Projects
Culina
AI-native food and meal-planning app — know what you have, what you can cook, and keep a plan that adapts when you deviate or when things are about to go bad.
Context
Meal planning apps exist, but they don't connect to what's actually in your kitchen. You end up with plans that ignore what you already have, food that goes bad before you use it, and no easy way to adapt when life gets in the way. We wanted one place where your pantry, your meals, and your plan stay in sync — without manual tracking.
What I'm building
I'm building Culina with a small team. I own strategy, product, and engineering; day to day that centers on the AI integration and the interaction model:
- Photo-based intake: Snap a photo of your shopping; AI extracts items, expiry dates, and quantities.
- Multi-modal interaction: Click, chat, or voice — whichever fits the moment (hands full in the kitchen vs. planning at the table).
- Adaptive meal plans: Plans adjust when you skip a night, eat out, or when something's about to expire.
The goal is an experience that feels purpose-built for how people actually handle food at home, not a generic app with AI bolted on.
Outcome
We're in development: concept is clear, active development, MVP is taking shape. Building toward a product people use daily — pantry tracking, AI-generated meal plans, and adaptive rescheduling.
Key Insights
- AI works best when it's invisible — the value is in the result (a meal plan that makes sense), not in showing off the tech.
- Food is personal. The system has to learn preferences without parroting them back — if you like Mexican food, it shouldn't suggest Mexican every time. Learn and vary, not prescribe.
- The hardest part isn't the AI — it's keeping the data right without making it a chore. If someone shops and doesn't track it, or uses something up without logging it, the whole app breaks. Making that interaction smooth and invisible is the core challenge.
Proof: Product website — concept and direction. Prototype walkthroughs on request.