#19: Big Tech Makes Its Health Move

OpenAI goes consumer. Anthropic goes enterprise. Clinicians get the downstream effects.

In the span of a week, OpenAI and Anthropic both launched healthcare-specific products.

This isn’t “AI in healthcare” as a concept.

This is AI companies choosing where to plant their flags in the health stack:

Same category. Different endgames.

OpenAI launched two things and announced an acquisition:

ChatGPT Health allows users to connect medical records and wellness data, then ask for summaries, trend analysis, and visit preparation. This is not framed as diagnosis. It’s framed as orientation—helping patients understand what they already have.

That positioning matters. It keeps OpenAI clear of FDA territory while still becoming the place where patients begin forming a narrative about their health.

ChatGPT for Healthcare is the enterprise version—GPT-5 models with “HIPAA‑supporting” infrastructure and BAAs, already rolling out at Cedars-Sinai, Memorial Sloan Kettering, Stanford Children's, and UCSF. Based on the implementation, it can draft notes, pulls evidence with real citations, and integrate with your existing systems securely.

Torch was acquired for $100m. OpenAI’s acquisition of Torch is best understood as an acquihire, not a feature drop.

Torch’s value wasn’t a standalone product that OpenAI intended to ship as-is. It was the team. The founders came from Forward, where they spent years building at the intersection of healthcare delivery, AI, and consumer-grade software. That experience is rare—and hard-earned. They’ve lived inside the realities of clinical operations, patient expectations, and the brittleness of health data infrastructure.

Just as important: the Torch team had already wrestled with the unsexy but decisive problem in healthcare AI—integration.

Connecting labs, prescriptions, visit artifacts, longitudinal records, and consumer health tools isn’t an intelligence problem. It’s a plumbing problem. A permissions problem. A standards problem. A workflow problem.

Torch brought OpenAI people who understand that mess firsthand. In other words, this wasn’t about buying a product.

It was about accelerating OpenAI’s ability to operate inside real healthcare complexity.

OpenAI’s healthcare leadership and product hires reflect a consumer-first orientation—people who know how to scale platforms, shape user behavior, and build products patients actually adopt.

Taken together, the message is clear: OpenAI is aiming to be where patients start—and where their health story increasingly lives.

Anthropic followed four days later with Claude for Healthcare. It takes a very different angle.

Rather than going after consumer-facing health narratives, Anthropic is leaning into enterprise-grade and life sciences use cases: administrative workflows, payer/provider infrastructure, and integration with trusted medical and scientific data sources.

This includes support for tasks like prior authorizations, documentation, and clinical operations—areas where speed, consistency, and auditability matter more than consumer UX.

They also launched connectors that include ways to pull in medical records and Function Health (the big winner from both announcements?).

Anthropic’s bet is that healthcare adoption will flow from the inside out: through institutions, payers, research, and regulated environments—rather than through patients first.

Same industry. Different gravitational centers.

Whether it’s ChatGPT Health or similar tools that follow, more patients will walk in with a synthesized view of their data and an AI-generated interpretation of what it might mean

That can be an asset: fewer missing records, more prepared patients, better questions.

But it also shifts the clinician’s role. Less information delivery. More interpretation, prioritization, and correction.

We are already seeing early regulatory experiments—like AI-enabled prescription renewals in limited jurisdictions—that would have been unthinkable a few years ago.

These tools are staying just outside diagnosis today, but the pace of change suggests that traditional lines around licensure and scope will be tested sooner than most clinicians expect.

The strategic question isn’t whether AI will replace clinicians.

It’s which tasks get absorbed into software—and which become more valuable because of it.

Many clinicians practice knowing they don’t have the full story: fragmented records, missing labs, partial histories, disconnected wearables.

These tools—especially when driven by consumer aggregation—make it increasingly possible to see the whole patient, not just the slice that shows up in one EHR.

That doesn’t replace judgment. It raises the ceiling on what judgment can be applied to.

This week wasn’t about AI “doctors.”

It was about where intelligence enters the system:

Clinicians will feel both—through more informed patients and faster-moving systems.

The opportunity isn’t to resist that shift.

It’s to step into the role that becomes more valuable as everything else accelerates: sensemaking, sequencing, and trust.

That’s the work no platform can automate—no matter how powerful the model.

Before you integrate anything, run every AI use case through these three filters:

AI hallucinations are real. Even in favorable internal benchmarks, Claude still gets a non‑trivial fraction of complex medical calculations wrong. For medication dosing, the acceptable error rate is zero.

The rule is simple: AI drafts, you decide. Every note gets reviewed. Every patient message gets eyes before sending. Every citation gets verified if it matters clinically.

This isn't paranoia. It's how you stay out of liability trouble while still capturing the efficiency gains.

Here's where most clinicians get sloppy.

Consumer ChatGPT and Claude are not HIPAA-compliant. Pasting patient data into the free version—even "de-identified" data—is a problem. The enterprise versions with BAAs are different. Know which you're using.

If you're in a hybrid model (insurance-billed care plus cash wellness memberships), you're likely touching both HIPAA-regulated PHI and state-regulated consumer health data. Washington's My Health My Data Act, for example, covers wellness data that HIPAA doesn't.

The practical move: Create a one-page internal policy. 

This is the question nobody's asking.

Adding AI tools doesn't automatically make your practice better. It can just as easily create more cognitive load, more things to check, more systems that don't talk to each other.

The win isn't "more AI." It's AI that makes your intake, workup, intervention, and follow-through feel like one continuous logic—to you and to your patient.

Before adopting any tool, ask: does this help my clinical thinking flow, or does it interrupt it?

This week, we share a cross-posted podcast from Fitt/Insider, featuring Sunita Mohanty, Co-Founder and CEO of Vibrant Practice.

Fixing healthcare’s foundational flaws, Vibrant Practice is building the AI-native electronic health record for preventative longevity practices — with tools built for personalized, data-driven care

In this episode, we discuss building the operating system for modern medicine.

We also cover:

Get ahead in 2026 with Vibrant, the AI-powered, all-in-one EHR built specifically for personalized medicine. Schedule a demo with our team to learn more about how we can help you extend your clinical brain and deliver great personalized care.

AI Meets Longevity Tech at CES 2026 AI isn’t just entering clinics — it’s showing up in your bathroom. The NuraLogix Longevity Mirror, unveiled at CES 2026, uses AI-driven imaging to analyze a 30-second selfie for signs of cardiovascular, metabolic, and physiological health, then produces a longevity score and personalized insights. The standalone device, expected to ship early this year, also reports sub-scores for heart health, metabolic age, and stress, and may someday supplement preventive health monitoring if core vitals measurements receive regulatory clearance. 

Is giving ChatGPT Health your medical records a good idea? As OpenAI rolls out ChatGPT Health — letting users upload medical records and link apps like Apple Health, Function, and MyFitnessPal for personalized interpretation — privacy and accuracy concerns are dominating the conversation. Experts recognize the potential for more informed patients, but caution that sensitive health data in general-purpose AI tools raises questions about confidentiality, reliability, and how patients might act on AI interpretations without clinical context. 

5 Takeaways from LongevityFest 2025 At LongevityFest 2025, clinicians and longevity experts reiterated that the future of practice isn’t just extending lifespan, but preserving functional capacity — and that includes learning to work with AI and data tools. Among the top takeaways for 2026 practice success: focus on metabolic resilience, brain health, and muscle preservation, while integrating trusted AI-enhanced approaches into clinical workflows and patient engagement strategies. 

The Modern Clinician is written for functional, integrative, and longevity-focused physicians who want to scale their impact and deliver cutting-edge care.

If you liked this one, share it with a colleague! We appreciate you spreading the word.

To learn more about the why behind this newsletter, start with our first post introducing The Modern Clinician.