AI scribes cut the note.
The after-hours work stayed.
The largest controlled study to date found ambient AI scribes save documentation time but barely move the total day. Here is what the April 2026 JAMA data showed, why after-hours charting did not drop, and what athenahealth practices should do about it.
- 16
- fewer documentation minutes per 8 hours of care
- 0
- significant change in after-hours EHR time
- 8,581
- clinicians studied across five health systems
Real savings, but the total day held.
Published in JAMA on April 1, 2026, the study tracked 8,581 ambulatory clinicians across Mass General Brigham, Emory, UC San Francisco, Yale New Haven Health, and UC Davis. It is the largest controlled evidence to date on whether ambient AI scribes change clinician workload.
- 13 min
Fewer minutes of total EHR time per eight hours of scheduled care, a 3% relative reduction.
- 16 min
Fewer minutes of documentation time per eight hours of care, a 10% relative reduction.
- +$167
Added monthly billing per clinician, with about half an additional visit per week.
- No drop
After-hours EHR time, the pajama time clinicians do at night, did not change significantly.
The note was never the only hard part.
A scribe drafts the note. The study authors point to the obvious explanation for the flat total: time saved on documentation gets reallocated to the rest of the work a visit creates, not given back as free evenings. That work is specific and unavoidable.
- 01
Coding still has to happen
Turning the encounter into correct, defensible ICD-10 and CPT codes is its own task. A drafted note does not produce a claim.
- 02
Orders still have to be entered
Labs, medications, and referrals have to land in discrete fields, correctly. Most scribes suggest at best; they do not post.
- 03
The inbox still has to be cleared
Results, refills, and patient messages keep arriving. The scribe never touches them, so they wait until after hours.
What a scribe finishes, and what still lands after hours.
The note is the first 20% of the burden. The study suggests the other 80%, the part that actually fills the evening, is still waiting once the scribe is done.
| The work a visit creates | Ambient AI scribe | What it still takes |
|---|---|---|
| Draft the clinical note | Drafted from the conversation | Clinician review and edits |
| ICD-10 / CPT coding | Suggested at best, model-dependent | Coder time, or after-hours coding |
| Order and prescription entry | Drafted in newer tools, not posted | Manual entry into discrete fields |
| Inbox: results, refills, messages | Not addressed | Triage and action, often after hours |
| A closed, billable, coded visit | No | The clinician finishes the job |
Evaluate the day, not the draft.
athenahealth is rolling its own ambient scribe, athenaAmbient, into athenaOne at no extra cost, which makes documentation drafting a built-in baseline. It does not, by itself, close the after-hours gap the JAMA data exposes. Four ways to evaluate ambient AI on the whole day.
- 01
Measure the whole day, not just note time
Track after-hours EHR minutes and inbox volume, not only documentation time. The study shows documentation savings can hide a flat total.
- 02
Drive adoption past the 50% threshold
Only 32% of adopters used scribes in at least half of visits, the level tied to the largest gains. Training and workflow fit matter as much as the tool.
- 03
Look past the draft to the action
Evaluate whether a tool also codes, enters orders, and helps clear the inbox, or whether it stops at the note and hands the rest back to the clinician.
- 04
Demand EHR write-back, not suggestions
For coding and orders, a suggestion still leaves the work. Codes and orders written back into discrete fields are where minutes actually return.
Start where the scribe stops.
CarePilot reads the visit, codes it to 98% accuracy, and writes ICD-10 and CPT codes back into discrete fields inside athenaOne, then carries order entry and the inbox to done. It is built to return about 78 minutes per clinician per day by finishing the work a visit creates, not just drafting the note.
- 98%
- coding accuracy
- 78
- minutes back per day
- 1 to 2
- business days to go live
After-hours questions.
Do AI scribes reduce after-hours charting?
Not significantly on their own. The 2026 JAMA study of 8,581 clinicians found ambient scribes cut documentation time by about 16 minutes per eight hours of care, but after-hours EHR time did not drop significantly. The time saved shifts to coding, orders, and the inbox.
How much time do AI scribes actually save?
In the controlled JAMA study, about 13 fewer minutes of total EHR time and 16 fewer minutes of documentation time per eight hours of care. High-frequency users and primary care clinicians saw larger gains, up to roughly 25 minutes per day.
Why doesn't a scribe give clinicians their evenings back?
Because a drafted note is only part of the visit. Coding, order entry, and inbox work still have to be finished, and the minutes saved on documentation get reallocated to that work rather than returned as free time.
Is athenaAmbient enough on its own?
athenaAmbient drafts notes, and in newer testing diagnoses and orders, inside athenaOne at no extra cost. It sets a documentation baseline, but drafting is not the same as completing coding, orders, and the inbox, which is where the after-hours hours go.
What closes the after-hours gap?
Completing the downstream work, not just drafting the note. Tools that write codes and orders back into discrete fields and help clear the inbox address the part of the day a scribe leaves untouched.
Finish the visit, not just the note.
Coding, orders, and the inbox, completed inside athenaOne. See it on your encounters. Book a 30-minute demo.