Nurelio

Connect what you eat with how you feel — and improve it.

Comment

Inspiration

I kept feeling off after some meals and could never figure out which ones. I tried keeping notes, but one day doesn't tell you anything. You need to look across weeks, and I couldn't hold that much in my head. Calorie apps didn't help either. They told me how much protein I ate, not why I felt tired afterwards.

What I really wanted was something that would just point it out: "you usually lose energy after this", "bloating tends to show up after that". So I hired the only developer who couldn't say no: my husband! 😁 The pay is terrible, but dinner is included, and now every dinner gets logged. Together we started building the app I'd been wishing for.

What it does

You log a meal with a photo, your voice, a quick text, search or a barcode, and Nurelio fills in calories, macros and micronutrients. Then you tap how you feel: energy, digestion, sleep, mood, cravings, and more.

After a week or two it starts noticing things, like bloating showing up after dairy on 4 of the last 5 days. There's also an organ view on a realistic body figure (heart, liver, gut, kidneys and more), a weekly health score, and a chat where you can ask questions about your own history. It syncs with Apple Health both ways.

How we built it

Nurelio is an iOS app built with React Native and Expo, with RevenueCat handling subscriptions, the paywall and the trial. Everything personal lives in an encrypted database on the phone, and the pattern engine runs there too. A small backend on Cloudflare Workers handles AI food recognition and the coach one request at a time, without storing anyone's diary, and barcode and search results come from real food databases (USDA and Open Food Facts) rather than AI guesses.

Challenges we ran into

The hardest part was making the patterns honest, we had to fine tune probabilities and time intervals to make it work as we expect. Food DB was another problem, especially naming of products in USDA database and missing images for products, so we ran local LLM through the items and added emoji category to each and rewritten names to more human readable.

Accomplishments that we're proud of

Nurelio is live in 136 countries, and it doesn't feel like regular AI calorie tracker. We're proud of our UI and UX, it still has room for improvement, but we think it's better than many. Also, we're focused on privacy, nonjudgmental approach, real results (our calorie goal is adaptive to current progress!) and want to provide real value to users, and to ourselves, its most loyal and most demanding users 😄

What we learned

Price for when the product pays off, not for when you'd like the revenue. Patterns need about two weeks, so we skipped a weekly plan and made the trial 14 days. It cost us revenue during Shipaton, and we'd do it again. And test on real data early: our demo data looked perfect, real life was much messier. And of course, distribution is a new king, with so many apps out there it's much harder to get first users.

What's next for Nurelio

Our goal is simple to say and hard to build: the most complete tool for understanding how food affects your body — not the average person's, yours. The same bowl of oatmeal leaves one person full and focused and sends another into a crash an hour later. We want Nurelio to know which one you are.

That takes better signals. Today it learns from what you log and what Apple Health already knows: sleep, activity, heart rate. Next we want devices that measure your response directly, starting with continuous glucose monitors, so you can see what a meal actually did instead of guessing from how you felt. Eventually, a model personal to you — one that learns how your body tends to react and can tell you before you eat how a meal is likely to land.

Built With

  • astro
  • bun
  • cloudflare-workers
  • drizzle
  • expo.io
  • gemini
  • healthkit
  • hono
  • maestro
  • meilisearch
  • nativewind
  • open-food-facts
  • openrouter
  • posthog
  • react-native
  • reanimated
  • revenuecat
  • sentry
  • sqlcipher
  • sqlite
  • tailwind
  • typescript
  • usda-fooddata-central
  • vercel-ai-sdk
  • vitest
Share this project:

Updates

Submission history