KrishiBotAI

Inspiration Agriculture is the silent engine of our world, yet for many farmers, it remains a high-stakes gamble. We were inspired by a simple question: How can we demystify farming for the everyday cultivator? We realized farmers need foresight, not just data. We wanted to move beyond static dashboards and create an immersive "Digital Twin" ecosystem. This led to KrishiBotAI—a fusion of serious agronomy and gamification, giving farmers a risk-free environment to simulate outcomes while carrying powerful diagnostics in their pockets.

What it does KrishiBotAI is a comprehensive agricultural ecosystem that:

Gamifies Farming: Features EcoFarm, a hyper-realistic farm simulator driven by live environmental data, allowing users to test crop strategies before planting. Predicts Risks: The Forecasting Hub analyzes weather and soil data to generate PDF reports on crop viability and potential risks. Diagnoses Instantly: Uses a Disease Detection System and Soil Type Prediction System to identify plant health issues and soil properties from simple images. Connects Farmers: Offers a Community Forum for knowledge sharing and a Personalized AI Chatbot with two-way voice communication for accessible, hands-free advice. How we built it We used a modular "hub-and-spoke" architecture:

Core Brain (FastAPI): A high-performance backend orchestration layer that handles everything from TensorFlow inference to game state management. Simulation Engine: We built a growth algorithm where crop health is a mathematical function of real-time variables—Temperature (Open-Meteo), Rainfall, and Soil pH (NARC API)—rather than random chance. Intelligence Layer: We integrated specialized ML models for disease detection and soil analysis, alongside Google's Gemini API to power the natural language voice assistant. Challenges we ran into Data Synchronization: syncing the NARC Soil API with live Open-Meteo weather data to drive the game engine without breaking the simulation logic was complex due to differing data scales. Voice Latency: Achieving smooth, natural two-way voice communication required significant optimization of audio buffer streams to minimize delay. Model Optimization: Serving heavy TensorFlow models alongside a real-time game engine required careful resource management to prevent the backend from hanging during inference. Accomplishments that we're proud of Successfully creating a "Digital Twin" simulation that reacts to real-world geography (NARC soil data). Building a truly farmer-centric design by implementing two-way voice communication, making high-tech insights accessible to users regardless of literacy levels. Integrating multiple complex systems—gamification, ML diagnostics, and forecasting—into a single, cohesive "Liquid Glass" UI. What we learned We learned that gamification is a serious learning tool; users understand the impact of pH and weather far better by "playing" with the variables than by reading manuals. We also discovered that for an agri-tool to be effective, accessibility (voice support, simple UI) is just as critical as the accuracy of the underlying machine learning models.

What's next for KrishiBotAI Market Integration: Adding a direct marketplace for farmers to sell produce based on yield forecasts. IoT Connectivity: Integrating with physical soil sensors for real-time, on-site calibration of the simulation. Offline Mode: Optimizing the lightweight TFLite models to run entirely on-device for remote areas with poor connectivity.

Built With

• 15
• api
• css
• fastapi
• gemini
• git
• google
• leaflet.js
• narc
• next.js
• open-meteo
• postgresql
• python
• shadcn
• soil
• sqlalchemy
• tailwind
• tensorflow
• typescript
• ui