AgriGPT AI for Farmers, Anywhere

Helping farmers in their own language with crop advice, government schemes, and emergency guidance powered by cutting-edge AI technology.

Comment

Inspiration

Farmers are the backbone of every nation. Yet, millions of small and marginal farmers lack access to timely, localized agricultural guidance. Language barriers, limited internet literacy, and high consultation costs prevent them from receiving expert advice. Our inspiration was simple: What if every farmer had a personal AI companion — one that could listen, speak, and guide in their own language, anytime, anywhere?

What it does

AgriGPT is an AI-powered farming assistant that helps farmers:

🎤 Ask questions by voice in local languages (Hindi, Punjabi, English).

🧠 Get real-time AI answers for crop care, soil, pests, and weather.

📜 Access simplified government schemes & subsidy info through retrieval.

🔊 Listen to responses back in their own language with Text-to-Speech (TTS).

📱 Use a clean, mobile-first web app with a beautiful and intuitive UI/UX.

How we built it

Frontend: React + TailwindCSS + Framer Motion → stunning and responsive UI.

Backend: Node.js / FastAPI with REST APIs.

AI Stack:

Whisper API → Voice to text.

Hugging Face GPT-OSS-20B → Smart text generation.

Coqui TTS → Text-to-speech in local languages.

LangChain + FAISS → Retrieval of government scheme documents.

Integration: Frontend chat connects with backend endpoints: /stt, /ask, /tts, /retrieve.

Challenges we ran into

Making the system multi-lingual (especially for rural languages).

💻 Balancing between offline wow factor and API-based MVP for hackathon.

⏱️ Time limits: ensuring a working demo instead of just a concept.

⚙️ Integrating different APIs (Whisper, Hugging Face, Coqui) smoothly.

Accomplishments that we're proud of

Built a fully working voice-to-voice AI companion in hackathon time.

Created an aesthetic, user-friendly landing page that feels production-ready.

Managed to integrate free APIs to avoid heavy infra dependencies.

Designed with real-world farmers in mind, not just as a tech demo.

What we learned

The power of OSS models like GPT-OSS-20B when integrated creatively.

Importance of UI/UX for adoption: even the best AI is useless if the interface is not friendly.

How to balance cloud APIs vs. offline models in real-world hackathons.

That building AI for social good is both impactful and inspiring.

What's next for AgriGPT AI for Farmers, Anywhere

📲 Offline-first version with on-device inference (for rural areas without internet).

🌎 Expand language support to 10+ Indian and global languages.

📊 Add predictive analytics for yield, soil health, and market prices.

🤝 Partner with government & NGOs for real-world pilot programs.

☁️ Launch as a progressive web app (PWA) for low-end smartphones.

Built With

Share this project:

Updates