One of the challenges for healthcare journalists this year is to use AI to determine where they are on the Gartner Hype Cycle. We are firmly beyond the “innovation trigger,” where breakthrough advances increase interest, and we appear to be squarely on the cusp of “heightened expectations,” where initial hype can lead to unrealistic expectations. We have not yet reached the valley of disillusionment. It will probably come.
And nearly every vendor product announcement describes how their solution is “AI-driven,” making it difficult for reporters to tell the wheat from the wheat. At Healthcare Innovation, we have and will continue to take an approach that aligns with health system CIOs, CMIOs, and chief AI officers as they prioritize use cases and solutions and quantify measurable quality and efficiency gains. We also try to interview venture capitalists and consortium leaders focused on AI startups in the healthcare space.
This month, we announced our annual package of 10 transformational trends. This year, we dedicated the entire package to AI, from governance and regulation to how AI will impact areas such as oncology, revenue cycle, and cybersecurity. While we encourage you to spend some time with this package, we'd like to take a look back at how the use cases are evolving and what some health tech executives and investors predict will happen next.
In October, venture capital firm Menlo Ventures released a report on AI adoption in the healthcare sector. According to the study, 22% of healthcare organizations are using domain-specific AI tools, a 7x increase from 2024 and a 10x increase from 2023. Health systems have the highest adoption rates at 27%, followed by outpatient providers at 18% and payers at 14%.
The $4.9 trillion healthcare industry, which accounts for one-fifth of the U.S. economy but just 12% of software spending, is currently deploying AI at more than twice the rate (2.2x) than the overall economy, according to Menlo Ventures. Spending on healthcare AI this year reached $1.4 billion, nearly triple the amount invested in 2024.
Two categories that address urgent operational challenges and deliver measurable ROI are ambient clinical documentation ($600 million) and coding and billing automation ($450 million). Other fast-growing categories include patient engagement (+20x year over year) and prior authorization (+10x year over year).
In September, I interviewed Murray Brodzinski, a partner at Aegis Ventures, a New York-based venture studio that partners with entrepreneurs and health systems to launch and scale health tech startups. He noted the ebb and flow of enthusiasm around AI. “We all got excited about the zeitgeist of ChatGPT, but when it came time to implement these things, we went through a bit of a trough of disillusionment,” he said. “But that's definitely what everyone's thinking about most: How do we apply generative AI or agent AI to workforce problems? We absolutely think that's a top priority. How do we implement it for efficiency and effectiveness? Historically, the focus has been on efficiency. How can we do what we're doing better? We feel like there's a lot of opportunity there, but we actually think there are problems that we couldn't solve before that we can revisit now.”
In June, I spoke with Amy Trainor, RN, CIO at Ochsner Health, about how the health system is expanding its relationship with a company called Latent to develop clinical AI tools to support pre-authorization for specialty, infusion, and retail pharmacies.
During that conversation, Mr. Trainor said something that intrigued me. “What we don't want to do, and what we don't want to do at Latent, is have a bot war between the payer's AI and our AI and see who wins. That's not a success for the patient, it's not a success for the payers, it's not a success for us as an organization.”
A few weeks later, I discussed the same topic with Fawad Butt, founder and CEO of startup Penguin Ai and former chief data officer at UnitedHealthcare, Kaiser Permanente, and Optum. His company says its flagship platform combines task-specific small language models (SLMs) and digital workers and agents with a healthcare-specific AI platform to streamline processes such as prior authorizations, claims processing, medical record summarization, and appeals management.
I asked Mr. Butt about this idea of a battle between AI agents on the payer and provider sides.
“That war has begun,” he said. “The agency war is here. It's not some futuristic thing that's going to happen. It's happening today. I sat down with the CEO of one of the largest community health plans in the country, and what he's seeing is that, in some ways, the provider side is adopting agents much faster than the payer side. The process on the payer side… Because it's more complex. In some scenarios, he said, a small network of providers that used to appeal 5 percent of denials is now doing well.''He said he believes the health plan has eight representatives who appeal 100 percent of every denial the group sends them. So how do they win it? ”
Monitor AI performance
In 2025, many health systems continue their efforts to establish AI governance teams and processes, and we reported on several startups formed to support those efforts. They include Vega Health, whose CEO is Mark Sendak, MD, who previously served as director of population health and data science at the Duke Institute for Health Innovation. His company partners with health systems to implement solutions in local environments and objectively monitor performance. Vega Health also works with health systems and innovators to commercialize effective AI solutions across the healthcare ecosystem.
Another company entering this space is Qualified Health. In a June interview, Kedar Mate, MD, co-founder and chief medical officer of the company, explained how the company is helping health systems build infrastructure to support their generative AI efforts.
Mate, former CEO of the Institute for Healthcare Improvement, said the initial approach to governance in most health systems was to have a technology governance committee, but now they have subcommittees focused on generative AI or large-scale AI solutions. “We call it analogue governance for analogue technology in an analogue era. But we need true digital governance,” he stressed. “By their very nature, these AI tools require digital governance. The underlying underlying models evolve over time, which can cause specific applications used in your institution to perform better or worse, and you may not know it unless you regularly monitor the performance of those algorithms.”
He added that one of the company's assumptions is that automation will be the basis for driving productivity improvements. “Competitiveness in the market will be determined in part by how we deploy AI in the future. If we can deploy AI faster and better than the systems on the other side, we will have a better chance of cornering different aspects of the market in the future,” Mate added.
Returning to my interview with Ochsner's Amy Trainor, I asked her whether she feels that large, established companies, such as the major EHR vendors, are moving fast enough with AI tools.
“Six months ago, I would have said no,” she replied. “I don't know what happened there, but it seemed like a big shift. For most things, we look to our partners first. If they don't meet our needs, we look elsewhere. Same thing with AI. We want to make sure we understand the value of what we're doing. Generally speaking, we look internally first. I think they're moving a little faster. Epic hasn't done this yet, but they probably will in the next six months.
“Enabling AI is a completely different conversation than changing the color or a new button on someone’s dashboard. We need to make sure this is safe and follows responsible AI principles,” Trainor added. “I would say the ChatGPT era is the fastest-moving era of public technology in my 20 years of working in health IT.”
Look for AI development to move even faster in 2026.
