Krushit is a state-of-the-art AgriTech platform designed to bridge the gap between traditional farming and modern technology. By combining Internet of Things (IoT) sensors with Artificial Intelligence (AI), Krushit provides farmers with a "Command Center" for their fields—accessible in their native language and available even on low-spec mobile devices as a PWA.
Our mission is to reduce crop loss, optimize resource usage (water/fertilizer), and provide actionable intelligence to the farmers of tomorrow.
A unified interface where farmers can monitor their entire operation at a glance.
- Farm Health Score: A proprietary AI-driven metric calculating overall farm status.
- Real-Time Node Architecture: A state-of-the-art Push-based Supabase Realtime Engine completely hydrates identical sensor payload data globally without ever refreshing.
- Urgent Alerts: Instant notifications for disease outbreaks or irrigation needs.
Leveraging TensorFlow.js and high-accuracy neural networks to identify plant pathology.
- Instant Scans: Detect 38+ common crop diseases with >95% accuracy.
- Persistent Tracking: A locally mirrored and permanently synced diagnostics history lets farmers trace crop lifecycle improvements over time.
- Native Document Export Engine: Built natively with
jsPDFandjspdf-autotable, generate strictly formatted, A4 professional vector reports outlining detected pathologies, granular budget estimations, and mitigation protocols immediately onto portable devices.
A smart companion powered by Gemini 2.5 Flash.
- Voice & Text Input: Farmers can speak or type in English, Hindi, or Marathi.
- Context-Aware: The bot knows your crop type, soil moisture, and local weather.
Real-time monitoring through distributed field sensors.
- Smart Irrigation Windows: Tells you the exact hour to water to minimize evaporation.
- Nutrient Roadmap: Tracks NPK and pH levels to optimize fertilization cycles.
Helping farmers bridge the financial gap securely.
- Smart Personalized Engine: A dynamic matching logic reads the user's localized risk profile and maps the optimal federal/state subventions out of dynamic registries directly pushing them based on farm scale and conditions.
- Direct Apply: Direct links to official portals (PM-KISAN, PMFBY, KCC).
- Globally Tracked Real-Time GIS: Submitting an active hazard payload via native GPS coordinate mappings immediately forces a push broadcast to all surrounding
DiseaseMapcomponents natively out of the central Supabase nexus. - Anonymized Reporting: Contribute to the local data to prevent regional outbreaks securely.
- Framework: App Router, Server Components.
- State Management: Zero-refresh mapping through integrated
Supabase WebSocket channels, unified Hooks, + i18next global state. - Engine Exports: Pure client-side highly formatted vector text extraction via
jsPDF. - Aesthetics: Glassmorphism, smooth micro-interactions, and visual data cards.
- i18n: Fully localized in English, Hindi, and Marathi with instant toggle.
- Architecture: Asynchronous REST API.
- Database: Supabase/PostgreSQL for scalable, heavily optimized real-time Row-Level-Secured data storage.
- Auth: Secure JWT-based authentication.
- AI/ML Integration:
- TensorFlow: For image-based disease prediction.
- Google Gemini: For natural language processing and advisory.
- Data Pipeline: Real-time telemetry processing dynamically routed via secure push-model WebSockets logic straight into the dashboard without destructive polling endpoints.
- Analytics: Historical trending of farm health metrics.
We believe technology should speak the farmer's language. Krushit is built from the ground up to support:
- English (Standard)
- हिंदी (Hindi) (Primary)
- मराठी (Marathi) (Native Support)
Selected language persists through sessions and synchronizes across the Dashboard, Chatbot, and Reports.
The Krushit platform is a distributed system designed for resilience, speed, and linguistic accessibility.
- Frontend: A Progressive Web Application (PWA) built with Next.js 14, providing a smooth, app-like experience on mobile and desktop.
- Backend: High-performance FastAPI services handling real-time requests, system logic, and AI orchestration.
- Database: Supabase (PostgreSQL) for secure, real-time data storage of farmer profiles, crop records, and advisory history.
- AI/ML Module: A hybrid system using TensorFlow.js (client-side) and Gemini Flash (server-side) for crop disease detection and smart reasoning.
- External Services: Integrated IVR system capabilities for supporting farmers using feature phones (non-smartphones).
graph LR
A[Farmer: Mobile / IVR] --> B[Frontend PWA]
B --> C[Backend API]
C --> D[Database]
C --> E[AI/ML Model]
E --> F[Crop Insights]
graph TD
%% Entry Points
F_User[Farmer: Mobile PWA]
A_User[Admin: Web Dashboard]
subgraph Farmer_Experience [Farmer Dashboard]
F_Scan[AI Disease Detection]
F_Weather[Smart Weather Advisor]
F_IoT[IoT Sensor Analytics]
F_Comm[Community Disease Map]
F_Voice[AI Voice Assistant]
F_Gov[Government Schemes]
end
subgraph Admin_Experience [Admin Command Center]
A_UserMgmt[User & Farm Audit]
A_Heatmap[Regional Outbreak Trends]
A_Realtime[Live System Health]
A_Reports[CSV/PDF Report Generation]
end
subgraph Intelligence_Layer [AI & Decision Logic]
TF[TensorFlow.js: Vision Model]
Gemini[Google Gemini: LLM Advisor]
Advisory[Smart Irrigation Engine]
end
subgraph Backend_Services [FastAPI Middleware]
Core_API[Core Service Hub]
Auth[JWT Security Layer]
IoT_Hub[Sensor Data Pipeline]
end
subgraph Data_Persistence [Cloud Layer]
Supabase[(Supabase: PostgreSQL)]
C_Storage[Cloud Image Storage]
end
%% Relationships
F_User --> Farmer_Experience
A_User --> Admin_Experience
Farmer_Experience --> Core_API
Admin_Experience --> Core_API
Core_API --> Auth
Core_API --> Intelligence_Layer
Core_API --> IoT_Hub
Intelligence_Layer --> TF
Intelligence_Layer --> Gemini
IoT_Hub --> Supabase
Core_API --> Supabase
Core_API --> C_Storage
Detailed technical setup and configuration guides are available in the /docs folder:
frontend-setup.md– Installation, local development, and build steps for the Next.js app.backend-setup.md– Python environment setup, API documentation, and server execution.local-setup.md– A comprehensive step-by-step guide to running the entire project ecosystem locally.
These documents are designed to minimize onboarding time for new contributors.
The system relies on several security keys and configuration strings. All required variables are documented in:
agritech-app/.env.local.example(Frontend)backend/.env.sample(Backend)
Action: Copy these .example files to .env or .env.local and substitute your own configuration values before starting the servers.
To maintain code quality and system integrity, all contributors should follow these standards:
- Version Control: Regularly push staged code to GitHub with descriptive commit messages.
- Documentation First: Update relevant
.mdfiles in the/docsfolder whenever a new feature or architectural change is introduced. - Standardization: Adhere to the established directory structure and linting rules (standard JS/TS and Python PEP 8).
- Node.js (v18+)
- Python (v3.9+)
- Supabase Account
- Gemini API Key
cd agritech-app
npm install
npm run devcd backend
pip install -r requirements.txt
python main.pyCreate .env files in both directories following the provided .env.sample templates.
- Market Price (Mandi) Link: Real-time crop pricing.
- Drone Integration: Automated field monitoring.
- Fertilizer Calculator: Precision calculation based on soil health cards.
- Offline Mode: Enhanced IndexedDB support for zero-connectivity zones.
Transforming the soil, one byte at a time.