Fitbit’s revamped app, with Gemini-powered health coach, rolls out to Premium users (techcrunch.com)

Google is rolling out a public preview of a Gemini-powered “Coach” in the revamped Fitbit app to Premium subscribers in the U.S. on Android (iOS coming later). Coach uses Google’s Gemini LLM to generate personalized fitness and sleep plans—creating routines based on goals, preferences and available equipment, adjusting workouts in real time from user feedback, and modifying plans if you report an injury. The updated app also reorganizes data into four tabs (Today, Fitness, Sleep, Health), surfaces cardio load and weekly metrics, offers AI-driven sleep insights and aggregates vitals like breathing rate, SpO2 and resting heart rate for quick review. Some features (nutrition tracking, cycle logging) remain unavailable at launch. For the AI/ML community, this is a concrete example of large-model personalization applied to continuous health telemetry: Gemini acts as the decision layer that translates biometric and behavioral signals into adaptive coaching. That raises opportunities for more context-aware guidance and longitudinal behavior change, but also underscores integration challenges—sensor fidelity, model input engineering, privacy and clinical validation. Access requires an active Fitbit Premium subscription and a compatible Fitbit device or Pixel Watch, positioning Google to differentiate Fitbit through LLM-driven UX while testing real-world model-in-the-loop health workflows.