Microsoft Copilot Health Turns Your Body Data Into AI Advice

Your Oura ring knows your sleep. Apple Health tracks your steps. Your doctor's portal has lab results you forgot about six months ago. None of them talk to each other — until now, apparently.

Microsoft launched Copilot Health on March 12, a dedicated space inside Copilot that pulls data from wearables, electronic health records, and lab tests into one place. Then it does what you'd expect an AI to do with all that data: finds patterns, surfaces insights, and tries to help you ask better questions at your next doctor's appointment.

What It Actually Connects To

The integration list is more ambitious than anything Apple or Google offers on the consumer side. Copilot Health connects to 50+ wearable devices through Apple Health, Oura, and Fitbit. It pulls electronic health records — visit summaries, medication lists, test results, clinical notes — from over 50,000 U.S. hospitals and provider organizations through a partnership with HealthEx. And it imports comprehensive lab results from Function.

That last piece is the one that matters. Wearable aggregation is table stakes in 2026. EHR access at scale is not. HealthEx's network covers 52,000+ healthcare organizations via direct FHIR-endpoint exchange, plus another 12,000+ through TEFCA services. That's a genuinely massive health data pipeline feeding into a consumer AI product.

The Margaret Demo

Microsoft showed Copilot Health working with synthetic patient data. Margaret is diabetic with a history of high blood pressure. The AI analyzed her wearable sleep data, flagged insufficient deep sleep, and connected it to cardiovascular risk. It then cross-referenced lab results showing poorly controlled diabetes, wearable readings showing elevated blood pressure, and cholesterol numbers from recent bloodwork.

The pitch: instead of you manually correlating data across three different apps and a patient portal, Copilot Health does it automatically. It also helps you formulate specific questions for your doctor — which, if you've ever sat in an exam room blanking on what to ask, is arguably the most practical feature here.

Not HIPAA-Covered, and That's Worth Noting

Copilot Health operates as a direct-to-consumer service, which means it is not subject to HIPAA. Microsoft is upfront about this. Your health data is isolated from general Copilot, encrypted at rest and in transit, and — critically — not used for model training. You can disconnect any data source or delete your information at any time. The product has ISO/IEC 42001 certification, the international standard for AI management systems.

Still, "your health records in a consumer AI product that isn't HIPAA-covered" is a sentence that will make some privacy advocates uncomfortable. Microsoft built this with an external panel of 230+ physicians from 24 countries, partnered with AARP (38 million older Americans) and the National Health Council (180+ patient advocacy groups). The clinical governance is serious. The regulatory framing is worth watching.

The Bigger Ambition: Medical Superintelligence

Behind the consumer product sits something more provocative. Microsoft published a companion piece on its path to "medical superintelligence" — health AI that combines a generalist's breadth with a specialist's depth. Their diagnostic orchestrator, MAI-DxO, uses multi-agent reasoning where virtual specialist agents debate differential diagnoses.

The numbers: paired with OpenAI's o3, MAI-DxO hit 85.5% accuracy on 304 complex New England Journal of Medicine case benchmarks. Experienced physicians scored 20% on the same cases. That's a 4x gap — though the caveat matters. NEJM cases are deliberately hard teaching cases, not routine checkups. The system also reduced unnecessary lab and imaging orders by 28% in simulation.

None of this is in Copilot Health today. But it signals where Microsoft sees this going: from "here's a pattern in your sleep data" to actual diagnostic reasoning. Copilot Health is the data collection layer. MAI-DxO is the intelligence layer waiting in the wings.

Who Gets It

US only, adults 18+, English language, waitlist. Microsoft hasn't announced pricing beyond an initial free trial. The competitive field is getting crowded fast — OpenAI launched ChatGPT Health in January, Amazon is expanding health chatbot access through One Medical, and Anthropic announced Claude for Healthcare the same week. But none of the competitors match Copilot Health's EHR integration depth. Fifty thousand providers is a moat, if the product delivers on the promise.

The real question isn't whether AI can synthesize health data — it obviously can. It's whether people will trust a Microsoft product with their medical records the way they trust their doctor. Copilot's enterprise momentum is undeniable. Consumer health is a different kind of trust.