Summarize this blog post with:
Key takeaways
- AI scribes are priced per clinician. One January 2026 JAMA Health Forum viewpoint put ambient scribes at roughly $300 to $500 per clinician per month, which is $3,600 to $6,000 a year.
- A human scribe at the average $12 an hour reported in an AAFP pilot comes to about $24,960 a year for a full-time 2,080-hour schedule, before recruiting, training and turnover. Practices that hire through a scribe service should expect about double that rate.
- Published time savings are modest but real: about 1.5 minutes per appointment in a Hawaii Pacific Health pilot and about 1.6 minutes per note in a large emergency department study. Human scribes saved about 3.3 minutes per note in that same study.
- Neither type of scribe raised hourly clinical productivity in the emergency department study, so build your case on time and burnout, not on extra visits.
The short answer: an AI scribe usually costs a fraction of a full-time human scribe and scales with each clinician you add. Whether it pays back depends on one number you can measure yourself, the minutes it returns per note. The rest of this guide shows what is known, what is not, and the formula to use.
How much does an AI scribe cost?
There is no single list price. Vendors price per clinician per month, sometimes with setup fees or tiers by feature. The best public benchmark we found comes from Nong and Neprash in JAMA Health Forum (January 2026), who put ambient scribes at roughly $300 to $500 per clinician per month. Treat that as a reference range, then compare quotes on the same basis: price per clinician, what is included, contract length and exit terms.
| Cost line | AI scribe | Human scribe |
|---|---|---|
| Base cost | About $300 to $500 per clinician per month (JAMA Health Forum, 2026) | $9 to $17 an hour, average $12, about $24,960 a year full-time at $12 (AAFP pilot, 2016) |
| Through a service | Subscription | About double the hourly rate (AAFP) |
| Setup and training | Onboarding and workflow setup | Recruiting, training, supervision |
| Turnover risk | None per person | Scribes are often early-career staff who move on |
| Review time | You review every draft note | You review every note they write |
| Scales with | Number of clinicians | Number of scribes you can hire and manage |
The AAFP figures are from 2016, so use them as a floor and replace them with your local rate. The annual figure is our arithmetic: $12 x 2,080 hours.
How much time does an AI scribe save per note?
Two studies give the clearest numbers, and they agree on the order of magnitude.
- Hawaii Pacific Health (JMIR Medical Informatics, 2026): 79 providers, more than 25,600 notes, three months. Among high users, who used the scribe for at least 60 percent of notes, time in notes per appointment fell from 7.14 to 5.63 minutes, a 21 percent reduction. Time in notes per day fell from 66.24 to 52.63 minutes. The pilot had no control group.
- Mass General Brigham (Annals of Emergency Medicine, 2026): more than 198,000 emergency department encounters at four sites. Ambient AI scribes cut median documentation time by about 1.6 minutes per note and produced longer notes. Human scribes cut it by about 3.3 minutes. Neither raised hourly clinical productivity.
Both results put the saving per note at roughly 1.5 to 1.6 minutes for AI scribes. Time saved per note sounds small until you multiply it.
A worked ROI example you can adapt
The inputs below are example values for illustration. Replace them with your own.
| Step | Example | Result |
|---|---|---|
| Minutes saved per note (from the studies) | 1.5 | — |
| Notes per day | 20 | 30 minutes a day |
| Working days per year | 220 | 6,600 minutes, or 110 hours a year |
| AI scribe cost at $500 a month | $6,000 a year | About $54.55 per hour returned |
| AI scribe cost at $300 a month | $3,600 a year | About $32.73 per hour returned |
If an hour of your clinical time is worth more than the cost per hour returned, the tool pays back on time alone. If it is not, the case rests on after-hours work and burnout. The 110-hour figure assumes the study average holds at your practice and that you still review every note.
The after-hours and burnout case
In the Hawaii pilot, providers reporting eight or more hours a week of after-hours EMR work fell from 32 percent to 7.6 percent. Self-reported mental demand fell 46 percent and time demand 58 percent. Burnout fell from 59.5 percent to 50.6 percent, but that change was not statistically significant, and the pilot reported that self-reported time savings were larger than the roughly 13.6 minutes a day measured in the EHR. Count after-hours relief as a benefit, but do not promise a burnout number.
Can a scribe pay for itself with more visits?
The evidence is mixed, so do not rely on it. In a 2016 AAFP human scribe pilot, visits per month rose about 12 percent, from 152.8 to 171.4, and the authors said most physicians can cover scribe costs by seeing two to four additional patients a day. The 2026 emergency department study found that neither human nor AI scribes increased hourly clinical productivity. The JAMA viewpoint illustrated that a $500 monthly cost could be recouped by about four extra level-4 established-patient Medicare visits a month, while warning that using scribes for billing, not just documentation, could raise spending and recommending clinician-in-the-loop claims auditing.
AI scribe vs human scribe: which is better value?
Human scribes saved more time per note in the emergency department study, about 3.3 minutes against 1.6. They are also a person to hire, train and retain. AI scribes save less per note but work on every clinician's schedule and do not leave. No head-to-head accuracy trial between the two in primary care was found, so review every note either way.
| If this describes you | Lean toward |
|---|---|
| Complex visits and a scribe who has stayed for years | Human scribe |
| Several clinicians and uneven schedules | AI scribe |
| Coverage across sites and shifts | AI scribe |
| You want evidence behind suggested E/M levels and diagnoses | An AI scribe that shows it, such as SteerNotes |
What you get with SteerNotes
SteerNotes is Steer Health's Agentic Clinical Intelligence for clinical documentation, so it covers more of the visit than a transcription-only scribe. It prepares a concise summary of history, results and open items before the clinician walks in, listens during the visit and drafts the note in your templates, shows E/M level and diagnosis suggestions with the supporting evidence inline, queues labs, imaging and prescriptions mentioned aloud in the EHR for confirmation, and routes referrals and care gaps into scheduling. The clinician reviews and signs, and nothing fires without that signature. It comes with a BAA, encryption and audit logs. When you compare cost, compare what each tool does for the whole visit, not just the note.
How to run your own numbers in 30 days
- Record two weeks of baseline: minutes in notes per visit and after-hours EHR time, from EHR audit data if you can.
- Run the tool for two weeks with three to five clinicians.
- Calculate minutes saved per note, then multiply by notes per day and working days.
- Divide the annual cost by the hours returned.
- Audit 20 signed notes per clinician to confirm quality before you scale.
Want your own version of this table built from your visit types? Book a SteerNotes demo and bring your baseline.
Book a SteerNotes demo | Explore SteerNotes
Sources
- JAMA Health Forum viewpoint, Nong and Neprash, January 2026: https://jamanetwork.com/journals/jama-health-forum/fullarticle/2843723
- JMIR Medical Informatics, Hawaii Pacific Health ambient AI pilot: https://medinform.jmir.org/2026/1/e84104
- Annals of Emergency Medicine (2026), Mass General Brigham: https://mgriblog.org/project/ambient-ai-scribes-and-human-scribes-reduce-emergency-physician-documentation-burden-by-varying-degrees/
- AAFP Family Practice Management, scribe pilot: https://aafp.org/fpm/2016/0700/p23.html
- SteerNotes: https://steerhealth.io/steer-notes/




