AI moves from pilots to operations: Insights from HIMSS26
Artificial intelligence is leaving the lab and getting real work done inside hospitals. Speaking at HIMSS26 in Las Vegas, HIMSS President and CEO Hal Wolf said AI is already delivering results in operations, while most systems are still figuring out how to turn promise into consistent outcomes in care delivery.
Progress is clear in targeted areas. The bigger lift is translating tools into redesigned workflows, workforce readiness and strong governance.
Where AI delivers today
"If you're looking inside hospital operations, AI is being effectively deployed in efficiency - bed utilization, supply chain, staff utilization," Wolf said. These tools analyze operational data and surface decisions that improve throughput and resource use.
But there's no instant fix. Many organizations are still reshaping processes and workforce models to actually capture the value. As Wolf put it, "I understand the frustration that it's not a silver bullet."
Adoption is picking up
Two years ago, fewer than 5% of organizations had AI in production. Now, capabilities are showing up as embedded features in enterprise platforms. We're still early on clinical decision support, but the operational scope is expanding fast.
Governance and data quality come first
Wolf's warning is blunt: "With any new tool that's introduced, it needs to be objectively looked at by governance before the tool is used." Bad inputs taint outcomes. "Bad data in, bad information out."
Treat evaluation, safety and quality assurance as a product discipline, not a committee checkbox. Define approval gates, monitoring, and rollback plans before go-live.
Redesign the work, not just the tech
Technology without workflow change doesn't stick. "If you're going to make the investment but not redesign processes to take advantage of it, then it fails because we don't adjust our people or our processes," Wolf said.
- Stand up cross-functional AI governance (nursing, physicians, IT, operations, quality, legal).
- Prioritize a short list of high-value use cases with clear baselines and KPIs (LOS, throughput, denials, overtime, readmissions).
- Map the current workflow, remove steps, and integrate decisions into the EHR and team communication tools where work actually happens.
- Upskill clinicians and ops teams; create super users and competency checks tied to each use case.
- Pilot with guardrails, compare to baseline, then expand based on safety, equity and ROI.
- Vet vendors: data provenance, bias controls, model update cadence, human-in-the-loop, monitoring, and fail-safes.
Nurses at the center of digital work
Nurses sit closest to patients and to daily clinical systems. Their feedback decides whether a tool reduces burden or adds friction. Include bedside and charge nurses in discovery, design and rollout.
- Reduce cognitive load: fewer clicks, clearer recommendations, fewer interruptions.
- Target real pain points: documentation time, care coordination handoffs, admissions and discharge flow.
- Quick wins: ambient documentation with human review; bed management suggestions visible to charge nurses; staffing forecasts tied to patient acuity, not just averages.
Interoperability will decide scale
Point solutions can solve narrow problems but struggle to spread. Wolf is optimistic that better interoperability and AI-driven data integration will make it easier to connect and scale what works across systems. That ease of adoption ultimately benefits patients too.
Standards matter here. For context, see HL7 FHIR for data exchange basics that underpin scalable integrations.
What to execute this quarter
- Pick one operational use case (e.g., bed flow). Run a 12-week sprint on two units with a defined baseline.
- Formalize an AI governance rubric (safety, privacy, equity, explainability, escalation).
- Do a data quality audit for the chosen use case (completeness, timeliness, bias checks).
- Hold weekly co-design sessions with bedside nurses and operational leaders.
- Publish KPIs and learning updates every two weeks to build trust and adoption.
- Plan training, change management, and a measured path from pilot to production.
If you're leading strategy or execution, this resource can help organize the work: AI for Healthcare.
Governance reference
For a practical governance frame, many teams look to the NIST AI Risk Management Framework to shape risk controls and monitoring.
Bottom line from Wolf: digital transformation succeeds when it equips people to make a better decision tomorrow than they did today. Nothing about healthcare is easy-so keep the focus on data quality, workable workflows and the clinicians who live in them.
Your membership also unlocks:
