Project Go
AgentathonX 2026 — India's First Online Agentathon
Built solo by Srishti Suman Gupta, BTech CSE Year 1
India generates 62 million tonnes of waste every year.
80% of it ends up in open dumps or rivers.
Most Indians simply don't know which bin to use — not because they don't care, but because no one ever showed them how.
Project Go fixes that with one photo.
Project Go is an AI-powered waste classification agent. You photograph any piece of waste, and the AI:
- Identifies exactly what the item is
- Classifies it into the correct waste category (wet, dry, hazardous, e-waste, sanitary)
- Tells you which bin to use and how to dispose of it
- Calculates the real environmental impact of recycling it — carbon saved, energy saved, wildlife protected
- Finds the 3 nearest recycling centers to you
- Rewards you with EcoCoins redeemable for real-world impact (plant a tree, protect sea turtles, clean ocean plastic)
- Tracks your real money earned from selling scrap to the kabadiwala
- Educates you on how your actions connect to the UN's 17 Sustainable Development Goals
Project Go uses Gemini 2.5 Flash (via a secure Netlify serverless function) with three specialized agent calls:
| Agent Call | What It Does |
|---|---|
analyze |
Vision model identifies waste item from photo, classifies category, gives disposal instructions |
impact |
Calculates environmental impact — carbon %, energy, water, wildlife effect, fun facts |
centers |
Finds 3 nearest real recycling centers using geolocation |
The API key is never exposed to the frontend — all AI calls go through a Netlify serverless function.
| Layer | Technology |
|---|---|
| Frontend | React (Hooks), Recharts, pure CSS |
| AI | Gemini 2.5 Flash (Vision + Text) |
| Backend | Netlify Serverless Functions (Node.js) |
| Build | Vite |
| Deployment | Netlify (CI/CD via GitHub) |
| Geolocation | Browser Geolocation API + OpenStreetMap Nominatim |
| Storage | localStorage (per-user, no backend needed) |
| Fonts | Fraunces (serif) + Outfit (sans) via Google Fonts |
- Upload photo or use live camera (back-facing, flippable)
- Scanning line animation for camera mode
- Real-time AI classification with confidence score
- Bin recommendation (Green / Blue / Red / E-Waste Center / Black)
- Environmental impact panel with carbon %, energy saved, wildlife impact
- Nearest recycling centers with Google Maps links
- Log real money earned from selling scrap
- 12 preset items with typical kabadiwala rates (plastic, newspaper, iron, copper, etc.)
- Auto-calculates suggested earnings with "Use this ↗" shortcut
- Charts of earnings over time
- Full trip history
- 10 real-world rewards redeemable with EcoCoins
- Plant a real tree (SankalpTaru NGO)
- Fund sea turtle protection (Wildlife Trust of India)
- Solar lamp donation for rural families
- Eco certificates, discount coupons
- Covers 11 of the UN's 17 Sustainable Development Goals
- Tap any SDG to learn how your recycling actions contribute
- Educational, beautiful, interactive
- Level system (EcoCoins → Levels)
- Calendar view of scan history
- Charts: waste type breakdown, EcoCoins by category, carbon over time
- 7 unlockable badges (First Scan, Green Hero, EcoWarrior, etc.)
- Animated G🌍 logo cycling: Go Green → Go Clean → Go Smart → Go Local → Go Zero → Go Earth
- Full dark/light mode toggle
- Desktop (3-column sidebar layout) + Mobile (bottom tab bar) responsive
- Floating leaf animations, orb backgrounds, glass cards
# Clone the repo
git clone https://github.com/Srishti-Gupta74/project-go.git
cd project-go/wastewise
# Install dependencies
npm install
# Set up environment
# Create netlify/functions/.env or set in Netlify dashboard:
# GEMINI_API_KEY=your_key_here
# Run dev server
npm run dev
# For full AI features, run with Netlify CLI:
npm install -g netlify-cli
netlify devproject-go/
├── netlify.toml # Netlify build + redirect config
└── wastewise/
├── src/
│ ├── App.jsx # Entire React app
│ ├── App.css # Global styles & animations
│ ├── main.jsx # React entry point
│ └── index.css # Base CSS
├── netlify/
│ └── functions/
│ └── ai.js # Serverless AI proxy (Gemini)
├── index.html # HTML entry
├── package.json
└── vite.config.js
Project Go directly contributes to 11 of the 17 UN Sustainable Development Goals:
SDG 1 · SDG 3 · SDG 6 · SDG 7 · SDG 8 · SDG 11 · SDG 12 · SDG 13 · SDG 14 · SDG 15 · SDG 17
Team Prakriti — Srishti Suman Gupta
BTech CSE Year 1 · Solo participant · AgentathonX 2026
"62 million tonnes of waste. Most Indians don't know which bin to use. Project Go fixes that with one photo."