HIMSS26: 4 Strategies Hospital Leaders Are Deploying to Scale AI

Health system executives at HIMSS discussed how they're embedding the technology into workflows while maintaining clinician trust.

As AI adoption accelerates across healthcare, hospital leaders believe the primary challenge is figuring out how to deploy the technology in ways that tangibly improve care delivery and clinician workflows.

In various sessions at HIMSS26 in Las Vegas, executives described the internal strategies their health systems are using to operationalize AI, from empowering frontline staff to experiment with new tools to embedding AI directly into clinical workflows.

Here are four approaches they shared.

Start with operational friction points

One of the most consistent strategies executives highlighted was to begin AI deployments with operational or administrative challenges before tackling complex clinical use cases.

Intermountain Health vice president and chief health informatics officer Tamara Moores Todd said her organization initially targeted “back-end bureaucracy,” or processes that create administrative burden for clinicians but pose relatively low clinical risk.

Examples include documentation support, workflow automation, and other tools designed to reduce time spent on nonclinical tasks.

Leaders said these early use cases generate quick wins for clinicians while giving health systems time to build governance structures and evaluation frameworks before expanding AI into clinical decision-making.

The strategy is a reminder of the industry-wide reality that even though enthusiasm for AI is growing, providers must still prove the technology can improve efficiency without disrupting care delivery.

Let frontline staff drive use cases

Another lesson shared at HIMSS26 was that the best AI ideas often come from employees who use the technology every day.

Cherodeep Goswami, chief information and digital officer at Providence, described how his health system enabled staff to experiment with generative AI tools within a secure environment.

Rather than prescribing use cases, leaders encouraged employees across departments to explore how the technology could help them complete everyday tasks.

The approach quickly surfaced practical applications that leadership teams had not initially anticipated, especially around administrative tasks and information retrieval.

Several executives said this bottom-up model helps organizations identify the most promising use cases while building workforce familiarity with the technology.

Build AI into existing platforms

Executives also emphasized the prioritization of AI capabilities that integrate directly into existing platforms and clinical workflows.

At Brown University Health, chief digital information officer Adam Landman shared how the organization takes a platform-first approach when evaluating new technologies.

Instead of immediately reaching for standalone tools, the system first looks at capabilities embedded within the platforms clinicians already use, such as EHR systems or enterprise productivity tools.

This approach helps reduce workflow fragmentation and improves the likelihood that clinicians will actually use the technology.

Integration challenges remain one of the biggest barriers to AI adoption across healthcare, with many providers struggling to connect new tools to existing systems and workflows.

Measure success through clinician adoption

Finally, leaders said hospitals must rethink how they measure AI success, shifting away from deployment volume and moving towards clinician adoption as the key metric.

Landman said adoption data can reveal whether a tool is truly improving workflows. For example, providers can measure how often ambient documentation tools are used during patient encounters or how frequently clinicians rely on AI-assisted workflows.

Executives cautioned that organizations often get only one chance to introduce a new technology to clinicians. If the tool is inaccurate or slows down workflows, physicians may abandon it quickly and become reluctant to try future innovations.

Due to that dynamic, the most successful implementations center on usability and workflow integration as much as the technology itself.

Looking ahead

Projecting forward, leaders expect AI systems to take on increasingly complex tasks. Several predicted the next phase of adoption will involve “agentic” AI, with systems capable of performing multi-step tasks autonomously, like coordinating workflows or routing information across systems.

For now, these systems will likely focus on lower-risk tasks such as administrative coordination or workflow automation before expanding into more complex clinical applications.

At the same time, executives noted that some of the most effective uses of AI may remain largely invisible. In areas like diagnostics, AI is already embedded into workflows and delivering value without requiring direct interaction from clinicians, representative of a model some believe will expand in the coming years.