Nova AgriSense AI

The graphs
This is when Dash board is going to respond to Kinyarwanda
This is the dashboard in Kinyarwanda
The Data of situation
The other graphs
The updates
Dashboard can speak both languages English and Kinyarwanda

Inspiration

In Rwanda, many farmers still struggle to get fast, practical agronomy support in a language they are comfortable with. We built Nova AgriSense AI to make expert guidance accessible through a simple digital assistant that works in Kinyarwanda, English, or both.

What it does

Nova AgriSense AI is an AI-powered agriculture assistant that helps farmers:

Ask crop questions and get actionable guidance
Analyze crop photos for disease and nutrient issues
Receive recommendations on planting, fertilizer, and weather-aware decisions
Interact in Kinyarwanda, English, or bilingual mode

How we built it

We built the project using Next.js + TypeScript + Tailwind CSS for the frontend and API routes. The backend AI integration uses Amazon Bedrock (Nova models) to generate agricultural responses and image analysis results. We deployed the app on Vercel and configured AWS credentials securely via server-side environment variables.

Challenges we ran into

Fixing incomplete and broken API routes
Resolving build issues in frontend and TypeScript
Connecting the chat UI to real backend AI services
Solving AWS IAM and Bedrock access permission errors
Ensuring response quality across Kinyarwanda, English, and bilingual output modes

Accomplishments that we're proud of

Production-ready deployment with a live public URL
Full AI chat + image analysis pipeline working end-to-end
Farmer-focused multilingual UX (Kinyarwanda / English / both)
Clean GitHub repository ready for submission
Stable build and verified API routes in production

What we learned

We learned how critical cloud IAM permissions are in real deployments, how to design multilingual AI prompts for practical outputs, and how to transform a partially completed codebase into a working, deployable product under real constraints.