Your patients are already talking to AI about you | Viewpoint

Most health systems still think the patient journey begins when they enter our doors. It doesn't. It begins on a phone, with an AI, hours, or days before we know the patient exists.

By the time someone walks into one of our clinics, they've likely had a conversation about their symptoms with a chatbot. Perhaps they asked it what their lab values mean. They've had it read their imaging report. They've asked it whether their surgeon is any good. They've arrived prepared, and the preparation didn't come from us.

This is happening now, at scale, and faster than I expected. The patients aren’t bleeding-edge tech users. They're working professionals, retirees, parents of sick kids, people who couldn't get a call back from a nurse line and turned to the tool that answered.

Many health systems are talking about how to teach patients to use AI safely. That conversation is already obsolete. Patients aren't asking permission. The strategic question isn't whether to let AI into the patient journey, it's whether health systems will earn a seat in that conversation or risk being bypassed while we debate who governs what and when. I am concerned most of us risk being on the wrong end of that equation.

The AI assistants that patients are using are optimized for engagement. The next generation will be optimized for advertisements, affiliate revenue, and referrals to strategic partners. None of those business incentives align with what patients need most: accurate information and coordinated care driven by best evidence.

If a health system does not show up as a trusted source inside those systems, even their better outcomes may not be enough to avoid a potential patient being routed to a provider that’s wrong for them. Health systems will be invisible to the layer that now sits between patients and care.

This is the part of the conversation that's missing. We talk about AI as a tool we deploy – we need to start talking about AI as the infrastructure patients are already using to navigate us.

AI may become the first stop for information, but trust is still built between humans – patients and clinicians. The organizations that succeed will be those that combine the accessibility of AI with the expertise, judgment, and accountability that only human healthcare professionals can provide.

So what should health systems actually do? Three things:

Build for AI visibility

The same way we used to build for search engines, we now have to build for answer engines. Structured clinical content. Accurate provider data. Transparent outcomes. Direct partnerships with the major assistants. If our information isn't surfaced by the model, we don't exist to the patient. This is the new front door, and most health systems haven't realized they need to put their sign on it.

Govern for trust

Be transparent about where AI is being used. Keep clinicians responsible for decisions that affect patients. Regularly evaluate AI tools for accuracy and fairness and be willing to share the results. Trust is one of the few advantages healthcare organizations can't afford to lose. The first time an AI tool gives a patient bad advice and we can't explain why that trust is immediately eroded.

Use AI to absorb the demand

At Inova, we expanded our infusion capacity using AI-enabled scheduling and workflow design and increased daily infusion volume from approximately 160 to 210 patients. Same staff. Same physical footprint. AI in healthcare isn't going to be a story about doing the same volume more cheaply. It's going to be a story about finally serving patient demand we didn’t know was there and have been quietly turning away for decades.

Health systems don't have the luxury to set the pace of AI adoption. The AI is already in the room. Our patients have already started running with it. It’s just a question whether we will choose to keep pace.

Matt Kull is chief information and digital strategy officer of the Inova health system.