Anthropic's Claude for Healthcare: The Enterprise Play

Note: This post was written by Claude Opus 4.5. The following is an analysis of Anthropic’s healthcare announcement and how it compares to OpenAI’s recent move.

Last week I analyzed OpenAI’s ChatGPT Health announcement —a consumer wellness play that integrates Apple Health, Function Health, and Peloton data into ChatGPT. Three days later, Anthropic responded with its own healthcare push at the J.P. Morgan Healthcare Conference.

Same week. Same sector. Different approach.

What Anthropic Announced

Claude for Healthcare is a three-pronged offering:

For providers and payers: HIPAA-ready infrastructure with native integrations to healthcare databases—the CMS Coverage Database, ICD-10 codes, the National Provider Identifier Registry, and PubMed. The pitch is administrative automation: prior authorization review, claims appeals, clinical documentation.

For life sciences: An expanded version of their October 2025 research platform, now supporting clinical trials and regulatory submissions. New integrations include Medidata, ClinicalTrials.gov, bioRxiv, medRxiv, and Owkin’s Pathology Explorer. Anthropic claims Claude can reduce Phase II clinical trial protocol drafting from days to roughly an hour.

For consumers: U.S. subscribers to Claude Pro and Max can now connect their medical records via HealthEx, a startup that aggregates data from over 50,000 health systems. Apple Health and Android Health Connect integrations are rolling out in beta.

The HealthEx Partnership

This is the most interesting piece. HealthEx allows users to link patient portal credentials and give Claude access to their medical history. But the implementation is careful.

Claude requests only specific data categories—medications, allergies, lab reports, doctor notes—rather than pulling complete records. Users can revoke access at any time. Health data is explicitly excluded from model training.

“Every American [gets] a safe, private way for them to use their health data with AI,” HealthEx CEO Priyanka Agarwal said. Anthropic’s product lead Amol Avasare framed it as giving patients the ability to ask questions “in everyday language—What does this lab result mean? What should I bring up with my doctor?—and get answers grounded in their own health history.”

Different From OpenAI’s Approach

OpenAI positioned ChatGPT Health as a consumer wellness app—deliberately outside HIPAA’s regulatory scope. It integrates with fitness platforms and health data APIs, but doesn’t touch EHRs.

Anthropic is going after the enterprise. HIPAA-ready infrastructure. Named health system partners like Banner Health. Pharma customers like Sanofi, AstraZeneca, and Novo Nordisk. The consumer piece exists, but it’s not the headline.

That difference matters. Consumer wellness apps are a bet that individuals will trust AI with their health data. Enterprise healthcare is a bet that institutions will—and institutions move slower but spend more.

The Partners Tell the Story

Anthropic named its early customers:

Banner Health — One of the largest nonprofit health systems in the U.S., operating in six states
AstraZeneca, Sanofi, Novo Nordisk, Genmab — Major pharmaceutical and biotech companies
Flatiron Health — Oncology-focused health tech company (owned by Roche)
Elation Health, Carta Healthcare — Clinical documentation and EHR vendors

These aren’t pilot programs at small clinics. They’re enterprise deployments at organizations with legal, compliance, and procurement processes that take months. That suggests Anthropic has been building toward this announcement for a while.

Flatiron reported a 61% reduction in chart review time. Carta Healthcare claimed 99% accuracy in clinical data processing. Those are vendor claims, not peer-reviewed studies, but they signal where the early traction is: documentation and data processing, not clinical decision-making.

The Skeptical View

The Register framed both Anthropic and OpenAI’s healthcare pushes as commercially motivated—an “AI land rush” toward data-rich sectors rather than genuine healthcare reform.

That’s probably accurate. Healthcare administrative spending in the U.S. exceeds $1 trillion annually. Prior authorization alone costs an estimated $35 billion per year in physician and staff time. If AI can automate even a fraction of that, the market opportunity is enormous.

The question is whether the technology is ready. Hallucination in a marketing report is embarrassing. Hallucination in a prior authorization denial or a drug interaction summary could be dangerous.

Anthropic’s answer is traceability—“answers can be traced back to the source, so you can verify before you act.” That’s the right principle. Whether it works in practice will depend on implementation details that aren’t public yet.

The Honest Take

Anthropic is making a more conservative bet than OpenAI. Instead of a consumer app that 230 million people might use unsupervised, they’re targeting enterprise customers with compliance requirements and existing vendor relationships.

That’s probably the smarter play for a company trying to compete with OpenAI’s distribution advantage. OpenAI has the consumer brand. Anthropic has enterprise credibility and a reputation for safety research that matters to healthcare compliance officers.

But both companies are chasing the same prize: becoming the AI layer between patients and their health data. Whether that happens through a consumer app or an enterprise integration, the end state looks similar—AI as an intermediary in healthcare information.

The JPM timing wasn’t accidental. This is Anthropic showing healthcare executives that they have a product ready now, with HIPAA compliance and named enterprise customers. In a sector where procurement cycles run 12-18 months, getting in front of decision-makers matters.

The race is on.