Inspiration

Every day, millions of Latin American families face the same question: what do we cook today? The answer is usually repetition, improvisation, or skipping home cooking altogether. Existing meal planners are generic, English-first, and disconnected from how Latin American families actually shop and cook.

I built CaseroAI because I live this problem every day — as a shift worker and father of two kids who resist new foods.

What it does

CaseroAI generates personalized 7-day meal plans using a constraint-satisfaction algorithm that considers:

  • Family size and children's food acceptance scores
  • Cooking time available per day (weekdays vs weekends)
  • Ingredient and category variety (no pasta twice in a row)
  • User preferences and foods to avoid
  • Gradual introduction of new dishes for picky eaters

Claude AI acts as a conversational chef — explaining the plan, giving cooking tips in natural language, and helping families introduce new dishes gradually.

The app also generates a consolidated shopping list and suggests optimal times to visit neighborhood stores based on peak hours.

How I built it

  • React + Vite for the frontend
  • Firebase Firestore for real-time data sync across devices
  • Anthropic Claude API as the AI conversational layer
  • Custom constraint-based algorithm for meal planning logic (not just a chatbot wrapper)
  • Vercel for deployment

Challenges

  • Designing a scoring algorithm that balances variety, kid-friendliness, and cooking time simultaneously
  • Keeping Claude API costs minimal while maximizing the value of each call
  • Making the app genuinely useful for Latin American cuisine context, not just translating an existing concept

What I learned

That domain expertise matters more than coding skills. Understanding the real problem — how families actually plan and cook — was more valuable than any technical decision.

What's next

  • Expanded recipe database with Argentine and Latin American dishes
  • Anticipatory cooking tips (e.g., "make pizza dough tonight for tomorrow's dinner")
  • Seasonal and cultural calendar awareness (no locro in summer)
  • Shopping route optimization based on neighborhood store locations

Built With

Share this project:

Updates