A new AI is here – and it can book patients’ appointments and buy their OTCs

Agentic AI in healthcare just became a reality. Here’s an honest look at where things stand — what it is, who’s in this space, what it can and can’t do (yet), and what it means for your healthcare website.

Following completion of a trial period, Amazon expanded Health AI to all US consumers in March 2026. The catch is that its downstream integrations are currently limited to Amazon’s own verticals — so any recommended treatment or appointment bookings are exclusive to One Medical providers.

So, we’re not quite at the point where AI health agents are out in the wild, accessing any healthcare provider’s website and booking appointments on a patient’s behalf. And in Europe, where regulation is stricter, we’re likely further behind still.

That said, the groundwork for broader adoption of agentic AI is moving faster than the public picture suggests. According to McKinsey, 1 in 2 health organisations are testing proof of concept internally, and up to 1 in 5 are actively integrating it into workflows, as of their survey findings published in April 2026.

Amazon is currently the only major player with a genuinely agentic healthcare product. But around 2 in 3 healthcare providers believe that agentic AI will soon meaningfully disrupt the patient-provider experience, and for the better.

Meanwhile, it’s far from quiet in the broader AI health space. OpenAI’s ChatGPT Health and Anthropic’s Claude for Healthcare both launched in early 2026. 

These are sophisticated health-focused AI assistants that can connect to medical records, answer detailed health questions and help patients navigate their care. But they’re not agentic in the true sense: they respond, they don’t act.

What these platforms are doing, though, is establishing the consumer habit. The infrastructure and consumer appetite that makes agentic healthcare viable is being built right now, just one layer below.

The use cases above are just the consumer-facing side of a much broader picture. Agentic AI is being trialled across clinical settings, from autonomous health coaching for cancer survivors, to ChatExosome, an agentic AI system that can independently analyse patient samples for cancer without a clinician initiating each step. 

What these early studies collectively represent is a significant and wide-ranging effort to make agentic AI useful across some of healthcare’s most complex challenges. 

Nobody will be surprised to hear that, in terms of the evidence base to support the use of agentic AI in healthcare, it’s a bit of a wild west out there. 

Most academic work acknowledges a very limited understanding of how agentic AI could impact patient health outcomes. There are studies, but they have very specific context and use cases, often rely on highly structured or simulated data, and they certainly haven’t been run in the real world.

Meanwhile, companies like Amazon are off at a million miles an hour. They claim to have rigorously tested and scrutinised their technology. Whether that will satisfy regulators remains to be seen, particularly in the UK and Europe.

Inevitably, these platforms are required to operate within HIPAA, GDPR and general consumer protection law, but as with generative AI, meaningful governance to protect patients and their health outcomes remains underdeveloped.

Then there’s the question of whether consumers truly have an appetite for these technologies. A temperature test suggests yes. We’re seeing an unprecedented amount of confidence when it comes to asking ChatGPT questions about health, with 40 million daily health queries despite the well-known misinformation risks.

That’s a significant number of people becoming comfortable sharing health data with AI and expecting it to give useful, personalised answers. Perhaps that’s the natural precursor that makes handing over the reins to an AI agent feel less of a leap. 

Without meaningful uptake data on platforms like Amazon Health AI, we simply don’t know yet whether AI health agents will be met with receptivity or resistance. But with big tech pushing ahead with agentic development, we can expect to see more product launches like Amazon’s Health AI in the coming months. 

The honest answer is that nobody needs to overhaul anything today. But there’s a difference between not acting yet and not knowing where you stand. Most healthcare companies are firmly in the latter camp – even in sectors where AI agents are just around the corner. 

If you’re an online pharmacy or OTC retailer, the use case is closer than you might think. Agentic commerce can research, recommend and purchase products on a user’s behalf, and are a natural fit for supplement subscriptions and OTC purchases. 

The infrastructure these agents rely on, like product feeds, structured data, transactional flows, is much the same infrastructure that powers conventional ecommerce, but with additional protocol and compatibility requirements on top.

The sensible move is to get under the bonnet of your data architecture now, before it becomes urgent.

For service providers, like clinics, diagnostics, specialist referral pathways, the timeline for implementation is less clear. The Amazon Health AI model shows where things are heading, but widespread agentic booking across independent providers is still some way off. 

But, it’s worth noting that the foundations of agentic readiness (clean structured data, accurate schema markup, machine-readable user flows), are the same foundations that improve your performance in conventional SEO and generative engine optimisation.

Wherever you sit, there are a few practical steps worth taking now — primarily, auditing how your website performs for non-human visitors. See our blog for a breakdown on what healthcare websites need to do to prepare for agentic AI.

Beyond the technical, it’s also worth starting internal conversations now, and identifying who owns important questions – for example, the compliance implications for your organisation if AI agents begin accessing your services on patients’ behalf.

Our Agentic AI in healthcare Readiness Audit gives you a clear and specific picture of where your health website currently stands, and exactly what it would take to be agent-ready. 

We test your set up and data architecture against the latest Universal Commerce Protocols, which are the emerging standards that govern how AI agents interact with commercial and transactional platforms. Specifically, we look at three things:

1. Protocol compliance – whether your website supports the UCP standards that allows agentic platforms to physically make a transaction.
2. Agentic product feed compatibility – can AI agents actually see your products, services, pricing, availability and booking rules? Or are they hitting a wall?
3. Structured data health – a thorough review of your schema markup to make sure agents are being served accurate, complete information at every touchpoint

The result is a clear picture of where you stand — what’s working, what isn’t, and what would need to change if and when agentic healthcare goes mainstream. 

These read-outs are also substantially useful for general AI visibility strategies, which make sure your health brand is discovered and accurately cited by AI search tools. 

If this is something you’re starting to think about, our readiness assessment will give you a clear and practical picture of where you stand. Get in touch to get started.