The UCLA randomized trial of ambient AI scribes found product-specific time savings and potential burnout improvements requiring confirmation in larger trials. These findings support careful evaluation of individual documentation tools before choosing one, with continued attention to accuracy.
Key Takeaways
- In the UCLA trial, Nabla reduced time-in-note by 9.5% versus control; DAX showed no significant change.
- The JAMIA crossover trial reported greater satisfaction improvements and documentation time reduction with Product B than Product A.
- UCLA reported occasional clinically significant inaccuracies and described burnout improvements as potential, requiring confirmation in larger trials.
What randomized trials of ambient scribes test
An ambient scribe turns encounter conversations into draft documentation. In the UCLA registry, both tools capture physician-patient conversations, generate transcripts, and integrate summaries into electronic health record notes. Neither tool provides clinical decision support.
That distinction sets the scope of the evidence. Read these findings as tests of documentation efficiency and clinician experience. Do not turn them into claims about diagnostic performance. For that separate discussion, see our explanation of what clinical decision support can and cannot do.
Keep the outcome attached to every claim you discuss internally. The UCLA primary outcome was the change in mean time spent in notes per note, comparing month 2 with baseline on a natural log scale. A finding on that measure is not a finding about every part of a clinician’s working day.
How UCLA compared scribes with usual care
The UCLA study was a pragmatic randomized trial involving 238 outpatient physicians across 14 specialties. Its registry lists 79 physicians assigned to Nabla, 79 to Vendor B, and 80 to the no-scribe control group. The published abstract identifies the second product as DAX.
The trial ran from November 4, 2024, to January 15, 2025. Its published report, Ambient AI Scribes in Clinical Practice: A Randomized Trial, appeared online in NEJM AI on November 26, 2025. The registry marks the study completed and names the University of California, Los Angeles as sponsor.
Nabla users experienced a 9.5% decrease in time-in-note versus the control group. DAX showed no significant change on that measure. Keep both results together when discussing the study: the favorable finding belongs to a particular product and outcome.
Both platforms showed improvements in burnout metrics and reduced task load. However, the conclusion describes potential burnout improvements requiring confirmation in larger trials. Preserve that qualification in any procurement summary; do not rewrite it as a settled promise of reduced burnout.
The registry also lists productivity and electronic health record activity measures among secondary outcomes. Treat the presence of an outcome in a study plan separately from a reported result. The time-in-note finding is the result to discuss here, without attaching a billing or productivity claim.
A different comparison in the JAMIA crossover trial
The 2026 JAMIA report, Comparing ambient scribes: a randomized crossover clinical trial addressing ambient scribe technologies’ impact on physician burnout, describes an open-label randomized crossover study. It enrolled 160 outpatient clinicians at an academic medical center, with 136 survey respondents analyzed.
Participants were assigned to groups with crossover periods. The study measured workflow satisfaction on a 1-7 scale, burnout using the Copenhagen Burnout Index, and efficiency. Its abstract refers to the tools as Product A and Product B; keep those labels when discussing the results.
Product B’s satisfaction improvement was 2.51 points, compared with 1.91 points for Product A. The abstract reports Product B’s reduction in average minutes-in-notes per day exceeded Product A’s by 3.19 minutes. Both tools reduced personal and work burnout scores, but differences between tools were not meaningful. No meaningful differences were observed in pajama time or patient-related burnout.
Your comparison should retain those distinctions. Overall burnout scores and patient-related burnout are separately described findings in this report. Similarly, the satisfaction result should remain a satisfaction result, rather than becoming a substitute for documentation accuracy.
Place the studies side by side before drawing a purchasing conclusion. Their designs, measures, and product labels belong in the same conversation as their favorable results.
| Study | Design and participants | Reported findings |
|---|---|---|
| UCLA, NEJM AI | Three-group pragmatic randomized trial; 238 outpatient physicians across 14 specialties. | Nabla: 9.5% less time-in-note versus control. DAX: no significant change. Potential burnout improvements need confirmation. |
| JAMIA, 2026 | Open-label randomized crossover trial; 160 outpatient clinicians; 136 survey respondents analyzed. | Satisfaction improved by 2.51 points with Product B and 1.91 with Product A. Both tools reduced burnout scores. |
Time savings and accuracy concerns
The UCLA abstract reports occasional clinically significant inaccuracies. Its conclusion pairs documentation benefits with a call for continued accuracy monitoring. That pairing deserves space in your assessment, alongside the headline time reduction.
Keep a practical question open during your review: what would your team need to observe before accepting a draft note? Use our discussion of AI scribe note cleanup time to frame questions about reviewing and correcting documentation. Do not assign a numerical cleanup burden to these trials beyond their reported measures.
A useful evidence summary separates the tool’s task, the measured outcome, and the reported result. For UCLA, the task was documentation, the primary measure concerned time in notes, and the time findings differed by product. Accuracy concerns remain part of that same evidence.
What the evidence supports
The UCLA findings and the JAMIA findings should remain separate entries in your evaluation. UCLA compared scribe applications against usual care; JAMIA compared anonymized products through a crossover design. Avoid treating their results as a shared ranking of named vendors.
For an internal review, label documentation time, workflow satisfaction, burnout, and accuracy separately. Record which study supports each statement. Keep UCLA’s request for larger confirmatory trials attached to its potential burnout benefits, even when summarizing the findings for a busy decision-maker.
Apply our evidence-first editorial filter to the purchasing discussion: preserve the comparator, outcome, and qualification. Take questions about your organization’s legal or regulatory circumstances to counsel or the relevant regulator. These studies describe documentation tools and measured outcomes, rather than resolving your organization’s individual obligations.
Frequently Asked Questions
Has a randomized trial compared ambient AI scribes with no scribe?
Yes. The UCLA trial randomized 238 physicians to Nabla, a second scribe identified as DAX in the published abstract, or a no-scribe control group.
What did the UCLA trial find about documentation time?
Nabla users experienced a 9.5% decrease in time-in-note versus control. DAX showed no significant change on that measure.
Has a randomized trial compared two ambient scribes directly?
Yes. The 2026 JAMIA crossover trial enrolled 160 outpatient clinicians. Product B showed greater satisfaction improvements and documentation time reduction than Product A.
Do these trials conclusively establish a burnout benefit?
The UCLA conclusion describes potential burnout improvements requiring confirmation in larger trials. The JAMIA abstract reports that both tools reduced personal and work burnout scores, but differences between tools were not meaningful; retain each study’s wording.
Sources
- UCLA ambient scribe trial registration, NCT06792890. ClinicalTrials.gov; study dates November 4, 2024, to January 15, 2025.
- Ambient AI Scribes in Clinical Practice: A Randomized Trial. NEJM AI, November 26, 2025; doi:10.1056/AIoa2501000.
- Comparing ambient scribes: a randomized crossover clinical trial addressing ambient scribe technologies’ impact on physician burnout. JAMIA, 2026; doi:10.1093/jamia/ocag018.
This article is for general information and is not legal, regulatory or medical advice.