Amazon Just Made AI the Front Door to Primary Care. Here's What DPC Doctors Should Know.

There’s a new way to ask a health question at 2 AM, and it lives on Amazon.com. One Medical launched Health AI across Amazon’s website and app, rolling it out to customers now with a stated goal of reaching all U.S. customers soon. It changes what “access to primary care” is going to look like for a lot of people.

If you run a DPC practice, this is worth paying close attention to. Not because Amazon is coming for your patients directly. But because it’s reshaping what those patients expect before they ever walk through your door.

How Health AI Works

The system runs on Amazon Bedrock and uses multiple AI agents working together. A core agent talks to patients. Specialized sub-agents handle tasks like prescription renewals and appointment booking. Auditor agents review conversations in real time. Sentinel agents monitor the whole system. When the AI reaches its limits, a human provider takes over.

Prime members get up to five free direct-message consultations with One Medical providers for more than 30 common conditions. Amazon puts that at up to $145 in value, bundled into a Prime subscription those members already pay for. Health AI itself isn’t gated behind Prime, though: Amazon says you do not need to be a Prime member or a One Medical member to use it. Beyond the introductory offer, One Medical’s pay-per-visit pricing starts at $29 for direct-message care and $49 for video care, and Amazon notes prices vary by condition and visit type. A full One Medical membership runs $99/year for Prime members, down from $199. Family members can be added for $66/year each.

One Medical CMO Andrew Diamond laid out the vision at HIMSS26. “People are asking lots of really good questions that would, in many cases, require them to see or message a clinician.” The AI handles the front door. Clinicians handle what comes through it.

Amazon is also building specialty care bridges, partnering with health systems like Rush and Cleveland Clinic to route patients from primary care into specialized services.

The Word That Barely Appears

Dr. Douglas Farrago at DPC News noticed something telling about Amazon’s announcement: “the term physician is never used in this whole article. Not once.” That reads a shade stronger than the announcement does. The word turns up twice, once in the American Academy of Physician Associates survey Amazon cites and once in the opening where co-author Dr. Andrew Diamond is described as having spent nearly two decades as a primary care physician. What’s missing is the word in the one place it would matter most. When the announcement describes the person a patient actually reaches, it says “One Medical provider,” “licensed health care professionals,” and “human provider.” It does use “clinician,” but only as a benchmark for the AI — learning “just as human clinicians learn from real cases,” and meeting “clinician-level performance.”

That language choice isn’t accidental. It reflects a model where the AI is the relationship and the human clinician is the escalation path. That’s the inverse of how DPC works. In a DPC practice, the physician is the relationship, and technology supports that connection.

This distinction matters more than it might seem at first glance. When you abstract the doctor out of the language, you’re signaling that any licensed provider is interchangeable. DPC is built on the opposite premise.

The Pressure and the Opportunity

This isn’t a story about AI replacing doctors. Amazon’s own CMO says that isn’t happening. Diamond noted that “most people that interact with AI in a primary care clinic are not finding that their interaction with a physician has been replaced.”

But Amazon is redefining what patients expect from primary care access. When a health question at midnight can be answered inside the same app people already use to order paper towels, and the answer draws on their medical records, that becomes the new baseline. The bar for “access” just moved.

You might feel pressure from two directions at once. Patients who’ve used Health AI will come to your practice with more information and higher expectations for responsiveness. They’ll expect you to be reachable outside business hours. They’ll expect fast answers to straightforward questions. If your practice doesn’t offer that kind of access, Amazon is showing them someone who will.

But there’s a real opportunity here too. Amazon’s model is built on scale. It handles volume by putting AI at the front door and a clinician behind it. That works for a UTI prescription or an allergy refill. It doesn’t work when someone needs 45 minutes to talk through a new diabetes diagnosis. It doesn’t work when symptoms don’t fit a neat algorithm. And it definitely doesn’t work when the real issue isn’t medical at all.

DPC is built for the things AI can’t do. The long conversations. The relationships that develop over years. The clinical judgment that comes from knowing a patient’s full story, not just their data points.

What This Means

Amazon isn’t trying to compete with your DPC practice head-on. They’re competing for the definition of primary care itself. Their bet is that most primary care interactions can be handled by AI with human backup, and that patients will accept that trade-off for convenience and price.

If you’re running a DPC practice, the move isn’t to out-tech Amazon. You can’t, and you don’t need to. The move is to make the case for what technology-mediated care can’t deliver. Continuity. Depth. A physician who knows your name without checking a database first.

The practices that will thrive aren’t the ones ignoring AI. They’re the ones using it to handle administrative noise while doubling down on the thing Amazon can’t replicate at any scale. The relationship.