Scan any barcode. Know its planet cost. Instantly.
GreenLens turns any product barcode into an environmental report card in under 5 seconds. Scan or type a barcode and get an A–F eco grade, three sustainability sub-scores (CO₂, water, packaging), the specific reasons behind the rating, and greener product categories to look for — all powered by a two-stage AI pipeline analyzing real data from the Open Food Facts database.
No signup. No install. Works on any smartphone or desktop browser.
GreenLens is a mobile-first web app that turns any product barcode into an environmental report card. Scan or type a barcode, and within seconds you get an A–F eco grade, three sustainability sub-scores, the reasons behind the rating, and greener product alternatives — all powered by AI analyzing real product data from the Open Food Facts database.
No signup. No install. Works in any modern browser.
Consumers worldwide make over 500 billion grocery purchasing decisions every year with virtually zero environmental context. People want to make sustainable choices, but existing tools are fragmented, slow, or require an account — none of them work at the speed of a checkout line.
Certification labels are scattered across dozens of schemes. Environmental databases are behind paywalls. Scanning apps focus on nutrition, not ecology. The result: sustainability is something people care about in the abstract, but abandon the moment they're standing in a supermarket aisle.
| Challenge | Status quo | GreenLens |
|---|---|---|
| Getting eco data fast | Search multiple sites manually | Scan once, results in seconds |
| Understanding environmental impact | Dense certification jargon | A–F grade anyone can read |
| Knowing what to buy instead | No in-context suggestions | AI-generated greener swaps, instantly |
| Access to product data | Fragmented, paid databases | Open Food Facts — 3M+ products worldwide |
| Nuanced AI scoring | Keyword-based rules engines | LLM reasoning over ingredients, NOVA group, packaging, and food miles |
graph TD
A[User scans barcode] --> B[Open Food Facts REST API]
B --> C[formatProduct — normalize response]
C --> D{Product found?}
D -- No --> E[ErrorCard: product not found]
D -- Yes --> F[Serverless Function /api/score]
F --> G[Groq: Scorer Agent]
G --> H[Groq: Validator Agent]
H --> I{Score valid?}
I -- No --> J[Fallback: use Stage 1 score]
I -- Yes --> K[Animated Result Page]
K --> L[GradeCard + ScoreBars + Reasons + Greener Swaps]
L --> M[useScanHistory: persist to localStorage]
M --> N[scanCache: 7-day barcode cache]
Security note: The Groq API key is never exposed to the client. All AI calls are proxied through a Vercel serverless function ().
Two-stage AI pipeline with self-validation: A scorer agent generates the initial environmental assessment. A second validator agent then audits the output for logical consistency — enforcing domain rules (ultra-processed NOVA 4 products cannot score highly on CO₂; beef products must have a low water score) and aligning grades with Open Food Facts certified ecoscore data when available. If the validator detects an inconsistency, it returns a corrected score. If the validator itself fails, the system gracefully falls back to the Stage 1 result. This produces significantly more reliable scores than single-pass LLM scoring.
- Barcode scanning — Type a barcode or use your device's camera with native
BarcodeDetector(EAN-13, EAN-8, UPC-A, UPC-E, Code 128) - AI eco scoring — Groq-powered LLM analyzes product ingredients, processing level (NOVA group), packaging materials, and supply chain signals to produce a nuanced grade
- A–F grade + three sub-scores — Overall grade plus individual scores for CO₂ footprint, water usage, and packaging, each 0–100
- Specific reasons — Three data-driven sentences explaining exactly why the product received its grade
- Greener swaps — Up to three alternative product suggestions with their estimated grades
- Scan history — Last 10 scans stored locally; tap any entry to revisit its results without re-scanning
- Animated result cards — Grade letter springs into view; score bars fill progressively with color-coded thresholds
- Honest limitations — An always-visible disclaimer that scores are AI estimates, not certified assessments
User scans barcode
│
▼
Open Food Facts API ← 3 million products
(product name, brand,
ingredients, NOVA group,
packaging, ecoscore_grade)
│
▼
Groq AI (LLM prompt) ← ingredients + NOVA + packaging + food miles
│
▼
Structured JSON response
{
grade: "B",
overall_score: 74,
co2_score: 80,
water_score: 65,
packaging_score: 72,
reasons: [...],
greener_swaps: [...]
}
│
▼
Animated result page
The AI prompt explicitly asks the model to consider:
- CO₂ score — Food miles, processing level (NOVA 1 = unprocessed, 4 = ultra-processed), animal vs. plant-based origin
- Water score — Water-intensive crops (almonds, beef, avocado score low), production water use
- Packaging score — Glass and cardboard score high; single-use plastic scores low
| Grade | Label | Score Range |
|---|---|---|
| A | Excellent | 85 – 100 |
| B | Good | 70 – 84 |
| C | Average | 50 – 69 |
| D | Poor | 30 – 49 |
| E | Very Poor | 15 – 29 |
| F | Harmful | 0 – 14 |
Open the deployed URL in any modern browser on your phone or desktop. No installation required.
# 1. Clone the repo
git clone https://github.com/code-withkrishna/greenlens.git
cd greenlens
# 2. Install dependencies
npm install
# 3. Install Vercel CLI (required to run serverless functions locally)
npm i -g vercel
# 4. Add your Groq API key
cp .env.example .env
# Edit .env and set GROQ_API_KEY=your_key_here
# Get a free key at https://console.groq.com → API Keys → Create Key
# 5. Start the dev server (Vite + serverless functions together)
vercel devOpen http://localhost:3000 in your browser.
Note:
npm run devstarts the Vite dev server only and cannot run the/api/scoreserverless function. Usevercel devfor a full local environment with AI scoring enabled.
npm run build
# Output is in /dist — deploy to any static host (Vercel, Netlify, Cloudflare Pages)| Variable | Required | Description |
|---|---|---|
VITE_GROQ_API_KEY |
Yes | Your Groq API key. Free tier is sufficient for development. Get one at console.groq.com |
| Layer | Technology |
|---|---|
| UI framework | React 19 |
| Routing | React Router 7 |
| Styling | Tailwind CSS |
| Fonts | DM Sans, Playfair Display |
| Build tool | Vite |
| AI scoring | Groq API (LLM inference) |
| Product data | Open Food Facts REST API |
| Camera scanning | Native BarcodeDetector Web API |
| HTTP client | Axios |
| State persistence | localStorage (scan history) |
| Feature | GreenLens | Yuka | Open Food Facts app | Google Lens |
|---|---|---|---|---|
| Environmental grade | ✅ | ❌ (health only) | Partial | ❌ |
| AI-generated reasons | ✅ | ❌ | ❌ | ❌ |
| Greener swap suggestions | ✅ | ❌ | ❌ | ❌ |
| No account required | ✅ | ❌ | ✅ | ✅ |
| Open source | ✅ | ❌ | ✅ | ❌ |
| Sub-scores (CO₂/water/packaging) | ✅ | ❌ | ✅ (ecoscore) | ❌ |
| Works in browser (no install) | ✅ | ❌ | ❌ | ❌ |
- Open Food Facts — Open, crowd-sourced product database with 3M+ entries including ingredients, packaging tags, NOVA group, and ecoscore
- Groq + LLM — Ultra-fast inference for structured environmental scoring
greenlens/
├── index.html # Entry point
├── .env.example # Environment variable template
├── src/
│ ├── main.jsx # React root
│ ├── App.jsx # Router setup (/, /result, /about)
│ ├── pages/
│ │ ├── Home.jsx # Barcode input + scan history
│ │ ├── Result.jsx # Grade card + sub-scores + swaps
│ │ └── About.jsx # Methodology + limitations
│ ├── components/
│ │ ├── BarcodeForm.jsx # Text input + camera trigger
│ │ ├── CameraScanner.jsx # BarcodeDetector integration
│ │ ├── GradeCard.jsx # Animated A–F display
│ │ ├── ScoreBar.jsx # Animated score bars
│ │ ├── HistoryList.jsx # Recent scans
│ │ └── ErrorCard.jsx # Error states
│ ├── hooks/
│ │ ├── useCamera.js # BarcodeDetector + getUserMedia
│ │ └── useHistory.js # localStorage scan history
│ └── api/
│ ├── openFoodFacts.js # Product lookup
│ └── score.js # Groq AI scoring prompt
└── dist/ # Production build
GreenLens scores are AI estimates based on available product data — not certified environmental assessments. Data quality varies by product (some entries on Open Food Facts are incomplete). Use the scores as a directional guide, not a definitive rating.
- PWA support with offline caching
- Share result as a card (Web Share API)
- Aggregate household footprint tracker
- Certified ecoscore crosscheck when available
- Multi-language support
Contributions are welcome. Please open an issue first to discuss what you'd like to change. Pull requests should include a clear description of the change and its motivation.
MIT — see LICENSE for details.