Clinical, operational, and technical staff from health systems are working together to turn their ideas into agents that can help improve patient outcomes, increase patient engagement, reduce administrative burden, and more.
Teams from 25 health systems are in Verona this week for the second Agent Factory Build-a-Thon, where each team is building an AI agent that can carry out a complex task from start to finish.
Agent Factory is Epic’s platform for health systems to build, monitor, and evaluate AI agents. For example, using Agent Factory, ECU Health built an agent that summarizes incoming transfer requests and identifies which facilities can take each case—saving 20 hours of staff time each week.
Each health system’s Build-a-Thon team pairs clinical and operational leaders with technical staff, and Epic developers work alongside them. “Someone with direct care experience sits right next to the person building the agent, saying ‘that’s not quite it, try this,’” said Derek DeYoung, who leads Agent Factory development at Epic. “They try it, look at it, and try something else—all in hours, not months.”
For example, this week the health system teams are working on agents that can:
- Read nursing and physician notes for early signs of necrotizing enterocolitis—a serious intestinal condition in premature infants—and flag high-risk cases so the care team can review them sooner.
- Identify patients at risk of day-of-surgery cancellations, such as those who can’t get a ride, and reach out to help before the day of the procedure—helping reduce cancellations, unused OR time, and patient wait times.
- Check medication refill requests against clinical protocols and patient charts, then mark which ones are safe to approve and which need an appointment first—reducing chart review time for millions of these requests each year.
- Catch when critical care time hasn’t been documented, and draft a supporting note that a clinician can review and confirm right away instead of reconstructing it hours later.
Health systems are also building agents that work with Art, Epic’s AI for clinicians. One agent will review a patient’s medication history before admission and flag conflicts with new orders, so the care team can catch risky interactions before the first dose. Another team is teaching Art to assemble a timeline and draft a root cause analysis after a patient safety event to help investigators start fixing the gap sooner.
Participants spent Monday on brainstorming, and on Tuesday they’re building and testing. Wednesday is “show and tell”—the health systems will present to one another about what they’ve built.
Epic hosted the first Agent Factory Build-a-Thon in May 2026. Training for the first group of early adopters begins in October 2026, with broader availability planned for 2027.
