🌱 Inspiration Smallholder farmers across Sub-Saharan Africa face significant challenges—limited access to agricultural expertise, poor disease diagnosis, unreliable weather data, and minimal support in their local languages. In Kenya, over 75% of farmers depend on rain-fed agriculture. In Senegal and Côte d’Ivoire, more than 60% of the population works in agriculture, yet productivity remains low due to climate variability, knowledge gaps, and inaccessible technology. We wanted to empower these farmers with a simple yet powerful AI tool they could use with just a smartphone.
🤖 What it does Agri Assist is an AI-powered assistant for farmers that:
Diagnoses crop diseases through image uploads.
Provides localized weather forecasts.
Delivers conversational support in English and French with agricultural best practices.
Summarizes actionable recommendations in a farmer-friendly format.
🛠️ How we built it Frontend: Built using React, Vite, and TailwindCSS for a clean, fast, and mobile-responsive UI.
Backend: Gemini 1.5 Flash API for real-time image, voice and text generation.
AI: Prompt engineering to extract reliable, structured JSON from unstructured Gemini outputs.
Hosting GitHub for automatic deployment.
Multilingual Support: English and French localization to serve farmers in Kenya, Senegal, and Côte d’Ivoire.
🧱 Challenges we ran into Handling Gemini hallucinations and ensuring responses were consistently in structured formats.
Diagnosing diverse crop images with varying lighting and image quality from real farms.
Localizing weather and agricultural information for different regional contexts.
🏆 Accomplishments that we're proud of Building a real-time AI assistant that works from a smartphone browser with no app install.
Integrating both image voice and text-based AI to offer multimodal support.
Supporting both English and French, covering a majority of farmers in East and West Africa.
Deploying a live demo with and GitHub in under a 24 hours.
📚 What we learned AI can be incredibly powerful in agriculture, but it requires deep context and localization to be useful.
Prompt engineering is an art: we had to iterate carefully to make Gemini generate structured, useful responses.
A streamlined build process and GitHub-to-Firebase deployment saved time and allowed rapid updates.
🚀 What's next for Agri Assist Integrating offline capabilities via Progressive Web Apps (PWAs).
Mobile application on Android and IOS
Expanding support to other African countries like Nigeria, Ghana, and Uganda.
Partnering with local extension services and agribusinesses to offer real-time expert consultation.
Agri Assist is more than a tool—it's a step toward AI-powered resilience for African farmers. 🌍
Built With
- ai
- gemni
- react
- typescript
Log in or sign up for Devpost to join the conversation.