// the find
the-momentum/open-wearables
Self-hosted platform to unify wearable health data through one AI-ready API.
Open Wearables is a self-hosted middleware layer that normalizes health data from Garmin, Polar, Suunto, Apple Health, and others into a single REST API. The target is developers building health apps who don't want to maintain separate OAuth flows, data models, and sync logic for every device vendor. Also works for individuals who want their own data off proprietary cloud platforms.
The repository pattern with a proper data source priority system (when a user has a Garmin watch and an Apple Watch, you can configure which one wins for heart rate) is genuinely useful and something most people building this would skip. The mobile SDKs (iOS Swift, Android Kotlin, Flutter, React Native) for push-based HealthKit/Google Health sync are load-bearing — getting that data off-device is the hard part, and they've shipped it. Celery task architecture cleanly separates concerns: backfill, periodic sync, webhook processing, and provider-specific jobs are all distinct task files rather than one giant sync loop. Docker Compose with no external auth dependencies means the quickstart actually works in under five minutes.
Most of the headline features are labeled 'coming soon' — AI automations, the health assistant, embeddable widgets, and core health data endpoints (sleep, HRV, recovery scores) are not yet shipped. Of the eight cloud providers listed, only Garmin, Polar, and Suunto have confirmed working OAuth + data sync; the rest appear to be partial or planned. The backend was scaffolded from a generic Python AI Kit template, so there's boilerplate infrastructure not designed for health data — no mention of HIPAA/GDPR compliance considerations despite collecting sensitive biometric data. Self-hosting requires Postgres, Redis, and Celery workers, which is real operational weight; there's no documentation on backup strategy, data retention, or what happens when a provider changes their API.