Inspiration
BestBefore started from a simple but frustrating experience in my own life.
I often found myself discovering food in my fridge that had already passed its expiration date. Every time I threw away unopened groceries, I felt disappointed — not only because I had wasted money, but because perfectly good food had gone unused.
I realized the problem was not that I forgot to buy food, but that I forgot what I already had.
I looked for solutions, but most food tracking apps require too much manual input: entering product names, expiration dates, and categories every time you buy something. That process quickly becomes a burden.
I wanted to create something different: an AI assistant that could remember my groceries for me. That idea became BestBefore.
What it does
BestBefore is an AI-powered pantry assistant that helps users track their groceries and avoid unnecessary food waste.
Users can simply:
- Scan a barcode to identify packaged products.
- Take a photo of food packaging and let AI extract expiration dates.
- Use image recognition to identify fresh food and estimate shelf life.
- Automatically organize groceries into a digital pantry.
- Receive reminders before food expires.
- Ask AI what to cook based on ingredients that need to be used soon.
Instead of manually managing a food inventory, users can simply show BestBefore what they bought and let AI handle the rest.
How we built it
BestBefore was built as an AI-first application where intelligence is at the center of the user experience.
We used:
- OpenAI Vision to understand food images, recognize products, and extract expiration information from packaging.
- OpenAI GPT to reason about food categories, estimate shelf life, answer user questions, and generate personalized meal suggestions.
- Barcode scanning technology to quickly identify packaged products.
- Next.js for the application interface.
- Supabase for data storage and user pantry management.
- Vercel for deployment.
The key idea was to reduce manual input as much as possible and make food tracking feel natural through AI interaction.
Challenges we ran into
One of the biggest challenges was that food information is not always structured.
Expiration dates appear in different formats, locations, and languages depending on the product. Some products have clear labels, while fresh ingredients such as fruits and vegetables have no expiration date at all.
We needed to design AI workflows that could handle uncertainty — extracting available information when possible and making reasonable estimates when information was missing.
Another challenge was creating an experience that users would actually continue using. We focused on making every interaction simple: take a photo, scan a product, and let AI do the work.
Accomplishments that we're proud of
We are proud of creating a practical AI experience that solves a small but meaningful everyday problem.
With BestBefore, we:
- Built an AI-powered pantry assistant instead of another manual tracking tool.
- Combined computer vision, barcode recognition, and AI reasoning into one seamless workflow.
- Reduced the effort required to manage groceries.
- Created a system that not only remembers food, but helps users make decisions about what to eat next.
Most importantly, we transformed a frustrating daily experience into something simple, helpful, and sustainable.
What we learned
Building BestBefore showed us that the biggest opportunity for AI is not only automating complex tasks, but removing small everyday frustrations.
We learned that successful AI products need more than accurate recognition — they need to understand context and provide meaningful actions.
Knowing that a product expires soon is useful. Helping someone decide what to cook, what to use first, and how to reduce waste is much more valuable.
What's next for BestBefore
We want BestBefore to become a complete AI food companion for everyday life.
Future improvements include:
- Receipt scanning to automatically add groceries after shopping.
- Shared family pantries.
- Personalized meal recommendations based on available ingredients.
- Smart shopping lists that prevent duplicate purchases.
- Integration with smart refrigerators.
- Insights showing money saved and food waste reduced.
Our vision is to build an AI assistant that helps people make smarter food decisions effortlessly — one meal at a time.
Built With
- api
- barcode
- gpt-5
- next.js
- openai
- responses
- scanner
- supabase
- typescript
- vision
- zxing
Log in or sign up for Devpost to join the conversation.