Supermeal

What should I eat today?

That one question is where Supermeal started. Between work, training and everything else a day throws at you, there's rarely time to think through what healthy eating should actually look like for you. So you compromise — and then you're annoyed that your good intentions slipped again and that your fitness and wellbeing goals are barely moving, if at all. I know that feeling all too well, and at some point I wanted a way out of it.

So I built Supermeal. The app is meant to take that tedious decision off your plate — what to eat to stay on track — without complicated processes and without feeling like another chore.

Setup takes just a few seconds. From there the app adapts to your goals and to your eating and training habits, and gives you concrete recommendations for breaking bad habits and reaching your goals faster. The more you share with it, the better it gets to know you, and the more you get out of it. And if you connect your Apple Watch or Garmin, you can stop thinking about macros and calories altogether — that part just takes care of itself.

The tech stack

There's a fair amount of engineering behind Supermeal, and honestly that was the part I enjoyed most. I didn't want to click an app together — I wanted to build something properly thought through that grows with the user.

The app itself is a native SwiftUI app for iOS. It pulls your health data straight from your Apple Watch or Garmin via HealthKit, so everything important flows in automatically and you never have to type it in by hand.

All the logic behind it runs on a dedicated Food API that I built with Next.js and host on Vercel. It mattered to me that the app stays lean and that the heavy lifting happens in the background, where sensitive data is properly protected. Meals and ingredients live in MongoDB, which is a great fit — recipes and nutritional values rarely squeeze into a rigid schema. For auth, functions, messaging and the store I use Firebase, so I can focus on the product instead of on infrastructure.

The part I'm proudest of is the automated AI pipelines. They build and improve the meal database more or less on their own, so I don't have to maintain all of it manually. On top of that, background jobs run regularly to translate your profile into vectors and match it against the meals. Put simply: that's how the app learns which dishes actually suit you — your goals, your training, your everyday life. Because all of it is precomputed in the background, the app itself stays fast and light on its feet.

A passion project

For me, every single hour was worth it. Working on this felt like being a kid again, completely absorbed in something that really matters to you. I got to bring all my passions together — sport, nutrition, fitness, data, design and engineering — in one product, and I hope you can tell there's a person behind this app who poured their heart into it, not a template.

This first version is the beginning of something that's far from finished. It's currently in review with Apple via TestFlight, which means it's close to launching on the App Store. It feels like that moment just before you let go of something you've spent a long time building with a lot of love.