🌱 Inspiration Farmers in India and developing countries lose 20-40% of their crops to diseases annually. Most farmers can't afford agricultural experts and need instant, accessible solutions in their native languages. AgriSakhi was born to democratize agricultural expertise using AI.
🚀 What It Does AgriSakhi is a free, AI-powered Progressive Web App that:
Identifies 38+ plant diseases using Google's Gemini Vision AI Works in 4 languages: English, Urdu (اردو), Hindi (हिंदी), Kannada (ಕನ್ನಡ) Generates professional PDF reports with organic, chemical, and cultural treatment recommendations Works offline as an installable mobile app Completely free - No subscriptions, no ads, no barriers
🛠️ How We Built It Frontend & Architecture:
Next.js 16 with Turbopack for blazing-fast performance React 19 with TypeScript for type-safe code TailwindCSS for beautiful, responsive UI Zustand for state management PWA with service workers for offline functionality
AI Integration:
Google Gemini 1.5 Flash for real-time image analysis Custom prompt engineering for accurate disease detection Fallback system: Gemini → Hugging Face → Demo mode Multi-Language System:
react-i18next with 350+ translation keys RTL (Right-to-Left) support for Urdu/Arabic scripts Disease name translations for 20+ plant diseases Real-time language switching without reload Backend & Database:
Supabase for authentication and history storage PostgreSQL with Row Level Security Server-side API routes for secure AI calls PDF generation with jsPDF and autoTable
Key Features:
Disease Detection - Upload or capture plant images AI Analysis - Gemini Vision identifies diseases with confidence scores Treatment Recommendations - Organic, chemical, cultural, and prevention methods Knowledge Base - 60+ diseases with symptoms, causes, and treatments History Tracking - Save and review past detections PDF Reports - Professional multi-language reports Chatbot - Agricultural Q&A powered by Gemini
💪 Challenges We Faced
- Multi-Language Complexity
Translating disease names and technical agricultural terms Supporting RTL scripts (Urdu) while maintaining LTR (English/Hindi) Creating a seamless language-switching experience
- AI Integration
Training Gemini to output consistent disease format Handling model loading times and API rate limits Creating fallback systems for reliability
- Offline Functionality
Implementing PWA with service workers Caching strategies for 100+ MB of app assets Offline-first architecture with sync on reconnection
- PDF Generation Creating professional reports with embedded images Multi-language PDF support with proper fonts Keeping file sizes manageable (<2MB)
📚 What We Learned Advanced Next.js 16 features and Turbopack optimization Google Gemini Vision API and prompt engineering Complex i18n implementation with RTL support PWA development and service worker strategies Supabase real-time features and security policies Building accessible, farmer-friendly UI/UX
🎯 What's Next for AgriSakhi Voice Input: Add speech recognition in regional languages Soil Testing: Integrate soil health analysis Weather Integration: Provide disease risk predictions Community Features: Connect farmers for knowledge sharing Expert Network: Live consultation with agricultural experts Crop Planning: AI-powered crop recommendation system
🌍 ImpactAgriSakhi has the potential to help millions of farmers: Reduce crop losses by 15-30% through early detection Save costs on unnecessary treatments Bridge language barriers with native language support Democratize agricultural expertise for small-scale farmers Promote sustainable farming with organic treatment options
Built With
- 16
- 19
- ai
- gemini
- git/github
- jspdf
- next.js
- postgresql
- pwa
- react
- react-i18next
- supabase
- tailwindcss
- typescript
- vercel
- zustand
Log in or sign up for Devpost to join the conversation.