Inspiration
Indian farmers lose ₹40K–80K per season simply due to poor sell-timing. We wanted to bridge the gap between complex market analytics and everyday farming by building an intelligent assistant that tells farmers exactly when to sell, in their own language.
What it does
AgriBuddy is a predictive market-timing assistant. It evaluates market trends, weather patterns, and expert advisories to generate definitive "SELL or HOLD" verdicts. Farmers can also chat with the platform in plain language to ask about current mandi prices and market trends across 9 different Indian languages.
How we built it
Data Pipeline: We used Auto Loader to ingest APMC mandi prices into a Databricks Delta Lake medallion pipeline. Forecasting & RAG: We trained Prophet forecasts (tracked in MLflow) and retrieved ICAR expert advisories using FAISS RAG. AI & Interface: We combined this data so Sarvam could generate cited recommendations. Sarvam Mayura translates this into 9 languages, and Databricks Genie powers the natural-language conversational interface for seamless data querying.
Challenges we ran into
integrating the databrick genie
Accomplishments that we're proud of
What we learned
What's next for AgriBuddy
We plan to expand our dataset to include more crops and remote mandis. Next steps also include deploying a WhatsApp integration for voice-note queries and setting up real-time SMS alerts for sudden market price spikes.
Log in or sign up for Devpost to join the conversation.