"At Microsoft, we believe AI can shape a future for all people on the planet to live healthier," said Joe Petro, corporate VP of Microsoft health and life sciences solutions and platforms. "Our ongoing collaboration with Epic supports our vision to help transform the healthcare experience through innovation that delivers meaningful outcomes."

According to an Epic spokesperson, the company's "native AI charting is being developed, and we anticipate it will be ready for limited use by early next year."

"Epic made several big announcements around new products such as the CTMS and ERP, eliminating the need for bolt-on third party solutions. They have already begun to replace these over the past few years with new products such as Gallery and Hello World. It will be interesting what happens to the future of those companies and how they are able to compete in the market," said Katie Dunnivant, director of provider technology services at Optum Advisory.

According to Carol Chouinard, VP of provider technology services at  Optum Advisory, "Epic just made it clear it is not simply an EHR company anymore, it is building the operating system of healthcare. The announcements in Verona centered on three big shifts.

First, AI is no longer theoretical. Epic is rolling out a native AI scribe in partnership with Microsoft, embedding ambient documentation directly into the workflow. At the same time, Cosmos AI, a large language model trained on over 8 billion encounters, is opening new possibilities for predictive care and research. And through 'Art,' 'Emmie,' and 'Penny,' Epic is positioning AI as an everyday assistant for clinicians, patients, and revenue cycle teams. 

Second, the experience is getting smarter. MyChart Central, coming in November 2025, will allow patients to use a single login across all Epic organizations, a long-awaited step toward unifying the patient journey. Meanwhile, a redesigned clinician and patient UI, scheduled for 2026, will weave AI directly into the visit flow to support decision-making and reduce friction.

Finally, Epic is expanding well beyond the EHR. With EpicOps ERP, the company is moving into workforce, supply chain, and finance management, and its new clinical trials platform will bring research and care closer together starting in 2026." 

"The message was clear: Epic is no longer asking 'How do we digitize healthcare?' The real question now is, 'How do we redesign it?,'" Chouinard added.

"Regardless of your role, embracing AI is no longer optional — it's essential. The time to prepare and engage with these tools is now," Watts noted.  

Epic's annual meeting: 1 lesson learned (and 2 remaining questions)

By Ty Aderhold

Takeaway: Ambient AI is a step, not an endpoint. 

Although much of the pre-UGM hype was focused on Epic's announcement of their own ambient AI scribe, they didn't use the term ambient once. Instead, Epic very purposefully kept the focus on their larger suite of clinician-focused AI capabilities, particularly those that are enabled by ambient documentation. As the vendor market continues to grow, Epic and others in the ambient space are focusing on capabilities beyond note generation.

My biggest takeaway from the announcements and discussions around clinician support tools? Ambient documentation is a useful first step. But actual transformation of clinical workflows is going to depend on how effectively the documented information can be combined with other data and then used across a variety of processes — from chart summaries to in-basket response to discharge instructions.

Question: Are AI agents a realistic answer to existing technologic inefficiencies?

The evidence suggests both Epic and Oracle, which recently announced a new AI-driven EHR solution, believe the answer is yes. Time and again across the four days, conversations about AI and return on investment turned towards Epic's AI agents' ability to provide efficiency gains for administrative staff, clinicians, and patients.

But is simply augmenting (or even automating) a series of bad processes the right step forward? Historically, the answer to this question in healthcare has been no. But with organizations facing continued financial and workforce challenges while AI-powered solutions continue to improve, we might start to see more and more organizations choosing to automate inefficient processes.

Question: Are large medical models the path forward for clinical AI and personalized medicine?

Rather than ambient listening or AI agents, I think the biggest announcement Epic made at UGM was actually the announcement of the new Cosmos Medical Event Transformer (CoMET) models.

Cosmos is Epic's huge de-identified data set, which has upwards of 300 million unique patient records.  Epic, along with researchers from Yale, have now trained what is essentially a "large medical model" — built similarly to a large-language model (LLM). Instead of words being the training data and output like an LLM, health events are the training data and output.

LLMs work because they are extremely accurate at predicting the right next word and then putting those words into a sentence and paragraph. Similarly, this model could theoretically predict future health events, such as readmission, for a patient. Future use of big data models like this could have powerful implications for many parts of the industry, from clinical decision-making at the point of care to care management to coverage determinations.

*Advisory Board is a subsidiary of  Optum. All Advisory Board research, expert perspectives, and recommendations remain independent.