Inspiration

Farmers often face difficulties comparing crop prices across different mandis and finding the most suitable market to sell their produce. We were inspired to build MandiDarpan to make agricultural market information more accessible and help farmers make informed selling decisions using technology and AI.

What it does

MandiDarpan is an AI-powered mandi price intelligence platform designed to support farmers. It provides:

Mandi Price Checker: Helps farmers check available crop prices in different mandis.
Nearby Mandi Finder: Helps farmers discover nearby mandis and view routes.
Profit Estimator: Estimates potential earnings after accounting for transportation and other costs.
AI Assistant: Allows farmers to ask questions in Hindi or Hinglish and receive easy-to-understand responses.
Price Trends: Aims to show historical crop price movements when sufficient historical data is available.

How we built it

We built MandiDarpan using a combination of web development, data processing, AI, and mapping technologies.

Frontend: React, Vite, and Tailwind CSS for the user interface.
Backend: Node.js and Express.js to handle API requests and process mandi data.
Data Processing: Python and Pandas for cleaning and organizing agricultural market data.
AI Integration: A Gemma model hosted through Google AI Studio to help understand farmers' questions and provide explanations.
Market Data: Government mandi price data from data.gov.in.
Maps and Routing: Leaflet and OpenStreetMap for maps, with OSRM for route information.

Challenges we ran into

One of our main challenges was retrieving and filtering mandi price data accurately. We encountered situations where the API returned data successfully but filtered results were empty. Understanding the difference between data retrieval, pagination, and filtering was an important part of debugging.

Another challenge was integrating multiple technologies into one platform while keeping the experience simple and accessible for farmers. We also needed to ensure that AI-generated responses rely on actual market data rather than inventing crop prices.

Accomplishments that we're proud of

We brought together several useful features in one platform: mandi price discovery, nearby market exploration, estimated profit calculation, and an AI assistant designed for Hindi and Hinglish users.

We are also proud of building a project that applies AI and publicly available agricultural data to a real-world problem. Our focus is on making complex market information easier for farmers to understand and use.

What we learned

Through this project, we learned how to connect a frontend to a backend, work with APIs, process structured data, and troubleshoot filtering issues. We also explored how AI can help make information more accessible through natural-language interactions.

Most importantly, we learned that a useful AI application must be grounded in reliable data, communicate uncertainty clearly, and be designed around the needs of its intended users.

What's next for MandiDarpan

Our next steps are to improve data filtering and coverage, make mandi prices more current, and enhance the nearby mandi recommendation experience.

We also plan to add historical price charts when sufficient historical data is available, improve estimated transportation costs, expand language support, and make the platform more accessible on mobile devices. In the long term, we want MandiDarpan to become a trusted digital companion that helps farmers make more informed marketing decisions.