Health Systems Tighten AI Playbooks, Retire Misfiring Tools as Governance Becomes Standard
Leaders now demand measurable results with human oversight.
Overview
- Health systems reported sunsetting or switching AI tools that underdelivered, including FMOL Health canceling a billing coder pre–go-live, Baptist Health South Florida dropping a virtual navigator for mis-triage and low use, and City of Hope replacing commercial CDI AI with an internally built solution over privacy concerns.
- Others tightened scope rather than retire, with Mahaska Health ending isolated pilots and creating a Data Science and AI Center of Excellence, MUSC Health choosing Epic’s native denials AI over a niche vendor, and John Muir and Nationwide Children’s rebooting or slowing pilots to fix process and security gaps.
- CIOs described governance as a required operational discipline alongside clinician-led workflow redesign and reliability engineering, framing uptime, identity controls, and EHR integration as core to patient safety.
- Programmatic milestones underscored the operational turn, including UMass Memorial’s 3,000th Hospital-at-Home admission, Providence’s rollout of a proprietary network telemetry platform, and Billings Clinic–Logan Health selecting Oracle Health as its EHR.
- Pediatric and large systems cast AI as an enterprise capability under tight guardrails, with CHOP defining AI-driven transformation and Children’s National centering foundational data governance, as leaders position 2026 to scale what works and retire what does not.