Krishi AI

Krishi AI is a maintained, multilingual Android farming assistant that combines AI guidance, plant diagnosis, live mandi prices, voice, weather, and practical agricultural knowledge. I built it independently to make useful information easier for Indian farmers to reach in their own language and context.

Problem and intended users

Farmers often need timely guidance across crops, pests, processes, weather, and prices, but the information is fragmented and may not be easy to use in a local language or while working in the field. Krishi AI is designed for farmers and other agricultural stakeholders who want one accessible starting point for those questions.

My role and contribution

I am the founder and sole product developer. I handle product direction, Android development, backend services, infrastructure, releases, public documentation, and the feedback loop around the product.

Approach and technology

The Android application uses Kotlin, Jetpack Compose, MVVM, and clean-architecture patterns. Firebase and cloud functions support authentication, history, application services, and AI integrations. Live mandi prices come from India’s Open Government Data platform, while the product also includes plant diagnosis, local weather, crop knowledge, multilingual voice conversations, saved history, offline-aware behaviour, and PDF export.

The idea has a longer history. DigitalMandi reached 41,000 users in 2017 before I stopped maintaining it. Krishi AI is a return to that mission with a broader product and a more disciplined commitment to continuing the work.

Evidence and current outcomes

• The Google Play listing publicly shows 10K+ downloads and identifies me as the developer.
• The source repository documents the Android and cloud implementation.
• I publicly documented acceptance into the Google for Startups Cloud Program.
• I publicly announced a strategic partnership with Headway Earth to offer Krishi AI AgriTech services to its customers in Nepal.
• I shared the build journey in the talk From Village Fields to AI: Building Krishi AI with LLM Tools.

Limitations and status

Krishi AI is still evolving. AI-generated guidance can be incomplete or wrong, market and weather data can change, and the app should not replace local agronomic expertise or professional judgment in consequential decisions. Public milestones show reach and programme participation; they do not by themselves prove improvements in yield, income, cost, or environmental impact, so I do not claim those outcomes here.