Fitbit’s March update quietly turns its Gemini-powered personal health coach from a clever experiment into something that starts to look like an actual companion for your health — one that knows how you sleep, how your metabolism is doing, and, soon, what your doctor’s lab reports say about you. It is still officially a public preview, but the feature set is maturing fast enough that it no longer feels like a mere AI chatbot bolted onto your step counter.

The first big piece of this update is sleep, and Fitbit is treating it less like a vanity metric and more like a core vital sign. Fitbit is rolling out what it calls its most significant sleep-tracking upgrade yet, promising roughly a 15% bump in sleep staging accuracy for personal health coach public preview users. Under the hood, the models have been retrained on more diverse datasets so they can better differentiate between you trying to fall asleep and actually being asleep, and they do a better job at catching real-world nuisances like brief wake-ups, naps, and transitions between stages. That matters because the reworked Sleep Score is no longer just a single “good” or “bad” grade; it breaks down specific components such as how long it took you to nod off and how much of your night was spent in more restorative stages, feeding those details back into tailored coaching tips instead of generic “sleep more” advice.

In practice, this means the coach can start nudging you on very specific levers rather than hand-waving about “sleep hygiene.” If your data shows you take forever to fall asleep but get decent, deep sleep once you’re out, the coach can focus on your wind-down routine and late-night screen habits rather than suggesting earlier bedtimes you’ll just ignore. For shift workers or new parents whose sleep is a mess by design, better detection of naps and interruptions also means the app is less likely to misjudge your nights as catastrophic and more likely to recognize the patchwork reality of your rest. The upgraded tracking is rolling out first, with the refreshed Sleep Score experience following over the next few weeks for those in the preview.

Beyond sleep, the update leans heavily into metabolic and cardiometabolic health — essentially, how your body handles things like glucose and blood pressure over the long term. Google and Fitbit’s health teams just had a study published in Nature that uses everyday wearable data plus routine blood tests to predict insulin resistance, a precursor to type 2 diabetes that often goes undetected for years. In that study, data from Fitbit and Pixel devices — heart rate, activity, sleep duration and more — combined with simple lab markers like fasting glucose and basic demographics, allowed a model to flag insulin resistance with a level of sensitivity and specificity that suggests wearables could become an early warning system, not just a retroactive log. That same line of research is feeding into Fitbit’s personal health coach so it can eventually give more context-aware guidance like, “Your patterns look similar to people at higher risk; here are lifestyle changes that may help,” instead of staying in the realm of generic wellness platitudes.

Fitbit is also pushing deeper into cardiovascular research through its hypertension work. The company has been recruiting thousands of Pixel Watch 3 users into a long-term Hypertension Study Lab, using wearable data and ambulatory blood pressure readings to see whether a smartwatch can reliably spot early signs of high blood pressure. The aim is to use subtle changes in trends — resting heart rate shifts, activity patterns, sleep quality — as signals that your blood pressure may be heading in the wrong direction, paving the way for future features that could warn you earlier and prompt a conversation with your doctor instead of waiting for a crisis. For now, it is still framed as research rather than a diagnostic tool, but it slots neatly into the personal health coach’s mission: connecting long-term patterns in your data to practical guidance you can act on.

What makes this March update feel different, though, is how Fitbit starts to bridge the gap between consumer wearables and the formal healthcare system. Starting next month, public preview users in the U.S. will be able to link their medical records directly into the Fitbit app, pulling in lab results, medications, and visit history from participating providers. Fitbit is working with partners like b. well and CLEAR so you can either sign in to your health system’s portal or verify your identity using an IAL2-certified process — basically, a selfie plus a valid ID — and then let the system search for and sync records across different providers where possible. Once those records are in, the personal health coach can use them alongside your wearable data to answer questions with a level of relevance that hasn’t really existed in mainstream fitness apps until now.

Instead of asking, “How do I lower cholesterol?” and getting the same answer as everyone else, you’ll be able to ask, “How do I improve my cholesterol?” and the coach can summarize your own lab panels, highlight which values are out of range or moving in the wrong direction, and then offer wellness-focused suggestions that match your history and your current activity and sleep patterns. If your LDL is creeping up while your sleep and exercise are trending down, the coach can connect those dots and frame your next steps in a way that acknowledges your actual trajectory instead of a hypothetical ideal. In the coming months, Fitbit plans to let you share these summaries and even your records via a secure Smart Health Link or QR code, making it easier to loop in family members or clinicians without exporting messy PDFs or screenshots.

Of course, plumbing medical records into a consumer health app raises real privacy questions, and Fitbit seems very aware of how sensitive this looks on paper. The company is emphasizing that medical records will be stored securely, with clear on-device controls over how that data is used, shared, or deleted. Just as importantly, Fitbit is drawing a bright line around ads: medical records and other health data in Fitbit are not used for advertising, which is a critical promise if Google wants users to trust it with their lab history and prescriptions. Fitbit is also careful to stress that the personal health coach, even with access to clinical data, is not intended to diagnose, treat, cure, prevent, or monitor disease; it is a wellness tool that should complement, not replace, professional medical advice.

The other thread running through this update is how conversational the whole experience is becoming. Earlier this year, Fitbit started expanding the personal health coach preview to more countries — including the UK, Canada, Australia, New Zealand, and Singapore — and to iOS users, making the Gemini-powered assistant accessible on far more wrists and phones. The onboarding feels more like a chat with a trainer than flipping through a setup wizard: you spend a few minutes talking about your goals, preferences, and what equipment you actually have, and the coach uses that as context to build multi-week plans and answer follow-up questions. With the March update, that conversation gains a memory of your clinical history and deeper insight into your sleep and metabolic patterns, which should make responses feel less like generic blog posts and more like commentary on your actual life.

Fitbit is also experimenting with how AI can help during real clinical encounters. In the next few weeks, it is kicking off a “Get care now” Fitbit Lab research study with Included Health, a U.S. virtual care provider, to see how conversational AI can guide people through virtual visits more effectively. The idea is that an AI coach that already understands your wearable data and, potentially, your synced records could prepare you for the visit, help you ask better questions, and make sense of the recommendations afterward, rather than leaving you to decipher medical jargon alone. It is another example of Fitbit trying to move from “after-the-fact tracker” to “in-the-moment guide,” stitching together daily life, virtual care, and long-term trends into a single narrative you can actually follow.

For everyday users, the real test will be whether all of this complexity translates into simple, trustworthy nudges that feel worth paying attention to. If the upgraded sleep tracking means your bedtime reminders finally line up with how tired you actually feel, or if early metabolic signals give you enough warning to change course before a scary diagnosis, then Fitbit’s personal health coach could become more than just another tile in your app’s Today tab. The March update does not magically turn your watch into a doctor on your wrist — and Fitbit is clear it should not be treated that way — but it does push the ecosystem a step closer to what many people have been quietly hoping for: a health companion that understands both the data you create and the data your doctors collect, and can talk to you about both in plain, human language.