chartpath
Surveys and studies

Why clinicians look at AI scribes

Charting does not end when the patient leaves. Some physicians already use software that drafts the note. In one national survey, a larger share say they are interested. The figures below are surveys and studies. They are not a promise for your clinic.

Canada · one survey

Who uses an AI scribe, and who wants to

28%28% of physicians in this Canadian survey were already using at least one AI scribe.

Canadian Medical Association press release, January 26, 2026, on a survey with the Canadian Federation of Independent Business. Not a United States rate. Source & date

42%Another 42% said they were interested in an AI scribe.

Same Canadian press release. Interest is not the same as a plan to buy, and it is not a United States rate. Source & date

Why the note is the complaint

The same Canadian survey asked about paperwork, not about a brand of software.

Documentation and burnout

90% agree or strongly agree that clinical documentation contributes to burnout.

Canadian physicians and residents in the July 31–August 21, 2025 survey. Agreement is not a measure of hours. Source & date

Hours on paperwork

The same survey says physicians spend nine hours a week on administrative tasks. Family physicians report 9.9 hours, and other specialists 8.6.

Canada, July 31–August 21, 2025. 1,924 practising physicians and residents, plus 55 medical students with separate questions. Administrative time is wider than the visit note. Source & date

What users report back

Physicians who already use an AI scribe report saving 64 minutes a day.

Self-reported, in the same Canadian survey. The page also says 45% rank AI tools as a solution. This is not a timed study of a product on this list. Source & date

United States: AI for the chart

The American Medical Association asked about health AI, including charts and visit notes. That is wider than an ambient scribe, so these percents are not the Canadian 28% and 42%.

Any health AI

66% of physicians in the AMA survey used some health AI in 2024, up from 38% in 2023.

AMA news report, February 26, 2025, of a survey of nearly 1,200 physicians. This is any health AI, not an AI scribe. Source & date

Charts and visit notes

21% used AI for billing codes, charts, or visit notes in 2024, up from 13% in 2023.

Same AMA survey report. The question is broader than ambient scribes. 57% called reducing administrative work the biggest opportunity for AI. Source & date

Private practice

22% of private-practice physicians used AI for charts or visit notes, and 21% of employed physicians did.

AMA news report, May 27, 2025. Private practice here means solo practice plus single-specialty and multispecialty groups. Overall health AI use was 64% in private practice and 72% among employed physicians. Still any documentation AI, not a named scribe. Source & date

What a published study measured

Both studies are from health systems. A result for one product does not transfer to the next, or to an independent clinic.

Burnout, one vendor

After 30 days with the Abridge scribe, burnout in this survey fell from 51.9% to 38.8%.

Six US health systems, February 1–October 31, 2024. 263 ambulatory clinicians who finished both surveys. No control group. Surveys were not anonymous. Participation was voluntary. Abridge’s chief clinical officer is an author. Abridge is not on this product list. Not a result for the products compared here. Source & date

The same clinicians reported 0.90 hours less after-hours documentation time.

That figure is the study’s mean difference on its after-hours question, not a clock measured inside the EHR. The paper says the findings were subjective reports. Source & date

Nabla, one trial

Nabla users spent 9.5% less time in the note than usual care. Dragon Copilot, in the same trial, did not show a significant change (−1.7%).

UCLA, 238 outpatient physicians, 14 specialties, November 4, 2024–January 3, 2025. Nabla was used in 29.5% of 23,653 assigned visits. Clinicians reported occasional clinically significant inaccuracies. Burnout scores still need a larger multi-center trial. Funding note names UCLA. Dragon Copilot is not on this list. Source & date

What separates the products

Each line is the vendor’s own description, except the Nabla trial above. A fit line is not a rank.

Freed

Freed says it is built for community clinics, not large health systems, and that ICD-10 and CPT suggestions sit in the scribe.

Freed’s homepage. A coding suggestion is not a reviewed coding result. Headcount and recall claims on that page are not used. Source & date

Heidi Health

Heidi says the scribe supports 110+ languages and can draft the note with or without an EHR connection.

Heidi’s scribe page. The pricing page still lists EHR integration as Custom on Teams. A note without an EHR link is not chart writeback. Source & date

DeepScribe

DeepScribe lists specialty models for oncology, cardiology, gastroenterology, urology, orthopedics, and neurology.

A name on the list is not a price, a contract, or a test of note quality for an independent practice. Source & date

For oncology, DeepScribe says the scribe sits inside Flatiron OncoEMR, including prep and the note, without copy and paste.

DeepScribe’s OncoEMR page. Hours-saved and approval-rate lines on that page are the vendor’s and are not used. Source & date

Nabla

In the UCLA trial, Nabla was turned on for 29.5% of the visits assigned to it, and note time fell 9.5% versus usual care.

One health system. Clinicians still reported occasional important errors in the notes. This does not set the result for another clinic or another EHR. Source & date

Commure Scribe

Commure lists four levels: self-serve scribing, specialist-edited drafts, live support for complex visits, and autonomous CPT and ICD-10 coding.

Commure’s scribe page. The passage does not give a price for each level. “Best for” lines are Commure’s labels. Source & date

Sunoh

Sunoh names athenahealth, Cerner, eClinicalWorks, Epic, ModMed, Practice Fusion, and Tebra. A browser extension for other web EHRs requires a Sunoh subscription.

Homepage read November 14, 2026. An earlier read of the pricing page did not name these EHRs. A named connection is not a field-by-field writeback test. Source & date

Glass Health

Glass says the visit recording can also draft a differential diagnosis, history questions, and the next documentation.

Glass’s own guide. This is not a test of diagnostic accuracy. A clinic list price is still not published on the reviewed pages. Source & date

Sully.ai

Sully lists a receptionist, a scribe, a medical coder, and a consultant on the same platform.

Homepage names. Dollar savings and the safety chart on that page are vendor presentations and are not listed as results. Source & date

CharmHealth

CharmHealth presents CharmCopilot as its own assistant for notes and questions, inside the CharmHealth EHR.

Homepage. The page does not publish a CharmCopilot price or a connection to athenahealth, Epic, or eClinicalWorks. Source & date

DeepCura

DeepCura says a standard note is one credit, in 35 language modes, and a person approves what reaches the chart.

The seat still includes 1,000 credits a month on the price already listed. One credit is DeepCura’s unit. The page also says Epic, athenaOne, and AdvancedMD now ship a scribe, and that the scribe is not the reason to talk to DeepCura. Source & date

S10.AI

S10’s Solo plan lists ICD-10 and CPT coding and 60+ languages. Practice adds an AI receptionist. Enterprise lists deep Epic, Cerner, and Athena integrations.

Pricing page. Solo is $199 per provider per month on that page. “Most clinicians save 2+ hours” is not used. “Works with any EHR” on Solo is not the same bullet as the Enterprise deep integrations. Source & date

Vendor pages often say a scribe saves one or two hours a day. Those sentences are not the studies above.

Compare the products