• 

  Landing page - form to capture location data

• 

  feedback generated based on location

Project Story

Inspiration

SeedMatch AI was inspired by the need to empower smallholder farmers with localized agronomic knowledge. Many farmers still rely on guesswork or general advice when selecting seeds, often resulting in poor yields and economic loss. Having worked in agriculture and digital innovation, I saw an opportunity to combine AI and browser technology to give farmers instant, data-driven recommendations—right from their Chrome browser, even without strong internet access.

What it does

SeedMatch AI helps farmers discover the best seed variety suited to their specific altitude and location. The app uses Chrome’s built-in Prompt API to analyze a local dataset of seed varieties and generate personalized recommendations directly on the browser. It also integrates the Translation API to present results in Kiswahili, making the tool accessible to a wider range of users.

How we built it

The project is built with Next.js and TailwindCSS, using Chrome’s built-in AI APIs for reasoning and translation. The browser’s Geolocation API collects location and altitude data, which is matched against a Supabase or JSON-based seed dataset. The Prompt API identifies the most suitable seed variety, while the Translation API converts the response into local language. All operations run within the browser for privacy and offline accessibility.

Challenges we ran into

Working with experimental Chrome AI APIs came with integration challenges, including limited documentation and feature stability. Crafting precise prompts to ensure accurate yet easy-to-understand recommendations required multiple iterations. Managing local datasets efficiently while keeping the app lightweight was also a key technical challenge.

Accomplishments that we're proud of

We successfully demonstrated how Chrome’s built-in AI can power a useful, offline-ready agricultural tool without any external AI model or heavy backend. SeedMatch AI runs privately in the browser, giving farmers meaningful insights in real time. Seeing it produce accurate, contextual advice for different altitudes and crops was a major milestone.

What we learned

This project deepened our understanding of edge AI and privacy-first web apps. We learned how to design prompts that yield localized, practical outputs and how browser-based AI can solve problems traditionally requiring cloud models. It also showed the importance of inclusive design, particularly the role of translation in breaking digital barriers.

What's next for SeedMatch AI

Next, we plan to expand the dataset to cover more regions and crop varieties and integrate offline caching for improved accessibility. We also aim to add a voice interface for low-literacy users and explore integration with agricultural APIs to update data dynamically. Ultimately, we hope SeedMatch AI can evolve into a full Chrome-based agricultural advisor for farmers across Africa.

Built With

• chrome-built-in-ai-apis-(prompt-and-translation)
• gemini-cli
• github
• next.js
• react
• supabase
• tailwindcss
• vercel
• visual-studio