AI health assistant
BetterDiet

Situation
The problem
Tracking nutrition and workouts requires constant data entry. Standard apps are cumbersome — users have to know exact calorie values and fill in forms.
Solution
How we got there
A personalized AI assistant that understands everyday language: the user describes what they ate or how they trained, and the AI understands, calculates, and adapts to the user's context and goals.
Result
What changed
Nutrition and workout tracking through natural conversation. Long-term user context lets the assistant give accurate recommendations.
Tech stack
Architecture and build
Next.js with Claude AI on long-term context and user history. Google sign-in. Data stored securely.
| Frontend | Next.js · React |
| AI | Claude API with long-term user context |
| Auth | Google OAuth |
| Data | User's nutrition and workout history |
AI / Automation
AI agent
Claude API with long-term context — the agent knows the user's goals, history, and current state. Recommendations adapt to how the person feels and recovers, not just to the numbers.
Numbers
Results
Data entry
Simplified
Forms and numbers → Natural conversation
Context
Persistent
Overwritten → Long-term
Recommendation accuracy
Personalized
Generic tables → Personalized