At its annual Users Group Meeting in Verona, Wisconsin, this week, Epic outlined its tech and AI strategy, signaling that the company has ambitions to be the operating system for healthcare.
What wasn't addressed on stage—and likely on the minds of many attendees—was recent leadership changes at the company following a key executive departure. Last month, former President Sumit Rana announced he was stepping down for personal reasons after 28 years at the company. Rana was long considered to be CEO Judy Faulkner's successor.
The company then announced that four executive leaders would take on expanded responsibilities to fill Rana's shoes when he leaves—Seth Howard, executive vice president of research and development, as well as Mark Lipsky, Erv Walter and Garrett Adams, all with the titles of senior vice president of R&D.
In the ensuing weeks, there have been media reports of other high-level executive departures at Epic.
In an interview with Fierce Healthcare on the sidelines of UGM, Howard said the decision to distribute Rana's responsibilities across multiple leaders, rather than replacing one-for-one, fits with Epic's leadership philosophy.
Epic employees tend to build unique careers and roles, so leadership transitions typically involve multiple people stepping into expanded responsibilities rather than a direct replacement, said Howard, a 19-year veteran at Epic.
"One of the things that I have really enjoyed at Epic, and I think a lot of people enjoy, is there is not a predefined career path, and so each employee invents their own pathway. It's not one ladder. It's you figuring out what's the combination of the things that you're good at, the things that you enjoy, and what our community and our company needs and finding that intersection," Howard said.
It's also Epic's philosophy that one of the most important responsibilities of a leader is to grow other leaders, he noted.
"Sumit took that responsibility very seriously, and I think he did a very good job of doing that. We have a monthly staff meeting and we celebrate anniversaries, major milestones. It's pretty amazing the number of of 20-, 25-, and 30-year anniversaries that get celebrated at that staff meeting. We're always looking at how do we grow leaders around us," Howard said.
Epic also focuses on developing leaders internally rather than hiring external executives.
"Judy [Faulkner] sometimes jokes, 'Oh, we'll start our search and find an executive search group that will identify a new person to fill the role'," Howard said, then laughed as he added, "Epic would never do that."
When asked how his specific role would change going forward, he said, "It's hard to describe exactly what any one of our responsibilities will be just because we're always taking on new things, taking on new challenges. I look forward to this next chapter and helping the Epic community continue to get the most value out of our software."
Many of Epic's major announcements at UGM were centered on its advancements with artificial intelligence—new AI tools for clinicians as well as workflow automation for patient engagement, workforce, payments, research and hospital operations along with capabilities for clinical decision-making at the point of care.
Howard views Epic's AI strategy as a continuation of a long-term roadmap, not a sudden shift, even as agile healthcare AI startups jump into the market to offer specialized tools.
"We've been pretty deliberate in building these capabilities. You heard last year even some of the foundational components. Curiosity was introduced last year. Art, Emmie and Penny were introduced last year. We've been extending the healthcare network to connect more parts of the ecosystem, and over the past year, you know things like Agent Factory have been developed," he said. "It's more of a deliberate continuation to build on the foundational capabilities that we have been building, not just for the last few years, but I would say for decades."
With the groundwork laid, Epic is now focused on building more advanced AI capabilities, and the company's advantage stems from its unified data platform, Howard asserted.
"The integration across applications, the approach we've taken where we build our software here. We don't acquire companies, and what that means is data is consistent. It's not duplicative. We don't have three different places to store date of birth. There's one date of birth. There's one set of allergies. There's one medication list. That data platform, the infrastructure, the agentic capabilities, the integrated workflows for people, and now AI agents, to do work together, really has been something we've been building towards," Howard said.
A big headline this week was the rollout of Epic's Agent Factory, which it previewed at the HIMSS26 conference in the spring. Agent Factory is Epic's platform for building, running and monitoring AI agents that can reason and act across workflows. It effectively serves as a development, automation and management platform for agentic AI.
As healthcare AI adoption rapidly grows, health systems increasingly want to customize AI tools for their own workflows and clinical protocols.
"It was a pretty natural evolution to say, our developers have access to these tools and are building agentic capabilities, and our customer community represents some of the most innovative health systems in the world, so it is not a big leap to think they will want to innovate on top of these types of agentic capabilities," Howard said.
Earlier this summer, Epic hosted an Agent Factory build-a-thon for health systems. "The team thought, 'Oh, maybe we'll get some interesting ideas.' They got over 1,000 ideas from customers. It's been amazing to see the interest from the community, and I'm really excited to see what the community builds," Howard noted, adding the expects customers to create and share agents across the Epic community, accelerating innovation collectively.
Epic is partnering closely with early-adopter health systems who want to be on the forefront of AI innovation and are willing to test new AI capabilities in real-world clinical settings, Howard noted.
"Healthcare is under a lot of pressure financially, from an access standpoint and workforce. We feel a great urgency to create useful tools for our customers to address those challenges, and AI is obviously a big component of that.
Those organizations help validate performance, refine prompts and evaluate reliability, generating evidence that gives the broader customer community confidence to adopt the tools.
"We learn together because what you learn in the field in real life in interacting with patients and interacting with clinicians you can't recreate in a lab. That part of the community that wants to move fast with us helped to actually solve the issue of how do we then validate and test for safety, for reliability, and then that provides the confidence to the rest of the community to say, OK, we've got a pathway now," he said.
The use of AI also is changing IT and software developer roles, he noted. This week, Epic unveiled a tool called analyst build assistant as an AI agent to help health system IT teams configure Epic EHR systems faster.
As AI automates routine coding, configuration and support work, that frees up analysts and IT staff to focus more higher-value tasks, he noted.
IT professionals will spend more time partnering with clinicians and operations teams to identify problems, design workflows and build custom AI tools. He said Epic has long emphasized these skills, hiring developers who understand user needs and manage projects from concept to deployment rather than simply writing code.
Earlier in his career at Epic, Howard was a software developer on the company's nascent anesthesia team, spending significant time in operating rooms observing anesthesiologists' workflows, decision-making and care handoffs to help design software around real-world clinical needs.
"Having that understanding, interacting with the users, I think you'll see as the field of software development changes, that's going to be the part that's most critical. As the effort to actually write the code continues to decrease, knowing what problems need to be solved, knowing what the right way to solve them, and being able to deliver that at scale will be the skillset that's necessary," he said.
Since last year's UGM, Epic's R&D teams have spent 23,000 days on clinical immersion trips to hospitals and health systems to observe real-world workflows, executives said at UGM.
Another key topic at UGM was Cosmos Curiosity. Epic leveraged Cosmos, its clinical research database, to build a generative AI model that uses real-world patient data to forecast outcomes such as hospital readmission and stroke risk. The model simulates patient health trajectories to predict clinical and operational outcomes. Cosmos combines clinical data from a community of Epic Systems health systems representing over 320 million patients and 23 billion medical encounters.
Epic plans to release Curiosity as an integrated component of its EHR software in March 2027, Howard noted.
Rather than forecasting a single outcome like readmission risk, Curiosity, as a generative AI model, is trained on longitudinal patient records and medical events, such as lab results, medications, admissions and discharges, to predict the next event in a patient's clinical journey.
"It's generalized to predict really anything that we have in the comprehensive database and that will help across a variety of different domains we think," Howard said. "There are other models that predict a readmission after an ED discharge. This can predict not just the readmission, but why the patient might be readmitted, the likelihood of each of those potential diagnoses, and then you could start to think through what changes do I make now that would reduce the likelihood."
As Curiosity is integrated into Epic's EHR software, it could function as a personalized clinical decision support tool, he noted.
"That's exactly the goal. You would feed it the patient's medical record and it would then predict for that specific patient what might happen next and that could help clinicians make better decisions," he said.
Epic's broader message to the provider community is that AI is now operationally viable, not experimental.
"AI is something that they can take and apply to their workforce, their operations, their clinicians, their patients, and see real practical outcomes," Howard said.
Related stories:
Epic expands AI ambitions with agent platform, Cosmos-powered predictions and workflow automation
Epic's AI scribe goes live as EHR giant touts strong adoption of built-in AI features
HIMSS26: Epic expands AI roadmap, previews Factory to build and orchestrate AI agents