

The 2AM Question: ChatGPT or MyChart?
A few threads have come up again and again in recent conversations this past month:
- Epic’s announcement of Emmie and Art — their new patient- and clinician-facing AI tools — as well as CoMET, a foundation model built on more than 100 million patient records in their Cosmos data ecosystem
- OpenAI bringing on Ashley Alexander from Instagram and Nate Gross from Doximity to lead their healthcare efforts
- Mark Bertolini discussing Oscar Health’s new large language model platforms on Invest Like the Best
All the big players are now showing their AI hands. Each with a chatbot. Each with a model. Each with a unique plan for how AI shows up in care delivery.
Anyone who’s asked ChatGPT about whether a rash is serious or whether a set of symptoms might be cancer can see where we’re heading. It is impossible to imagine a future of healthcare that doesn’t include GPT-driven care. The big question is not if, but how do we get there, and how quickly? Across these threads, we can see the stage is set for an Olympian clash: tech giants pressing against the old titans of healthcare.
Live Players vs. Dead Players
This is a perfect example of what Samo Burja describes as Live Players vs. Dead Players. Incumbents like Epic, payors, and large providers seem to be playing the same game they always have, just a bit sharper. We’ve already seen this with AI use in prior auth, RCM, and documentation. Even Epic’s new AI assistants won’t be rolled out until late 2026, underscoring how slow the old cycle turns.
Far more compelling is the potential for consumer-facing AI to create entirely new ways to receive care.
What’s more likely:
- That Epic intercepts a patient’s late-night worry and MyChart is where they go at 2am to type “Why can’t I sleep?”
- Or, that the same question goes to ChatGPT first, and ultimately the reply is “Some sleep problems can be related to a thyroid imbalance. Would you like me to order a TSH test for you?”
Consumers are turning to ChatGPT for every kind of question, and health is just another category. Once the door opens for real clinical action, we believe it will be a tectonic shift.
Where Innovators Can Play
This shift opens up three clear fronts for innovation:
- Service-based marketplaces: ChatGPT already offers to draft a note for your doctor. It’s a remarkably short leap to imagine it asking “Want me to arrange an X‑ray?” or “Schedule a home blood draw?” Companies like General Medicine and GetLabs are already positioning themselves as services are increasingly triggered directly from AI-driven encounters.
- The interop layer: For consumer-first AI to matter clinically, its insights and actions will still need to flow into the health system, and vice-versa. CMS is already putting a spotlight here, but there is likely even more value than we realize in making data portable between ChatGPT and an EHR, or between a personal health AI and a primary care record.
- Behavioral change (the holy grail): Ordering labs is relatively simple. Changing long-term habits is not. Tools that integrate into AI flows and reliably nudge better sleep, diet, medication adherence, or stress management could finally unlock sustained behavior change. That might be the hardest, and most valuable, problem in healthcare.
Why Now
Two ladders are leaning against opposite sides of the same wall. On one side, costs are only rising higher. Health costs and premiums are set for their steepest increase in 15 years with many looking at double-digit increases for 2026. On the other side, technology only keeps building, with AI platforms already outperforming best-in-class clinicians in some domains.
The wall is finite. At some point, those ladders will meet. When they do, patients will be more than willing to climb over to viable alternatives to escape rising costs.
To be sure, there’s plenty of room for smart AI inside existing structures. But the tidal wave will break around a simple question: who captures patients at the first point of search, and earns the right to guide them the rest of the way?