Three years after its $3.9 billion bet on One Medical, Amazon is deploying its healthcare AI assistant across its main website and shopping app, in a direct challenge to OpenAI’s ChatGPT Health and Anthropic’s Claude for Healthcare.
Amazon has opened its Health AI assistant to all US customers, removing the requirement to be a One Medical member or a Prime subscriber to access the product.
Previously available only inside the One Medical app, Health AI is now accessible through Amazon.com and the Amazon mobile app.
The assistant can answer general health questions without accessing a user’s personal medical records, but its core proposition is personalisation. Customers who consent to share their health data through the Health Information Exchange, the nationwide system for transferring patient medical records between providers, can ask Health AI to interpret their lab results, explain diagnoses, review their medications, and receive guidance that accounts for their specific health history.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now!
The assistant can also book appointments with One Medical providers, manage prescription renewals through Amazon Pharmacy, and connect users to clinical care via message, video, or in-person visit.
For Prime members in the US, Amazon is offering an introductory bundle: up to five free direct-message consultations with a One Medical provider, covering more than 30 common conditions including cold and flu, allergies, acid reflux, pink eye, UTIs, erectile dysfunction, anti-ageing skincare, and hair loss.
Amazon values this at up to $145 in care. Non-Prime users can access One Medical providers through a pay-per-visit option priced at $29. Prime members who want ongoing access can buy a One Medical membership at $99 per year, against the standard $199 rate.
Health AI runs on Amazon Bedrock, the company’s managed AI infrastructure service, and is described in the launch post as a multi-agent architecture: a primary agent handles patient conversations, sub-agents process specific clinical workflows, auditor agents review conversations in real time, and sentinel agents monitor the system end-to-end with escalation pathways to human providers.
Amazon says the system was evaluated against synthetically generated clinical conversations before deployment, and is required to meet or exceed clinician-level performance on safety-critical decisions before being released.
Amazon describes Health AI as designed to support rather than replace the relationship between a patient and their doctor. The assistant will refer users to a human provider rather than answer if it is uncertain about a clinical recommendation.
It is explicitly not intended for diagnosis or treatment without the involvement of a care provider.
The privacy posture matters here, and Amazon has been careful to lay it out, if not exhaustively. All interactions take place within a HIPAA-compliant environment and are protected by encryption and access controls, the company says.
Protected health information from One Medical and Amazon Pharmacy is not used for general merchandise advertising or sold to third parties. Amazon does, however, acknowledge that it trains Health AI on “abstracted patterns” derived from aggregated patient interactions, for instance, using patterns from multiple patients asking about medication interactions to improve future responses, while keeping individual identities private.
The company says this is permissible under HIPAA. TechCrunch reported it has asked Amazon for specifics on encryption and access controls but had not received a response at time of publication.
Researchers at institutions including Stanford and Duke have urged caution about sharing personal health information with AI systems, citing concerns around training data use and the reliability of AI health guidance.
Those concerns are not Amazon-specific, they apply to the entire category, but they are likely to intensify as Health AI grows its user base beyond the One Medical membership and into the much larger population of Amazon shoppers.
Health AI is entering a crowded market that is moving fast. OpenAI launched ChatGPT Health in January 2026. Anthropic followed a week later with Claude for Healthcare. Amazon now joins both as a major player competing directly for the AI health assistant user.
Unlike the other two, Amazon arrives with an integrated clinical network, a pharmacy, and an existing relationship with hundreds of millions of retail shoppers, a very different distribution advantage.
The assistant was co-developed with One Medical’s clinical team and is written about in the launch post by Prakash Bulusu, CTO of Amazon Health Services, and Dr. Andrew Diamond, Chief Medical Officer of Amazon One Medical.
Amazon says it intends to expand availability to all US customers in the coming weeks.
Get the TNW newsletter
Get the most important tech news in your inbox each week.