AI provider search is here. How can health orgs stay visible?

Understanding how AI provider search works -- for now

Currently, AI agents and chatbots can only draw on search as their foundational layer, according to Desai, meaning what appears online and in search results feeds the ChatGPTs and Geminis of the world.

Although many large language models are working on more sophisticated knowledge graphs, they are still very much grounded in search data.

That means healthcare organizations need to be aware of what information they make available to search engines, so they can also determine what's available to AI.

According to Desai, that information can fall into one of three categories: what organizations make available via their websites, what's said about organizations on online provider review websites or a combination of the two if the organization's website supports patient reviews.

Providers must pay attention to their websites, Desai stressed.

"Organizations need to ensure that they have the right information on their website, whether it is the hours of operation, the address or the location where somebody practices."

They also need to understand what's being said on other websites, as AI engines typically pull information from multiple sources to ensure reliability.

Take, for example, the query mentioned above asking for a cardiologist available at 4:30 p.m.

The practice's website confirmed the provider's name, specialty and location, but it's those online reviews that confirmed that the provider is a good fit. In that example, Desai said reviews mentioning later appointment times probably helped AI produce its response.

"Organizations also need to ensure that there is a good amount of user-generated content, such as provider reviews, showing up on their websites," he explained. "As far as trust is concerned, the one thing that we're seeing across these AI engines is that each of them will look at different sources of data when it comes to healthcare. They all start with the provider website, but they will include other sources."

Future-proofing for AI provider search

As noted above, many of these principles mirror best practices for appearing in search engines. That's a good start, Desai noted, but organizations that appear in AI provider search will also have safeguards built in for the future.

For example, many AI chatbots already provide, or are working on, agentic features that help patients book care. But to do that, the AI will need to understand how to work within an organization's website architecture to reduce friction for the patient.

"Look at how agents look at your website and your content, and start evaluating that, because the traditional metrics won't apply in a couple of years," Desai recommended.

Historically, there's been a strong emphasis on human factors and human-centered design to understand how people see and navigate a provider's website. In the future, it'll also be important how an AI agent experiences a health system website, Desai predicted.

Organizations should start asking themselves if AI agents would be able to find the right information and whether they have enough structured data to tell the agent what it needs to know.

Health systems should also stay abreast of evolving search engine capabilities, including on Google and Bing. For example, Google recently introduced a button that providers can add to their Google profiles that lets users schedule an appointment, Desai said.

"If you connect the dots enough, that scheduling button is going to become the way you schedule appointments through Gemini in the future," he predicted. "Opt into that, because you want your information to show up."

AI has been disruptive, both for better or worse. But Desai maintained that in this case, it's for the better.

"The opportunity for the health system is having a more educated patient when they're walking through the door," he asserted.

Desai doesn't just mean patients who come prepared with more information about their symptoms or a list of questions for their clinicians that AI helped them write. He also means the ability for patients to come prepared administratively.

With the right information on a provider's website, AI provider search tools could start preparing patients for registration or simply remind them to arrive at their appointment 15 minutes before it starts. That could be a boon for an industry often defined by its complexity.

"Even if the healthcare institutions try -- and they've been trying really hard to ensure patients are educated -- everyone has different levels of proficiency when it comes to health literacy," Desai said.

Better-prepared, more informed patients are equipped to navigate medical care with confidence, which Desai said is invaluable as medicine continues to invest in better patient experiences.

"Where these agentic experiences, these assistive experiences have an opportunity is to lower the barrier to what it means to be an educated patient," he concluded. "For the health system, the chances of delivering exceptional experiences go up by doing some upfront work on the consumer side."

Sara Heath has reported news related to patient engagement and health equity since 2015.