Bellynote

The problem

People with food sensitivities, IBS or stomach problems that are hard to pin down are usually told to keep a food journal. In practice journals are too much work to keep up with, and even when people do log, spotting which foods line up with bloating, headaches or energy dips is hard.

Bellynote makes logging take seconds and does the pattern-finding for you. It presents patterns as leads to test, not diagnoses.

What I built

• AI logging. Snap a meal, a nutrition label or a barcode. Supabase Edge Functions call OpenAI's vision models to identify ingredients, look up barcodes and sort foods into 12 groups such as gluten, dairy, nightshades and FODMAPs.

• Symptom correlation. Heatmaps line up foods, food groups, additives and menstrual cycle phases against symptoms over a 4 to 24 hour lookback window, with 0 to 10 severity trends across 7, 30 and 90 days.

• Sage, an AI assistant. Answers questions about your own logs in plain language.

• Native integrations. Camera scanning, read-only Apple Health and Health Connect sync for cycle data, Sign in with Apple and Google, multi-factor auth, and push notifications for logging reminders, streaks and weekly insights.

• Subscriptions. RevenueCat in the mobile app and Stripe on the web.

• Web platform. A Turborepo monorepo with three Next.js 16 apps (the bellynote.app marketing site, a web app and an internal admin dashboard) sharing UI, subscription and Supabase type packages.

• Shipping. EAS builds for the App Store and Google Play, over-the-air updates between releases, and Sentry for crash reporting.