The insight
There are approximately 150 million smallholder farmers in India. Most of them are on WhatsApp.
That is the starting point: not a custom app, not a web platform, not an IVRS call center — WhatsApp. Because that is where farmers already are, where their family groups live, where they share information and ask questions. The adoption problem is solved before you write a single line of code.
The advisory problem, however, is real. A farmer in Madhya Pradesh or Rajasthan deciding whether to apply an extra dose of urea, wondering what rate their wheat is fetching at the local mandi, or trying to identify what is wrong with their crop has very limited options for getting a reliable, timely answer. Extension workers serve hundreds of farmers each. The internet is there, but not always in Hindi, not always intelligible, and not always trustworthy.
What Maati AI does
Maati (माटी) means soil — earth — in Hindi. The name grounds the project.
Farmers can message the assistant in Hindi (or their regional language) and ask:
- Agri-input questions — "Which fertilizer should I apply to my wheat at this stage?" "Is my pesticide compatible with this crop?"
- Mandi bhav (market prices) — "What is the MSP for paddy this season?" "What is the rate at Indore mandi today?"
- Weather — "Will it rain this week?" "What does this affect for my kharif crop?"
- Crop health — "My tomato leaves are yellowing at the edges, what could it be?"
- Seasonal guidance — "When should I sow mustard in my region?"
The responses are in plain, conversational Hindi — not technical jargon, not formal text.
The technology
The core is an LLM with agricultural domain context served through the WhatsApp Business API. The design choices reflect the constraints of the user:
- WhatsApp-native: no app install, no login, no new interface to learn
- Vernacular first: Hindi input/output, with scope for regional languages
- Low-bandwidth friendly: text-based, no heavy media required
- Voice note support: farmers who prefer speaking over typing can send voice notes
Connection to Carbonify
Maati AI and Carbonify are two facets of the same underlying interest: improving the material conditions of Indian farming, both through better inputs (Carbonify) and better information access (Maati AI). They are independent projects but pointed in the same direction.
Current status
Actively building. The core advisory capability is working in Hindi. Mandi price integration and weather are in progress. Not yet publicly deployed — the goal is to trial with a small group of farmers in our network before broader release.