We're here
to help.

AI Health Insights reads directly from Apple HealthKit on your device. We never receive raw values — everything stays on your iPhone unless you enable iCloud sync, which uses your private CloudKit container.

Open iOS Settings → Privacy & Security → Health → AI Health Insights and make sure read access is enabled for the categories you care about (heart, sleep, workouts, etc.). Then pull-to-refresh the Dashboard.

Insights are produced on-device after enough recent data is available — typically within 24–48 hours of first sync. They refresh in the background and on each app launch.

No. We don't sell, rent, or share health data. Anonymous crash logs and aggregate event counts may be sent via Firebase if you've allowed analytics in iOS, but they contain no HealthKit values.

Open the iOS Settings app → your name at the top → Subscriptions → AI Health Insights → Cancel Subscription. Apple manages all billing; we never see your payment method.

Sign in with the same Apple ID, open the paywall, and tap Restore Purchase. Your subscription will reactivate within a few seconds.

No, but the app is significantly more accurate with one. Heart rate variability, resting heart rate, sleep stages, and workout heart-rate zones rely on Watch sensors.

Profile → Data & Privacy → Delete All Data. This removes local storage and queues a CloudKit deletion for any synced copy. Uninstalling the app also removes local data, but does not delete CloudKit copies.

AI Health Insights requires iPhone with iOS 18 or later. It is optimized for iPhone 12 and newer. Apple Watch (Series 4 or later, watchOS 11+) unlocks the most accurate HRV, heart rate, sleep staging, and workout data. The app runs on iPhone without a Watch, but accuracy for HRV and sleep stages is reduced without Watch sensors. iPad and Mac are not currently supported as primary devices but can receive iCloud-synced data from your iPhone.

The app requests read access to nine HealthKit categories: Steps, Distance, Exercise Time, Active Energy, Heart Rate, Resting Heart Rate, Heart Rate Variability (HRV), Sleep Analysis, and Workouts. You control which categories you allow. Enabling all categories produces the most complete readiness score and the most personalized AI insights.

Yes. Any app that writes to Apple Health — Strava, Nike Run Club, Garmin Connect, Oura, Whoop, and many others — can feed data into AI Health Insights. The app reads from HealthKit, not from those apps directly, so compatibility is automatic as long as the other app shares its data with Apple Health.

The readiness score (0–100) is a daily summary of how recovered your body is and how prepared it is for physical or cognitive effort. It is calculated on-device from your recent HRV trend, resting heart rate, sleep duration, sleep consistency, and activity load. A score of 70 or above typically indicates a good day for moderate to intense exercise. A score below 50 suggests prioritizing rest or light activity. The exact weighting adapts to your personal history — the app builds a baseline from your data rather than comparing you to population averages.

Apple Health displays raw data — it shows you your step count, heart rate reading, and sleep time as individual numbers. AI Health Insights synthesizes that data into a single daily readiness score, detects trends and anomalies automatically, identifies correlations between your behaviors and outcomes (like caffeine timing and deep sleep), and generates plain-language recommendations. It is designed specifically around the Apple Watch HRV tracking app use case that Apple Health's native interface does not address: turning biometric data into actionable daily guidance.

The AI engine runs entirely on your iPhone using Apple's Core ML framework. It contains 30 rules that analyze your HealthKit data for trends (changes over 7, 30, or 90 days), correlations (relationships between two metrics across time), and anomalies (single-day values significantly above or below your personal baseline). Each rule produces a candidate insight with a confidence score. Rules with sufficient confidence become the insights you see. The median inference latency is 50 milliseconds. No data is sent to any external server at any point.