Evidence record: ambient AI scribes in a randomized trial

On this page

By ChatGPT Health. AI assisted educational evidence summary. Source checked September 28, 2026. This is an abstract based record, not an independent clinical peer review.

Study and setting

A pragmatic randomized trial assigned 238 outpatient physicians across 14 specialties to DAX Copilot, Nabla or usual care. The trial ran from November 2024 to January 2025. Its primary endpoint was change in time spent writing notes.

Reported result

Nabla reduced time in note by 9.5% compared with control. DAX did not show a statistically significant change. Clinicians reported occasional clinically significant inaccuracies. These findings concern documentation and clinician experience, not demonstrated improvements in patient outcomes.

Source limits

The abstract identifies UCLA and other funding. Product version, detailed subgroup results and complete conflicts require the full report. Results apply to the tested products, period and setting.

How to use this record

For a local evaluation, separate the question of documentation time from the question of note quality. Define what counts as an omission or a consequential error before reviewing notes. Specify who resolves disagreements between reviewers and who is responsible for the signed record.

Document whether the software is optional, how often it is used, what consent process applies and what happens when a patient declines. A purchasing decision also needs local integration, retention, support and total cost information. This page does not rank vendors.

Primary source

Lukac and colleagues, Ambient AI Scribes in Clinical Practice: A Randomized Trial, NEJM AI, 2025. Read the full report and disclosures.