Microsoft has built Copilot Health into a dedicated section of its Copilot assistant, pulling together wearable metrics from over 50 devices — Apple Health, Oura, and Fitbit among them — alongside electronic health records from more than 50,000 U.S. hospitals through a partnership with HealthEx, and lab data via Function. The company says its consumer products already handle more than 50 million health-related questions a day; Copilot Health is its attempt to do something more useful with that demand than answer queries one at a time.
The product runs on MAI-DxO — Microsoft AI Diagnostic Orchestrator — a multi-agent system that combines general-medicine knowledge with specialist depth. Microsoft has described the direction of travel as "medical superintelligence," though Copilot Health ships with explicit disclaimers that it isn't a diagnostic tool. That framing is partly regulatory caution, partly commercial prudence: health AI sits in a grey zone where the line between "insight" and "diagnosis" carries real legal weight.
What separates Copilot Health from the Apple Health and Google Fit crowd isn't the wearable integrations — those are table stakes. It's the hospital EHR access. Getting structured clinical data from 50,000 hospitals into a consumer product is not a trivial engineering or business-development achievement. Whether that data depth translates into meaningfully better outcomes for users, or mainly a stronger sales pitch, will take time to judge. Microsoft has assembled a 230-physician advisory panel spanning 24 countries and secured ISO/IEC 42001 certification for AI management to shore up its credibility on that front.
The product launches in the U.S. via waitlist, English only for now, with other languages and geographies planned. For the broader agent space, Copilot Health is one of the cleaner examples yet of agentic health orchestration at consumer scale: multiple sensitive data sources, real-time synthesis, a single branded surface. The harder question — whether users will hand Microsoft their EHR data, and whether regulators will let the product's clinical ambitions outpace its safety case — doesn't have a clean answer yet.