About the Project
GreenReceipt started with a very simple thought.
I care about the environment… but when I shop for groceries, I don’t think about packaging at all.
I think about price. I think about what I’m cooking. I think about what I’m craving.
And then I come home with plastic trays, wrappers, bags — and it all becomes waste. That disconnect bothered me. So I built something small. Practical. Grounded in real life.
GreenReceipt takes a grocery receipt and turns it into awareness. Not guilt. Not judgment. Just awareness.
What Inspired Me
Environmental change often feels abstract. But grocery shopping? That’s real. That’s weekly. That’s personal. I wanted to bridge that gap — between everyday grocery shopping and environmental impact. Because small decisions, repeated often, matter more than dramatic one-time actions.
How I Built It (And How I Used AI)
GreenReceipt is built using:
Python Streamlit for the interface SQLite for storing streaks and scores And an AI model for intelligent analysis
Here’s exactly where AI is used:
1. Receipt Understanding
Receipts are messy. They contain abbreviations like:
“ORG EGGS LRG” “BRD WHT 675G” “CHK BRST BNLS”
AI interprets this unstructured text and converts it into clean, readable grocery items. Without AI, this step would require rigid rules that would break easily.
2. Packaging Classification
After extracting items, AI helps determine likely packaging types based on product category.
For example: Eggs → typically carton or plastic Bread → usually plastic Meat → tray + plastic wrap
AI provides contextual reasoning instead of fixed assumptions.
3. GreenScore Calculation
The GreenScore is not randomly assigned. AI evaluates: Item category Confirmed packaging selection (paper vs plastic, etc.) Sustainability heuristics
It then generates: A structured environmental score An explanation of what influenced that score This makes the score dynamic and contextual — not hardcoded.
4. Item-by-Item Suggestions
AI generates personalized packaging-focused suggestions, such as:
Swap to paper packaging Look for refill formats Reduce single-use plastics Recycling guidance when swapping isn’t possible
These suggestions are contextual to the specific receipt.
5. Recycling Guidance
If users cannot swap packaging, AI provides practical disposal guidance: Green bin (organic waste) Blue bin (recyclables) General waste
This keeps the solution realistic — not idealistic.
In short, AI powers:
Interpretation Classification Scoring Recommendation
The user experience feels simple. Underneath, AI is doing structured reasoning.
Challenges I Faced
The hardest part was control. AI can be powerful — but unpredictable.
I faced challenges like:
Making sure outputs were structured and consistent Preventing over-confident or exaggerated suggestions Keeping recommendations packaging-focused Designing a scoring system that felt fair and transparent Ensuring streaks calculated correctly without breaking flow
Another major challenge was balance. I didn’t want the app to feel judgmental. But I also didn’t want it to feel meaningless. Finding that middle ground took multiple iterations.
What I Learned
AI works best when solving a very specific friction point.
Environmental tools must feel supportive, not preachy. Small behavioral nudges (like streaks) can drive consistency. Intelligence in the background should not complicate the front-end experience.
GreenReceipt isn’t trying to solve climate change overnight.
It’s trying to build awareness — one receipt at a time.
Log in or sign up for Devpost to join the conversation.