Inspiration
India’s agriculture sector is the backbone of the country, yet millions of farmers still struggle with access to reliable guidance, weather warnings, pest control strategies, and awareness about government schemes. We were inspired to create KrishiKripa to bridge this gap — offering a friendly, always-available AI companion that empowers rural farmers using the latest in conversational AI, all while keeping things simple and localized.
What it does
KrishiKripa is an AI-powered chatbot designed to assist Indian farmers with: Personalized crop advice Weather-aware farming suggestions Early warnings for pests & diseases Government schemes (PM-Kisan, PMFBY, etc.) explained simply Mandi price trends (static for demo) Emergency response during floods/droughts Emotional and motivational support in times of distress
How we built it
We used Fetch.ai’s uAgents framework to create a decentralized, autonomous AI agent. The language model backend is powered by ASI:One LLM (asi1-mini) for contextual, natural conversations. We designed a custom prompt and memory architecture to preserve recent user interactions and maintain conversation flow. The application was deployed on Agentverse, enabling direct interaction with users via a web interface. For emergency assistance, we implemented a static helpline lookup system that detects state-based keywords and surfaces relevant support numbers.
Challenges we ran into
Context retention with limited short-term memory constraints posed a challenge when farmers asked multiple related questions. Designing a chatbot that is both technically intelligent and emotionally empathetic took careful prompt engineering. Providing accurate and location-specific data (e.g., weather, mandi prices) was challenging without real-time APIs in the hackathon timeframe. Localization remains a key challenge — integrating multilingual support for rural users is non-trivial.
Accomplishments that we're proud of
Created a fully functional conversational agent in a short span using cutting-edge decentralized AI frameworks. Successfully integrated emergency detection and state-wise helpline support. Developed a solution that is scalable and ready for real-world rural deployment with minor enhancements. Received positive feedback for the simple, farmer-friendly tone of the chatbot
What we learned
Learned how to build, test, and deploy uAgents using Fetch.ai’s Agentverse platform. Understood the power and challenges of integrating LLMs with contextual memory. Improved our skills in designing real-world conversational flows for users with varying levels of tech literacy. Explored the intersection of AI and agriculture, and how small innovations can lead to massive impact.
What's next for KrishiKripa
Multilingual Support We aim to support major Indian languages (Hindi, Tamil, Telugu, Marathi, Bengali, etc.) using language models fine-tuned for regional dialects, enabling better outreach to rural farmers. Integration with Real-Time APIs Connect to live APIs for weather forecasts, mandi prices, and satellite-based pest alerts to offer real-time, hyperlocal guidance. WhatsApp & SMS Integration Make KrishiKripa accessible via WhatsApp and SMS for farmers with basic phones and no internet access. Farmer Learning Mode Introduce a quiz-based, gamified module where KrishiKripa teaches sustainable practices and rewards farmers with tips, badges, or local NGO points. IoT & Sensor Integration Integrate with soil sensors, drones, and farm-based IoT systems for precision farming recommendations. Dynamic Government Scheme Updater Automate updates on newly launched or revised government schemes and eligibility criteria. Mental Health and Wellness Companion Add a dedicated module for emotional support, motivational tips, and rural mental health awareness. Collaborations with AgriTech Startups & NGOs Partner with organizations already working with farmers to pilot KrishiKripa in real field conditions. Farmer Dashboard Provide farmers with a personalized dashboard tracking their land, crops, weather, and scheme eligibility in a mobile-first interface.
Log in or sign up for Devpost to join the conversation.