Pulsewear 

Pulsewear makes a heart-rate and motion wearable beloved by serious amateur athletes. The hardware was excellent; the companion app was an afterthought — a glorified data dump that left users to interpret their own training.

With 300,000 active users and a 3.8 App Store rating, they hired us to turn raw telemetry into something that actually coaches.

Adaptive coaching means the app has to feel like it knows you — but most users open it sweaty, mid-session, for two seconds at a time. The intelligence had to disappear into an interface that demanded almost no attention.

We also had to handle the messy reality of wearable data: dropped Bluetooth connections, sensor drift, and workouts that don't fit any tidy category.

We designed around a single adaptive 'next session' card that reads your recent load, recovery, and goals and tells you exactly what to do today — with the reasoning one tap away for the people who want it.

Offline-first sync meant a dropped connection mid-run never loses data, and the coaching model retrains nightly on each athlete's own history rather than a generic population.

The relaunch took the App Store rating from 3.8 to 4.9 and roughly doubled weekly active sessions. Support tickets about 'what do these numbers mean' all but vanished.

More importantly, retention at 90 days climbed by half — the clearest signal that the app finally earns its place on the home screen.

Built in React Native with native modules for the Bluetooth and sensor-fusion layers, backed by a real-time sync engine that reconciles offline writes without conflicts.

The adaptive coaching runs on an on-device model for instant feedback, with heavier nightly retraining in the cloud — so the app stays responsive even with no signal.