FarmConnect — Agritech Marketplace & Advisory

FarmConnect is an integrated agritech platform that addresses three interconnected problems facing smallholder farmers in East Africa: access to markets, access to inputs, and access to agronomic expertise. Smallholder farmers — who produce seventy percent of Kenya's food — typically sell their produce through middlemen at prices far below retail, buy inputs at inflated prices from local agroshops without quality assurance, and receive agronomic advice that is generic, infrequent, and often wrong. FarmConnect solves these problems through three linked products. The marketplace connects farmers directly to institutional buyers — supermarkets, processors, export agents — enabling price discovery and transparent contract farming arrangements. The input marketplace allows certified agrodealers to list seeds, fertilisers, and crop protection products with verified quality certifications, and allows farmer cooperatives to aggregate orders for bulk pricing. The AI advisory engine provides personalised, localised advice based on the farmer's specific crop, soil type, location, rainfall pattern, and current market prices. The AI advisory engine is the most technically interesting component. We are building a fine-tuned LLM that has been trained on Kenya's agricultural extension service knowledge base, County agricultural officer reports, CIMMYT crop research publications, and aggregated farmer outcomes from our platform. The model will provide advice in Swahili and Kikuyu as well as English, and will be accessible via WhatsApp as well as the mobile app — critical for farmers who are comfortable with WhatsApp but have not yet adopted dedicated apps. The marketplace logistics layer integrates with motorcycle courier networks for last-mile delivery of inputs and a network of aggregation points for produce collection. The platform also includes a financial layer: farmers can access input financing through a partner SACCO, with repayment automatically deducted from their produce sale proceeds. The stack is Python/FastAPI for the backend, React Native for the mobile app, PostgreSQL for transactional data, and a vector database for the AI advisory knowledge base.