Epic doubles down on AI growth with rapid capability expansion

LAS VEGAS -- Epic's appetite for AI is only growing, as exemplified by the company's rapid expansion of AI capabilities integrated into its EHR. At this year's HIMSS conference, Epic shared metrics on the success of its core suite of AI tools and doubled down on plans to expand AI across a variety of healthcare workflows.

AI has once again been a prominent topic of discussion at HIMSS, with vendors and health systems embracing the tech at rapid speeds. As an EHR market leader, Epic is uniquely positioned to meet the demand by quickly developing and deploying AI tools at scale, leaders suggested.

"Really, we started this ten years ago. We built a cloud-based platform that could handle AI workloads," Phil Lindemann, vice president of data and research at Epic, said in an interview.

When generative AI tools like ChatGPT first debuted in 2022, "we had hundreds of customers already in a place where the technology was there to adopt it and scale it," Lindemann said.

"We've been working with AI long before genAI," added Jackie Gerhart, M.D., chief medical officer at Epic.

"So, from our perspective, it's always been innovate, innovate, innovate, go, go, go, try to make sure that we're really trying to decrease the burden. We started with clinicians because we know that part of their job is to continue to see more patients, but really we want to make sure that they're joyful about doing that and that they're providing good patient care."

Epic appears to have no plans to slow its AI expansion, with more than 85% of its customers now using Epic AI in some capacity.

Epic reports success metrics for core AI suite                                      

Epic's core AI suite consists of Art for clinicians, Penny for revenue cycle and operations, and Emmie for patients. According to data Epic showcased at HIMSS, Art is enabling clinicians to complete discharge summaries 20-30% faster.

At The Christ Hospital in Cincinnati, Ohio, Epic said that Art drove early detection of lung cancer at a rate of 69% versus the national average of 46%.

"We have an extracted findings tool for chest x-rays specifically, such that if you get a report back that says to follow up because there's an incidental finding, AI can read that and then not only put that into the patient's record, but also schedule an appointment to get that follow-up," Gerhart said.

"So, an actual patient outcome that I think is really meaningful, in addition to, of course, the reduced cognitive burden that people will have throughout using our tools."

Epic also said it plans to continue its rollout of Chart with Art, a built-in charting tool that drafts clinician notes and queues up orders. It launched in February across several outpatient specialties and will reach home care workflows in April.

At New Jersey-based Summit Health, Penny reportedly cut medication prior authorization submission time by 42%.

At other health systems, coding denials dropped by more than 20%, Epic said in a press release. Future iterations of Penny aim to complete coding session and appeal submissions independently.

Emmie has yielded high satisfaction rates, Epic also noted, with Louisiana-based Ochsner Health reportedly saving 750 hours of staff time.

Ask Emmie, a conversational AI assistant in MyChart that answers patient health inquiries within the context of their medical record, will be available more broadly in the coming weeks.

Emerging Epic AI capabilities

At HIMSS, Epic also highlighted expanded AI capabilities that were recently unveiled or on the horizon, including Agent Factory, an integrated platform for creating custom AI agents. Custom agents can be supplied with the organization's policies and knowledge bases and deployed on the organization's own schedule, Epic said.

"It can reach out to other systems, and it allows you to orchestrate and dream up new workflows that still have the guardrails and the safety of an Epic workflow so that things don't go off the rails," Lindemann said.

Empowering organizations to experiment with AI in a contained environment could facilitate new innovations that could be scaled across the community, Gerhart said.

This is a time of unlocking this creative potential that isn't just owned by software companies anymore.

"It feels to me like this is a time of unlocking this creative potential that isn't just owned by software companies anymore. It can really be owned by people who are doing the work," she said.

Curiosity, another emerging tool in Epic's AI suite, is a family of medical foundation models trained on anonymized patient records. Researchers gained access to Curiosity in a virtual lab in February 2026 to test new use cases.

"With Curiosity, you can take many real-world patient journeys and apply them to the patient that's in front of you and then help predict and understand what the most likely next medical event will be," Gerhart said. "And so, for me, that means probably shorter time to diagnosis, probably shorter time to specific treatment, maybe even keeping someone out of the ER."

Other planned expansions include collaborative visit agendas, a diagnosis checker, which analyzes patterns in patient systems and suggests next steps and conversational AI for clinicians.

Another area ripe for future AI innovation, Lindemann noted, is clinical trials.

"About 80-85% of the groups that we work with run clinical trials as an extension of the care that is offered at those facilities and making it easier to match patients with a trial that might work for them is an age-old problem," he said. "It's because a lot of what makes a patient eligible for a trial is tucked into notes or it's in different places where frankly the team didn't have the time and energy to set those things up."

Linemann sees AI filling that gap, helping organizations find patients and enabling scalability.

Emphasis on iteration, governance

Lindemann and Gerhart emphasized that Epic is continually iterating on existing AI tools to ensure they deliver value while maintaining safety and data quality. Software developers are required to complete immersions, during which they visit a site, understand the workflow, and rapidly iterate on feedback.

"The goal is to try to solve what our customers need as their biggest problems rather than just create software to create software," Gerhart noted.

While Epic has been rolling out new AI capabilities at a rapid pace, Lindeman added that actually building the tool is just one small part of the process. Evaluating the capability and assessing model drift and potential outcomes are crucial components that cannot be overlooked when responsibly creating an AI tool in healthcare.

"What we're finding is we are going to spend more development hours, more energy and time on telemetry, auditing, tracking and evaluation of AI than we actually spend building it," he said. "So, the infrastructure to scale and support AI is more effort than making the shiny thing. And that's often a really inconvenient truth for people to realize."

Despite Epic's enthusiasm for AI, Gerhart and Lindemann acknowledged that AI is not necessarily the answer to all of healthcare's problems.

When thinking about governance, Gerhart recommended that organizations first evaluate the problem they are trying to solve, and second, consider whether it is actually a good case for AI.

"Don't use AI as a hammer if you have a better tool that you can use," she said.

Still, Epic's expansion of its AI capabilities demonstrates that it sees AI as part of the solution to many clinical and administrative woes, and a valuable companion to patients.

Jill Hughes has covered health tech strategy since 2021.