For years, fitness trackers have been excellent at collecting data. They know when you run, how many steps you took, and whether you slept like garbage last night. The problem? All that information just sits there, largely meaningless unless you actually know what to do with it.
Google’s betting that Fitbit’s AI Coach can finally bridge that gap. This week at its annual health event, the company announced a bunch of updates that suggest it’s taking a different approach to wearable technology than most competitors. Instead of just tracking numbers, Coach is trying to understand the complete picture of your health.
Sleep Data Just Got 15% More Accurate
Let’s start with what Google is calling its “most significant update yet”: a 15% improvement in sleep stage accuracy. That might sound like a modest bump, but it’s actually meaningful if you’ve ever wondered whether your tracker was just guessing about your REM cycles.
The new algorithm can now tell the difference between when you’re trying to fall asleep versus actually being asleep. It’ll know if you’re napping, if something interrupted your sleep, or when you’re transitioning between stages. That sounds obvious, but getting it right requires real technical work.
What’s particularly clever is how this feeds into a revamped Sleep Score. Rather than just measuring total hours slept, the new system factors in how long it took you to actually drift off. More data means better insights, which means Coach can actually give you useful recommendations instead of generic “sleep more” advice.
Here’s Where It Gets Interesting (And a Bit Weird)
Starting in April, Fitbit Premium subscribers in the US can link their actual medical records to the app. We’re talking medications, lab results, doctor visit histories, and whatever else your healthcare provider has on file.
This is the kind of feature that sounds incredible on paper. Your Coach can now cross-reference your blood pressure reading from a lab test with the workout data from your watch and give you something resembling personalized medical advice. Theoretically, this is better than generic fitness guidance.
But here’s the catch that privacy experts keep hammering home: uploading medical records into an AI tool is fundamentally different from just syncing your step count. Yes, Google says it stores everything securely and won’t use your medical data for ads. Yes, you control what gets shared and deleted. That’s all reassuring. But the moment that data exists in the system, it exists. And we’ve all watched enough tech companies get hacked to know that “secure” is relative.
Glucose Monitoring Just Entered the Chat
Google’s also letting Fitbit users connect continuous glucose monitors through Health Connect. If you’re using a Dexcom or Abbott Lingo, you’ll be able to ask Coach how your meals and workouts actually affect your glucose levels.
This is where business strategy becomes apparent. Google’s investing heavily in health research around insulin resistance, hypertension, and AI in virtual care settings. These aren’t random topics. They’re breadcrumbs showing where the company wants to take Fitbit next.
The glucose monitor integration feels inevitable in retrospect. If Coach already has your sleep data, workout history, and medical records, why not throw glucose into the mix? The more data it has, the more useful it becomes.
But Let’s Be Real About This
Here’s what bothers me about the entire premise. AI health coaches are explicitly not doctors. Google says so right there in the announcement. They can’t diagnose or treat anything. You shouldn’t change your health routine without talking to an actual human who went to medical school.
So what’s the actual value proposition here? A smart summary? Better context for conversations with your doctor? A way to feel like your expensive wearable is doing something meaningful?
Maybe all three, or maybe Google’s just iterating on features it thinks will keep people subscribed to Fitbit Premium. The medical records integration is compelling enough that it could work. People want their data to matter, and linking it to their actual health history makes it feel significant in a way that step counts never could.
The real question isn’t whether Coach is useful. It’s whether you’re comfortable handing over that level of health information to a tech company, even a massive one with decent security practices. Because no matter how good the insights are, that’s the trade-off you’re making.
