Ambient artificial intelligence is changing how clinicians document patient visits, reducing hours of after-work charting and helping restore satisfaction in medical practice.
Early adopters report time savings and changes in care delivery
Three health systems in Pennsylvania and North Carolina are expanding ambient AI tools beyond initial tests, tracking both measurable gains and shifts in provider-patient interactions.
Penn Medicine in Philadelphia has introduced ambient documentation for about 1,300 ambulatory providers and over 600 nurses. The organization plans to extend the technology across inpatient, emergency, and home-care settings by fall 2026. Anna Schoenbaum, vice president and chief digital applications officer, shared the timeline.
Results show a 20% decrease in documentation time. After-hours charting fell for 79% of users. The most meaningful feedback came from clinicians’ families, who noticed evenings were no longer dominated by laptop work.
Schoenbaum recalled comments from spouses who said they regained time with their partners.
Related: State Cuts Medicaid for Trans Youth Care
Some outcomes were less clear. The system observed a rise in orders placed after ambient AI was implemented. Schoenbaum suggested the increase might reflect more thorough patient conversations but noted the cause remains under review.
The technology has since been adopted by 550 inpatient nurses and 80 ambulatory nurses. Early findings indicate faster documentation and less time spent in electronic health records. In ambulatory settings, Penn Medicine is testing the tool for nurse triage, an unconventional application that initially faced skepticism.
Schoenbaum acknowledged the trial was awkward but said nurses appreciated the effort. Adjustments are underway for a smoother rollout.
Specialties require distinct metrics and conversations
Temple Health in Philadelphia began testing ambient documentation in 2024. Dr. Benjamin Slovis, chief medical information officer and an emergency medicine physician, led the rollout to 400 users. While faster chart completion and reduced after-hours work are positive signs, Slovis said the true test is whether physicians feel less exhausted.
He wants providers to end their day with a sense of accomplishment and renewed connection to their work.
Expanding across specialties presented challenges. Comparing performance indicators between a gastroenterologist and a pulmonologist proved difficult. Temple Health used data to identify physicians who might benefit most and gathered input on clinical use. Specialists, for instance, requested better documentation for assessment-and-plan notes.
Related: HHS finalizes rule on information blocking
Slovis noted that workflow integration became complex as teams examined details. The tools effectively recorded daily plans but required clinicians to verbalize full summaries, which users found frustrating.
The technology also altered how Slovis conducts exams. Instead of silently listening to a patient’s heart and lungs, he now speaks his observations aloud. Patients appreciate the transparency, he said, as it demystifies the process.
For all three systems, the effort continues. None view ambient AI as a complete solution, but the evidence suggests progress.
Slovis said the data points to the right direction, even if the tools aren’t perfect yet.
Patients have raised concerns about oversight for AI listening tools in clinical settings, highlighting the need for clear guidelines as adoption grows.