Microsoft Dragon Copilot vs Nabla
This is the only pair in the category where both vendors were arms in the same randomized controlled trial, and the result favours the challenger. In Lukac et al, published in NEJM AI in November 2025, 238 UCLA Health physicians across 14 specialties were randomized to Dragon Copilot, Nabla or usual care over two months. Nabla reduced time in note by 9.5 percent and was the only arm to meet the primary endpoint; the Dragon arm reached 1.7 percent and was not statistically significant. Both improved burnout measures by roughly 7 percent. That does not settle the purchase on its own, because Dragon Copilot covers roles Nabla does not, carries deeper Epic embedding and publishes more about its own limitations. But when a buyer asks which of these two demonstrably saved documentation time in a trial neither of them ran, there is now an answer, and it is Nabla.
- The published documentation set is the most complete in this category and it names failure modes rather than gesturing at them: background noise and overlapping speech, omitted findings and hallucinated details in summarization, and underperformance on accents and dialects not well represented in training data.
- Roles beyond the physician. Distinct editions for nurses and radiologists, including bedside conversation converted into structured flowsheet entries, against a competitor built around the clinician note.
- Regulatory posture is stated in the markets that ask for it. Registered with the United Kingdom regulator as a Class 1 medical device and available across ten countries, which is a materially different position from a product that is unregulated everywhere it ships.
- It is the only ambient scribe in this index with a positive result in a randomized controlled trial it did not run, reducing time in note by 9.5 percent against control across roughly 72,000 encounters, and the only arm in that trial to meet its primary endpoint.
- The widest EHR footprint in the category, more than 15 platforms named including Epic, Oracle Health, athenahealth, NextGen, Greenway and Altera. For a system running mixed record systems across hospital and clinic, that is one scribe contract instead of several.
- The data lifecycle is published end to end, which almost nothing else here does. Audio is not stored, models are not trained on user data, clinical data is retained for a configurable 14 days sized around clinician review, and backups expire seven days after that so the deletion claim reaches the backup layer.
Side by Side
| Axis | M Microsoft Dragon Copilot |
N Nabla |
|---|---|---|
| AI Centrality | ||
| Autonomy and Oversight Model | ||
| Model and Technology Transparency | ||
| Clinical and Operational Evidence | ||
| AI Safety and PHI Stewardship | ||
| HIPAA and BAA Posture | ||
| Security Certifications and Trust Center | ||
| FDA and Regulatory Status | ||
| AI Governance and Bias Disclosure | ||
| EHR and Interoperability Depth | ||
| Deployment Model and Data Residency | ||
| Commercial Transparency | ||
| Setting and Specialty Coverage |
Related comparisons
Other published head to head assessments involving these vendors or their closest peers. The full set for this category is on the Ambient Scribes page.
The trial cited is Lukac et al, NEJM AI, 26 November 2025, registered as NCT06792890. It measured time in note and burnout over two months in an outpatient population at a single academic health system, so it does not establish note quality, coding accuracy or performance in inpatient and emergency settings, and a two month window may not capture proficiency gains.
Nabla moved off published per month pricing during 2026 toward a demo and contract motion, and its free tier is reported not to include a business associate agreement, so evaluate on the tier you would actually contract. Microsoft publishes no rate card.