Farm.ai — Agricultural Intelligence Platform

Know your land before you plant a single seed.

Farm.ai is a precision agriculture platform that analyzes soil composition, climate history, vegetation health, and crop compatibility for any property using USDA data, satellite imagery, and AI-powered recommendations.

🌾 Features

Core Analysis

• Soil Intelligence — USDA SSURGO soil classification, pH levels, organic matter, drainage
• Climate Profiling — 30-year historical normals, growing season length, hardiness zones, frost dates
• Vegetation Health — Sentinel-2 NDVI (Normalized Difference Vegetation Index) analysis
• Crop Compatibility Matrix — Suitability scores for 15+ crops with yield projections
• Economic Projections — 3 revenue scenarios (Max Yield, Low Maintenance, Pest Resistant)

Advanced Features (v2.0)

✨ All NEW in this release:

1. Perplexity-Style Dynamic Citations

• Inline glowing citation pills [1] [2] [3] next to every data point
• Expandable "Sources & Factuality" footer revealing all 7 government/scientific databases
• Transparency: See exactly which agencies powered each analysis

2. High-Yield Property Recommendations

• MapLibre integration showing 3-4 nearby parcels with similar soil/yield profiles
• AI-recommended parcels displayed as warm orange polygons with yield match %, soil match %
• Click popup for quick comparison, "Run Full Analysis" button for deep dives
• Auto-fetches recommendations when map loads

3. AI Agronomist Chat Interface

• Floating glassmorphism chat button (bottom-right, always accessible)
• Context-aware AI responses powered by Gemini 2.5 Flash
• Full property analysis bundled into every query for hyper-personalized answers
• Typing indicator, suggestion chips, message history persistence
• Framer Motion animations: spring drawer, slide-up messages, smooth transitions

4. AAA-Grade 3D Farm Visualization

• Real-time sky dome with sun position tracking (SunCalc integration)
• Volumetric cloud systems (clear: 3 clouds, rain/snow: 8 clouds)
• Dynamic particle systems: rain (3000 instances), snow (800 sparkles)
• Procedurally-generated grass-blade normal maps & roughness textures
• Weather-responsive materials (clear→wet→snowy)
• Cinematic post-processing: Depth of Field, Bloom, Vignette, Hue-Saturation boost
• Interactive OrbitControls with responsive terrain

---

🏗️ Tech Stack

Layer	Technology
Frontend	Next.js 16.1 (Turbopack), React 18, TypeScript, Tailwind CSS
3D Rendering	React Three Fiber, Three.js, @react-three/drei, @react-three/postprocessing
State Management	Zustand (lightweight, performant)
UI Animations	Framer Motion (spring, AnimatePresence)
Map	MapLibre GL (open-source, privacy-first)
Backend	FastAPI (Python), Google Gemini API
Data Sources	USDA SSURGO, Open-Meteo, Copernicus Sentinel-2, NWS, NOAA

---

📁 Repository Structure

farm-ai/
├── plantai/                          # Next.js frontend
│   ├── src/
│   │   ├── app/
│   │   │   ├── layout.tsx           # Root layout
│   │   │   ├── page.tsx             # Landing page
│   │   │   ├── map/page.tsx         # Drawing interface
│   │   │   ├── analysis/[id]/       # Main dashboard (7 tabs)
│   │   │   └── farm/[id]/           # 3D visualization
│   │   ├── components/
│   │   │   ├── chat/
│   │   │   │   └── AgronomistChat.tsx     # AI chat drawer
│   │   │   ├── farm/
│   │   │   │   ├── Atmosphere.tsx         # Sky, sun, fog
│   │   │   │   ├── WeatherSystem.tsx      # Rain/snow
│   │   │   │   ├── PostProcessing.tsx     # Effects
│   │   │   │   ├── TerrainMesh.tsx        # Textures
│   │   │   │   └── FarmScene.tsx          # 3D orchestrator
│   │   │   ├── map/
│   │   │   │   └── MapCanvas.tsx          # MapLibre + recommendations
│   │   │   └── ui/
│   │   │       ├── CitationPill.tsx       # Citation pills
│   │   │       └── SourcesFooter.tsx      # Sources footer
│   │   └── lib/
│   │       ├── store.ts              # Zustand state
│   │       └── apiClient.ts          # API calls
│   └── package.json
│
└── plantai-backend/                  # FastAPI backend
    ├── main.py                       # API endpoints
    ├── services.py                   # Data fetching
    └── requirements.txt              # Dependencies

---

🚀 Quick Start

Prerequisites

• Node.js 18+
• Python 3.9+
• Git

Installation

# Clone
git clone https://github.com/YOUR_USERNAME/farm-ai.git
cd farm-ai

# Frontend
cd plantai
npm install

# Backend
cd ../plantai-backend
pip install -r requirements.txt

Environment Setup

Frontend (.env.local in plantai/)

NEXT_PUBLIC_API_URL=http://localhost:8000
NEXT_PUBLIC_MAPTILER_KEY=YOUR_MAPTILER_KEY  # Optional

Backend (.env in plantai-backend/)

GOOGLE_API_KEY=YOUR_GEMINI_API_KEY

Run Locally

Terminal 1:

cd plantai
npm run dev
# http://localhost:3000

Terminal 2:

cd plantai-backend
python main.py
# http://localhost:8000/docs

---

📊 User Workflow

1. Enter Address — Search bar with Nominatim geocoding
2. Draw Property — Click points on map to outline field
3. Analyze — Backend fetches USDA, Open-Meteo, Sentinel-2 data in parallel
4. Explore Dashboard — 7 tabs: Overview, Soil, Climate, Crops, Economics, Agents, Plan
5. View Citations — Click pill [1] to see sources in expandable footer
6. Chat with AI — Click green button, ask context-aware questions
7. 3D Visualization — Click "3D View" for interactive farm scene

---

🔄 NEW API Endpoints

Chat (Context-Aware AI)

POST /api/chat
{
  "message": "Why tomatoes?",
  "history": [...],
  "context": { "soilData": {...}, "cropMatrix": [...] }
}
→ {"reply": "Your Cecil sandy loam with 6.2 pH..."}

Recommendations (High-Yield Parcels)

POST /api/recommendations
{ "lat": 40.79, "lng": -77.86 }
→ GeoJSON FeatureCollection with 4 nearby parcels

---

🎨 Design Highlights

• Glassmorphism — Blur, semi-transparency, subtle borders
• Dark Mode — #0a0d0a background, #4ade80 accent green
• Responsive — Mobile-first Tailwind CSS
• Smooth Animations — Framer Motion spring easing
• Accessibility — ARIA labels, semantic HTML

---

📦 Key Dependencies

Frontend:

• @react-three/fiber — R3F for 3D
• @react-three/postprocessing — Post-process effects
• maplibre-gl — Interactive maps
• framer-motion — Smooth animations
• zustand — Lightweight state

Backend:

• fastapi — Modern Python API
• google-genai — Gemini AI integration
• pydantic — Data validation
• shapely — Geospatial math

---

🧪 Testing

# Build verification
cd plantai && npm run build

# API test
curl -X POST http://localhost:8000/api/chat \
  -H "Content-Type: application/json" \
  -d '{"message":"Crops?","history":[],"context":{"soilData":{"name":"Loam","ph":6.5}}}'

---

🔐 Security Notes

⚠️ Demo API key in repo — Rotate before production!

Pre-deployment:

• Rotate all API keys
• Enable HTTPS
• Add authentication (OAuth2)
• Configure CORS origins
• Enable rate limiting
• Audit Gemini prompts

---

📈 Roadmap

• Phase 2: IoT soil sensors, subscription tiers, mobile app
• Phase 3: Drone imagery, real-time alerts, insurance optimization

---

📄 License

MIT License — See LICENSE file

---

👥 Contributing

1. Fork the repo
2. Create feature branch (git checkout -b feature/amazing)
3. Commit (git commit -m 'Add feature')
4. Push (git push origin feature/amazing)
5. Open PR

---

📧 Support

• Issues: GitHub Issues
• Email: contact@farm.ai

---

Made with 🌾 for farmers who think ahead.