CartRadar

Inspiration

My husband and I constantly have conversations like, “We’re out of eggs” or “We need to buy milk.” But when we actually get to the grocery store, we forget half the things we talked about.

We tried maintaining a shared grocery list, but that didn’t work either. Most of these conversations happen while we’re cooking or doing something else. We don’t always have our phones nearby, and we tell ourselves we’ll add things later. But we never do!

That’s what inspired CartRadar. What if our everyday conversations could automatically become a grocery list, without us having to do anything?

What it does

CartRadar is an AI-powered grocery shopping assistant that turns everyday conversations into smart shopping lists.

Using conversations captured through Bee, CartRadar automatically identifies grocery items, organizes them into a shared shopping list, and helps users remember what to buy.

• Context-Aware Shopping Intelligence: Uses OpenAI to understand purchase intent, conversational corrections, and completed purchases across Bee conversations, automatically adding, updating, deduplicating, or marking grocery items as purchased without manual intervention.

It also provides location-based reminders when users are near grocery stores, so they don’t forget what they need.

Just talk about what you need. CartRadar remembers it for you.

How we built it

We built CartRadar using:

• Bee API and CLI: Retrieve conversations captured through the Bee wearable.
• OpenAI: Extract grocery items from natural conversations and convert them into structured shopping lists.
• React Native and TypeScript: Build the mobile-friendly application.
• Supabase: Handle Google authentication, database storage, and shared shopping lists.
• Location Services: Detect nearby grocery stores and trigger shopping reminders.
• Vercel: Deploy the application for easy access across devices.

The workflow is simple:

Everyday conversation → Bee → AI extraction → Shared grocery list → Store reminder

Challenges we ran into

One of our biggest challenges was extracting grocery items accurately from casual conversations. People don’t always say, “Add eggs to my grocery list.” Sometimes it’s just, “Oh, we’re running low on eggs.”

Another challenge was implementing location-based notifications across different environments. Browser limitations and iOS development requirements made testing notifications more complicated than expected.

We also had to figure out how to integrate Bee conversations into our application and make the experience as seamless as possible.

Accomplishments that we're proud of

• Built an end-to-end application that transforms everyday conversations into actionable grocery lists.
• Achieved 98% accuracy in extracting grocery items during testing.
• Integrated Bee with an AI-powered extraction pipeline.
• Implemented Google authentication and persistent shopping lists.
• Deployed a mobile-friendly application accessible through a browser.

We’re especially proud that CartRadar solves a problem we personally experience every week.

What we learned

We learned how to integrate wearable AI with a real-world application, process unstructured conversations using LLMs, and build a complete application using React Native, OpenAI, and Supabase.

We also learned about the challenges of location-based notifications, mobile deployment, and integrating external APIs.

Most importantly, we realized that AI can be incredibly useful when it removes small, repetitive tasks from everyday life.

What's next for CartRadar

• Improving grocery item categorization and duplicate detection.
• Adding smarter store recommendations based on shopping lists.
• Making shared shopping lists update in real time for families and partners.

Our goal is simple: Never forget groceries again, even if you forget to make a list.

Built With

• bee-api
• bee-cli
• geolocation-api
• google-oauth
• openai-api
• react-native
• supabase
• typescript