Inside Fairview: A CMIO’s Perspective on Driving AI Innovation in Healthcare

On July 29, Nabla hosted a new session in its CompassionIT series, a CHIME-powered initiative bringing together healthcare leaders and technology experts to explore a key question: How can we ensure technology alleviates, rather than adds to, the burdens of care delivery?

One CMIO and her team have figured out how to harness the power of AI to drive real impact. That CMIO is Dr. Rebecca Markowitz, Chief Medical Informatics Officer at M Health Fairview and Associate Professor of Hospital Medicine at the University of Minnesota.

In this fireside chat with Nabla Co-founder and CTO Martin Raison, Dr. Markowitz shared a grounded and practical look at what it takes to lead AI adoption across an enterprise. From pilot projects to system-wide rollouts, from inpatient to outpatient settings, the conversation explored how M Health Fairview is applying AI across the organization, highlighting what’s working, what’s unique about their approach, and how the integration of academic and community care can serve as a powerful foundation for meaningful, scalable innovation.

Beyond the Pilot Phase: Turning AI Projects into Systemwide Success

One of the most pressing challenges facing CMIOs today is scaling innovation. At M Health Fairview, several AI initiatives like the deterioration index for hospitalized patients and diagnostic tools for dermatology have already moved beyond pilot into clinical practice. But as Dr. Markowitz emphasized, that success didn’t happen by chance.

“We had to learn the hard way that just launching a tool isn't enough. You need workflows, education, and buy-in,” she said.

When the team first introduced the deterioration index in their EHR, they assumed it would be self-explanatory. It wasn’t. Clinicians didn’t know how to use the tool or what to do with the information. Now, adoption strategies center on embedding AI tools into clinical workflows and clearly defining next steps. Dedicated monitoring teams and change management support are now built into every rollout.

Innovation with Intention: The Role of a CMIO

As CMIO, Dr. Markowitz sees herself as part translator, part educator, and part realist. “My role is to vet, educate, and push for solutions that align with our clinical reality,” she said. That means advocating for tools that solve real problems, not just checking the box for innovation.

Take ambient documentation.

“We approached ambient AI from a burden reduction perspective, not as a way to increase RVUs or throughput,” said Dr. Markowitz. “It was about bringing joy back into clinical practice. That made people curious. That made them want to try it.”

M Health Fairview piloted several solutions before selecting Nabla as its enterprise partner. The deciding factor? Usability and configurability.

“We landed on Nabla because it allowed for tailored prompts, which was key for specialists,” she said.

The result?

“Clinic is fun again. I rarely leave notes open at the end of the day.” She added with a big smile.

A System That Learns: Bridging Academic and Community Care

M Health Fairview spans both academic and community care settings, each with its own culture, challenges, and priorities. To bridge that complexity, the organization relies on its Learning Health System, a University of Minnesota–based framework that allows clinicians across the system to submit, test, and scale improvement ideas.

“Many of our AI projects started as LHS pilots,” said Dr. Markowitz. “Because we were able to show value at one site, we had data to help spread it elsewhere.”

She highlighted a retinal screening initiative that originated in community clinics where patients often faced barriers to specialty access. By enabling point-of-care screenings for diabetic eye exams, the tool brought care closer to patients. At the same time, it freed up appointments for more complex cases in academic centers.

“It’s a win-win,” she said, “just in different ways.”

Still, she was clear that community and academic settings aren’t interchangeable.

“They have different priorities, and that’s okay. What matters is finding shared value.”

Governance That Goes Beyond a Checkbox

With AI tools proliferating rapidly, governance is no longer optional. M Health Fairview established an AI oversight committee that includes leaders from clinical quality, equity, IT, and both academic and community care. It serves a critical gatekeeping function.

“It functions almost like an Institutional Review Board for AI,” Dr. Markowitz explained, referencing the formal committees typically used to review human research ethics. “But it’s not just for pre-launch approval. It’s continuous.”

The committee conducts ongoing monitoring to ensure tools remain safe, effective, and equitable over time. Through the Entrust AI initiative, funded by the State of Minnesota in collaboration with Mayo Clinic and the University of Minnesota, the team is evaluating AI model performance across specific subpopulations.

“Most tools are tested at the population level,” she said. “But patients are individuals. We need to know how models perform for different combinations of demographics and comorbidities.”

Reworking Documentation: A Shift Toward Joy in Clinical Care

Dr. Markowitz has long been at the forefront of improving clinical documentation and coding. One of her most impactful projects involved using discrete data to surface comorbidities that had been missed in clinician notes, ultimately helping the system lower its observed-to-expected mortality ratio from above 1.0 to around 0.6.

“It worked,” she said. “But it took a huge amount of effort. And those rules still have to be maintained.”

Ambient AI offers a smarter path forward.

“With tools like Nabla, we can present clinical context to providers in real time, not days later in a coding query,” she explained. “It’s a chance to reduce burnout and eliminate some of the worst parts of the job.”

She also shared how ambient has transformed her own clinical practice.

“I used to chart after hours every day. Even though I was good at the EHR, it took a toll. With Nabla, I start with a draft and just edit. It removes the cognitive burden.”

Patients notice the difference, too.

“When I tell them it helps me get home sooner, they’re genuinely happy for me. And the summaries are more complete than anything I ever typed while maintaining eye contact.”

Teaching the Next Generation: AI as Curriculum

As a leader in an academic system, Dr. Markowitz is shaping how the next generation of clinicians learns to work with AI. Her team piloted ambient documentation with family medicine and internal medicine residents, and the feedback was overwhelmingly positive.

“We’re now planning broader access, with programs setting their own parameters,” she said.

Still, some raised a familiar concern: shouldn’t residents learn how to write notes first?

“Absolutely,” said Markowitz. “Documentation is a form of critical thinking. But we’d be doing our trainees a disservice by not including AI literacy in their education. These tools are going to be standard, and we need to teach them how to use them responsibly.”

The AI-Clinician Partnership: Built on Listening

When asked what tech companies often get wrong, Dr. Markowitz didn’t hesitate:

“Vendors who don’t take the time to understand our workflows are the biggest challenge.”

One vendor, she noted, criticized her team for lack of onsite engagement, without understanding that their strategy relied on robust virtual support.

“We know our system,” she said. “What we need are partners who listen, adapt, and align with our goals.”

By contrast, she praised Nabla’s willingness to collaborate, particularly around rethinking dictation workflows to support ambient adoption.

“It’s been a true partnership,” she said.

Advice for Fellow CMIOs: Purpose, Oversight, and Pragmatism

As the conversation wrapped, Dr. Markowitz offered guidance for other CMIOs navigating the early stages of AI adoption:

• Define value upfront. Whether it’s time savings, improved documentation, or reduced burnout, know how your organization defines success, and make sure the tools support that.
• Build governance early. A formal review process is critical for consistency and long-term safety.
• Support the humans behind the change. Education, workflow design, and change management aren’t optional. They’re foundational.

“You’re changing how people work,” she said. “That takes support. And patience.”

A Final Word: Real Change, Made Practical

“AI isn’t about magic,” said Dr. Markowitz. “It’s about making things a little easier, a little better, every day. That’s how you create real change.”

Curious to learn more? Watch the full recording of Dr. Markowitz’s conversation with Martin here.