University of Utah Health’s Kawamoto Says Workbench Approach Scales Clinical AI Across the Enterprise

University of Utah Health’s AI strategy hinges on a newly created Chief Health AI Transformation Officer role, which centralizes AI initiatives across clinical care, research, and education. By leveraging existing standards like FHIR and Smart on FHIR, the team embeds AI directly into Epic’s electronic health record, avoiding costly custom builds. This standards‑based integration, combined with a rapid‑prototype "AI Workbench," lets clinicians generate custom prompts and analytics in days rather than months, accelerating the path from idea to usable tool.

A key cultural shift underpinning this technical agility is governance. Leadership moved from monthly to weekly decision‑making cycles, ensuring AI projects receive timely oversight without stalling. Pilot programs are deliberately scoped with engaged frontline users, providing continuous feedback loops that surface safety concerns and usability issues early. This disciplined yet flexible approach balances the need for rapid innovation with the realities of entrenched clinical workflows, preserving operational stability while introducing transformative capabilities.

The explosion of large language models has reshaped predictive analytics, allowing the health system to build mortality and length‑of‑stay models in a matter of days—far faster than traditional two‑year research cycles. This speed, however, challenges conventional academic publishing, pushing clinicians to rely on real‑time sources like X, LinkedIn, and community benchmarks. For health executives, the takeaway is clear: adopt standards‑based AI platforms, streamline governance, and embed rapid pilots to stay ahead in a landscape where AI advances faster than any textbook can capture.