AI scribes can boost RVUs and revenue – and paint a complete picture of care

As the debate continues about return on investment (ROI) for artificial intelligence (AI) scribes in healthcare, new research points to modest but scalable revenue boosts that stem from more accurately documenting what physicians do. 

Researchers from University of California San Francisco (UCSF) found AI scribe adoption was associated with increases in RVUs, or relative value units. These codes represent the resources necessary to perform medical procedures and provide clinical services, according to the American Medical Association.  

Three components go into calculating RVUs: The work a physician performs, the expenses a practice incurs, and the cost of managing liability for services provided (when necessary). The UCSF team found physicians using AI scribes integrated with their electronic health record (EHR) systems documented an increase of 1.81 RVUs per week, coupled with a 2.8% increase in encounters. That extrapolated to $3,044 in additional revenue per year per physician, based on the 2025 Medicare Physician Fee Schedule. 

Modest gains, and potential questions

Another recent study from UCSF (and Mass General Brigham) found AI scribe users in an ambulatory setting spend about 30 fewer minutes per day in the EHR and add approximately one visit every two weeks. Though this study didn’t explicitly explore whether AI scribe adoption impacts revenue, signs suggest gains remain modest.  

All told, with AI scribe subscription fees ranging from $200 to $600 per user per month, health systems appear to be breaking even at best. The research adds fuel to the fire favoring the argument that AI scribes generate minimal ROI because the time saved per encounter is so small – and, as the UCSF/MGB study found, there’s no “significant” change in clinicians’ after-hours EHR time. 

However, a JAMA Network Open commentary noted, large health systems can scale gains across thousands of clinicians (and even more encounters) to further recoup subscription costs. Plus, previous research noted AI scribes capture increased risk adjustment factors and Hierarchical Condition Category (HCC) scores. Both can increase revenue by documenting that physicians are treating more complex conditions that previously indicated. 

A more complete picture of care

Of course, more RVUs and higher HCC scores raise a valid question about upcoding, the JAMA commentary said. That’s especially true considering insurers’ moves to implement AI-driven code reviews and downcoding policies, which “may signal a coding arms race” that would offset providers’ revenue gains. 

California-based family physician Dr. Gigi Magan sees it another way. Writing on LinkedIn, she said the UCSF study shows “the documentation [is] catching up to what we always did” because AI scribes capture all the clinical threads of an appointment.  

For example, a visit may start with a complaint about pain in the toe but continue to cover diabetes and hypertension risk, stress at home, the toe’s impact on the patient’s ability to drive, and so on. Because manual documentation is “a time-limited act” – and one often done after hours – Magan said few of those topics typically made it into the clinical note. By recording the conversation, scribes change that. 

For Magan, the potential gains are greatest for primary care practices, safety-net providers, community-based organizations, and other entities where “visits routinely exceed their own complexity.” That’s because scribes ensure charts reflect the effort physicians put into meeting all the needs of their patients.  

Brian Eastwood is a Boston-based writer with more than 10 years of experience covering healthcare IT and healthcare delivery. He also writes about enterprise IT, consumer technology, and corporate leadership.