For Some Health Systems, AI Is Personal

Ambient AI may mean the difference between success and survival, says LCMC Health System's CMIO

For some health systems looking to use AI, cost plays second fiddle to survival.

At LCMC Health, a six-hospital non-profit operating in and around New Orleans, the shadow cast by Ochsner Health looms large. In order to stay competitive, the health system has to be on the cutting edge of AI integration.

Especially with ambient AI.

“We had to get this to our doctors,” says Damon Dietrich, MD, who’s been LCMC Health’s System Chief Medical Information Officer for roughly a decade. “It wasn’t about seeing more patients or making more money.”

“We are mission-critical about this,” he says. “This is about physician burnout, well-being, [and] the burden of documentation. … We have to do this to stay relevant and competitive in our market. We’re going to lose doctors to our competitor. They’re going to come in and hire them.”

The Value of 'Bake-Offs'

The health system jumped into the ambient AI sandbox back in 2024, launching three-month pilots with a handful of companies to test how the technology best fits into clinical workflows. By April of last year they’d settled on Nabla, which had the added advantage of synching with their Epic EHR platform, and by September they were going live.

It’s a strategy that many healthcare organizations – large and small – are embracing, especially as the AI market grows and the technology becomes more sophisticated. Health systems are turning to “bake-offs,” or small pilot projects that pit one product against another, or a few against each other.

Damon Dietrich, MD, System Chief Medical Informarion Officer, LCMC Health. Photo courtesy LCMC Health.

"That's going to be the challenge for our industry: Picking and choosing the right models for your organization," Chris Carmody, Chief Technology and SVP of Information technology at UPMC, told HealthLeaders in a November 2025 interview.

"We actually did a bake-off" between the two products, Carmody said of the program that split the health system's estimated 1,800 physicians involved in the project into two equal groups. "To be honest, I think it’s a great approach to always compare and contrast what works well and what doesn't work well."

Dietrich says LCMC Health focused on how clinicians would integrate ambient AI into their normal workflows, and which tool worked more smoothly for them. They compared the final results for ease of use and accuracy, as well as how often a clinician would have to go back into the record to add details or edit and entry. A tool in the Epic EHR allows them to also track efficiency, through time spent in the medical record, ordering and interpreting labs and tests, and checking messages.

Once they agreed to the partnership with Nabla, Dietrich said health system leadership organized AI adoption into stages. Employed doctors were given access to the tool in November in the first stage, while residents and attending clinicians at affiliated academic institutions (Tulane and LSU) were included in the second phase. The third wave will include any other clinicians, including those who may have been hesitant to join during the first two phases.

Dietrich says they’re targeting 50% adoption at first, and will start monitoring in February or March. If clinicians aren’t using the tool, he says, the health system will take the license and give it to someone else, while putting that person at the end of a waiting list. If adoption increases, the health system may seek more licenses.

Developing a Long-Term Strategy

Now, to say that cost doesn’t factor into the health system’s technology strategy doesn’t really tell the whole story. Dietrich says they’re very attuned to how much they’re spending on technology.

“Everyone wants to be fiscally responsible,” he notes. “I want to be fiscally responsible, but I also want to give the doctors what they need.”

Aside from stabilizing the workforce, Dietrich says the program has enabled the health system to end its dependence on scribes, both virtual and in-person.

That’s said, doctors need to use the tool consistently to prove its value to the health system. And while other health systems across the country are integrating coding and charge capture into their ambient AI tools, Dietrich says he wants their doctors to focus only on the clinical value to them.

“Doctors aren’t coders and coders aren’t doctors,” he points out.

Dietrich says it will take a while for ambient AI to show true value for the health system. That’s why it’s important to develop monitoring and governance strategy that keeps a close eye on the technology over the long term.

“What does year 2 look like and what does year 3 look like?” he asks. “We're not going to go with an AI tool because you think it's cool. We're going to … be deliberate, thoughtful, prescriptive and financially responsible in the decisions we make.”

Dietrich says the health system isn’t big enough to develop its own AI tools, so it relies on the vendor market. And that means asking some tough questions.

“How do you train your model?” he says. “How do you do quality assurance on your model to ensure that the accuracy is only improving? How do you generalize the model to make it fit for the common [patient] rather than the data set that you trained on in your training model?”

“You’ve got to be very thoughtful … because a lot of these companies won’t be here in three to five years,” he adds.