Inspiration
Calorie tracking often feels like a mess, while food and activity data live in separate apps. We wanted one simple place to understand both.
What it does
Wellspent lets users photograph or describe a meal, review its estimated nutrition, and compare calories eaten with real Apple Health activity data.
How we built it
We used Codex with GPT‑5.6 to build the Expo React Native app, TypeScript API, AI meal-analysis flow, USDA nutrition grounding, and HealthKit integration.
Challenges we ran into
Estimating portions from photos was difficult. HealthKit permissions, incomplete data, and keeping nutrition estimates transparent also required careful handling.
Accomplishments that we're proud of
We built a working end-to-end experience with editable meal estimates, HealthKit syncing, detailed activity views, and personalized progress insights.
What we learned
AI estimates need clear uncertainty and user control. We also learned that measuring model performance is far more useful.
What's next for Wellspent Health
Next, we’ll improve portion accuracy, test with more real-world meals, add secure cloud sync, and bring the experience to Android.
Built With
- 5.6
- codex
- gpt
- javascript
- node.js
- typescript
Log in or sign up for Devpost to join the conversation.