From a Clinician Perspective, Ambient AI Is Returning the ‘Joy Back to Medicine’

Penn Medicine Is an Early Adopter of Ambient AI

An early adopter of ambient AI through its Penn Sparks innovation program, Philadelphia-based Penn Medicine has developed ambient documentation for approximately 1,300 ambulatory providers and over 600 nurses and is expanding these solutions across physician and nursing teams in ambulatory, inpatient, emergency and home care settings, with full-scale deployment targeted for fall 2026, says Vice President and Chief Digital Applications Officer Anna Schoenbaum.

Future rollouts will include an AI platform for specialty clinicians, and the organization is testing an AI charting solution for Penn Medicine At Home.

Schoenbaum says the data shows a 20% reduction in documentation time, and decreased after-hours documentation for 79% of users, but the qualitative feedback has been equally striking. “We’ve had testimonies from spouses who said, ‘Thanks for giving my wife back, because she’s not going to the computer to work in the evening,’” she adds.

Metrics have also surfaced new questions, including a corresponding increase in orders placed.  

“Maybe that means we’re having more thorough conversations with our patients,” Schoenbaum says. “We’re trying to understand what’s driving that.”

READ MORE: Clinical workflow optimization creates better patient outcomes.

Penn Medicine recently rolled out ambient AI to approximately 550 inpatient nurses and 80 nurses in ambulatory. Early results show reductions in documentation latency and time spent in the electronic health record (EHR) system. In ambulatory settings, Penn Medicine is working with a vendor partner to integrate ambient AI into nurse triage workflows. When triage nurses encountered friction with a new nurse triage module, Schoenbaum introduced ambient AI as an alternative approach, despite the initial concern that the platform was not designed for that use case.

“We tried it, and it is clunky, but the nurses love it. So, we’re fine-tuning that, and hopefully, we’ll have that in production soon,” she says.

The work reflects a broader shift toward rethinking how technology enables care delivery, focused on practical innovation that reduces friction and supports clinicians in reimagining workflows across settings, says Schoenbaum.

Penn Medicine continues to study patients’ experience with AI scribes and plans to keep scaling into other disciplines.

“This is just the beginning, but it’s exciting that we finally have a technology that can bring a spark back into people’s workflow,” Schoenbaum says. “It’s about collaboration, as the implementation and support require a team of clinicians and experts in operations, data science, ethics, privacy, security, infrastructure and informatics, and we need everybody to make sure we’re doing it right.”

Early results of Penn Medicine’s ambient AI tool show reductions in documentation latency and time spent in the electronic health record system.

Advocate Health Tailors Ambient AI for Nurses

Charlotte, N.C.-based Advocate Health, which serves patients in six states, has deployed Microsoft Dragon Copilot to physicians and is now rolling it out to nurses. Recognizing that physicians and nurses deliver and document care differently has been crucial, says Tracy Breece, vice president of nursing innovation, AI and emerging technologies.

“Nurses are in motion, and we have quite a few distractions — telephone calls, messaging, interruptions — and this tool must work while we’re moving, while care is happening,” says Breece, who is also a registered nurse.

Moreover, an AI solution needs to file information the way a nurse would — into discrete sections of the EHR — and accommodate Advocate Health’s model of having nurses and nurse techs work together. The benefit, Breece says, is that “the documentation is happening in real time, so that alleviates a cognitive load and reduces latency.”

In 2024, Advocate Health collaborated with Microsoft on new, nursing-specific capabilities for Dragon Copilot, with nurses working alongside engineers from the tech partner and health system.

“Nurses know their work the best, so being at the table to have that synergy and insight, and to share, ‘This is how I work,’ really allowed for fast iterative cycles,” Breece says. 

The process also reinforced that change management is essential. “We learned a lot about understanding applied AI in a day in the life of Advocate nurses, and we were able to develop training that met us where we were at,” she adds.

In 2025, Microsoft released nursing features in Dragon Copilot, and Advocate Health rolled out the tool to approximately 130 nurses and nurse techs at two facilities. By summer 2026, an additional 450 nurses will have ambient AI scribes.

For Breece, the goal is less about efficiency and more about transforming bedside care, uplifting the art and science of nursing, and promoting nurse well-being. She knows it’s working when nurses report the qualitative shifts: “What I hear when I’m rounding is, ‘I get to clock out on time. I’m spending more time with my patients. My patients are having good conversations with me.’”

Click the banner below to sign up for HealthTech’s weekly newsletter.

Temple Health Scales Ambient AI Across Specialties

Philadelphia-based Temple Health started piloting ambient documentation when Slovis joined as chief medical information officer in 2024. He has overseen additional pilots and a Dragon Copilot rollout to 400 end users. While metrics are promising — less “pajama time,” faster chart closure — Slovis says the most important measure is whether physicians feel more efficient and less burned out.

“We want our providers to go home at the end of the day feeling that they’ve accomplished something, that they were able to focus on their patients and bring a little bit of joy back to medicine,” he says.

One scaling challenge has been developing comparable key performance indicators across specialties. “It’s easy to compare a gastroenterologist with a gastroenterologist. It gets harder when you start to look at the metrics of comparing a gastroenterologist to a pulmonologist or an ER doctor,” Slovis says.

WATCH: Tampa General Hospital rolls out ambient clinical documentation to nurses.

Temple Health has also used metrics to identify physicians who could benefit from ambient documentation tools and gathered feedback on clinical use — for instance, specialists wanted better documentation of assessment-and-plan notes. 

“Workflow integration became really interesting when we started to get into the weeds,” Slovis says. “These tools are great at documenting the plan for today, but unless you speak out loud that full summary statement, they don’t do that, so that was a big complaint from the end users.”

He’s also seen these AI tools change the patient experience in his own practice.

“Historically, I would auscultate a patient’s heart and lungs and not say anything, but now, I’m saying my observations out loud, and it changes the dynamic. Patients enjoy hearing what I’m experiencing during a physical exam. It takes a little bit of the mystery out of what I’m doing,” Slovis says.