Stephen O'Mahony, MD, SVP and Chief Medical Information Officer at RWJBarnabas Health, discusses the clinical adoption work behind an AI deterioration warning tied to an 18% reduction in the risk-adjusted odds of in-hospital death. He details the concept-to-data mapping done with the technical team, the suppression logic that keeps the system quiet when clinicians already know, and the silent background testing that cut daily alerts to a volume rapid response teams could absorb.

Tinu Tadese, MD, System VP/CMIO at Boston Medical Center Health System, discusses Project Unite, the six-month integration that brought two former Steward Health Care hospitals onto a mature Epic build. She explains the MVP+ approach that lifted an academic build into community settings, the day-one data decisions that kept patients safe, and why conversion time is the constraint an acquisition timeline cannot compress.

Julie Demaree, VP, Chief Technology and Digital Innovation Officer at St. Mary's Healthcare, Corey Zeigler, CIO at Helio Health, and Carrie Weigand, MD, clinical advisor at Heidi Health, discuss how a lean IT department gets ambient AI into clinicians' hands. They cover buying at-the-elbow support instead of staffing it, the customization decisions that break vendor alignment, peer champions as the fastest training path, and who owns an AI model once it is live.

Chris Paravate, EVP and chief digital and information officer at Northeast Georgia Health System, discusses the five-week rollout of Epic’s Chart with Art to nurses and patient care techs at five hospitals. He explains why a tool that fails safely earns a lighter implementation, how 88% adoption in 30 days replaced a business case for the AI shift-handoff summary, and where operations must set the standard for mandatory AI use.

Cassie Marcelle, VP of clinical informatics and CNIO at UF Health, Marc Perkins-Carrillo, CNIO at Moffitt Cancer Center, and Lisa Stephenson, CNIO at Cedars-Sinai, discuss why ambient AI built for physician notes cannot simply be pointed at nursing flowsheets. They cover device platform gaps, nurses who use discrete fields as a checklist, super user programs with paid training hours, and governance built to remove flowsheet rows.

J.D. Whitlock, CIO at Dayton Children's Hospital, breaks down the ambient AI decision waiting at the end of his Abridge contract and the test Epic's Chart with Art will have to pass. He explains why nursing ambient means a second vendor and a second bake-off, what reported take-private talks at Workday would do to the EpicOps timeline, and how 20 clinical governance committees keep the call out of his hands.

Allison Dempsey, PhD, PMH-C, Director of Technology and Data Governance at the Brain and Behavior Innovation Center at CU Anschutz, discusses why behavioral health met AI’s consent and privacy questions before the rest of the enterprise. She explains the response cost test she applies to every technology request, how Colorado’s new psychotherapy AI law added a mid-session consent step to ambient documentation, and what a one-page request does to governance cycles.

Shauna Overgaard, PhD, Senior Director of AI Strategy and Frameworks at Mayo Clinic, discusses an emerging approach that tests selected AI tools in the simulation center, with standardized patients, before clinical use. She describes the workflow shadowing that changed how Mayo's review board weighs proposals, the exploratory work on bringing clinicians into evaluation, and a reliability cue in development for patients a model may serve less well.

Susan Harkness Regli, PhD, Associate CHIO for Human-Computer Interaction at the University of Pennsylvania Health System, discusses how human factors work protects clinicians from AI tools that pass technical testing yet fail at the screen. She details Penn's staged alpha and beta exposure, the usage data that caught one role accepting AI-drafted patient replies unedited, and why governance approval sets a six-month review date.

Darin Prill, MBA, CHCIO, VP/CTO at WellSpan Health, discusses the 17-week migration that moved more than 300 applications and the full Epic environment to AWS ahead of a Broadcom renewal. He details the wave schedule that held Epic downtime to about two hours, the intake review deciding which third-party tools travel with Epic, and the data work that has to precede AI.