cohort is complete and the data is approved for release. That includes usage, language coverage, and outcome measures where farmers and partners have agreed that the results can be shared.

For media or partnership inquiries, write to hello@kissan.ai.