The real ROI of an AI scribe is whatever recovered time is worth to you, minus a small monthly fee. The honest 2026 studies put the time saved at about 41 seconds per note in one trial and roughly 16 minutes per eight clinical hours in another. At 20 notes a day, that’s somewhere around 14 minutes back daily, against $79 a month. The arithmetic clears the cost long before you get to the burnout you didn’t have.
Most “ROI calculator” pages inflate the inputs and skip the citations. We’re going the other way. Real figures from the 2026 randomized trials, the math shown step by step, and an honest section on when it doesn’t pay off so you can check your own situation against it.
Key takeaways
- A 2025 UCLA trial measured about 41 fewer seconds per note for one ambient tool; a 2026 JAMA study found ~16 minutes saved per 8 hours of patient care. Modest, real, well-sourced.
- At 20 notes a day, ~41 sec each is roughly 14 minutes back daily. Over 20 working days that’s about 4.6 hours a month.
- At $79/month and 400 visits, the scribe costs about $0.20 a visit. Compare cost per visit, not the monthly sticker.
- The break-even is tiny: recover about one hour of your time a month and the subscription has paid for itself at almost any clinician hourly rate.
saved per note in one tool, UCLA randomized trial (NEJM AI, 2025)
implied time back at 20 notes/day from that per-note figure
cost of a $79/mo subscription (annual Assist) across 400 visits
What does an AI scribe actually save you, in minutes?
Start with the studies, not the sales deck, because the inputs decide the answer.
The most rigorous 2026 read is a UCLA randomized trial published in NEJM AI: across 72,000 encounters and 238 physicians, one ambient tool cut average note-writing time by about 41 seconds, from 4 minutes 30 seconds to 3 minutes 49 seconds. A second tool in the same trial showed a smaller, statistically insignificant change. Separately, a JAMA study across five health systems found about 16 minutes of documentation time and 13 minutes of EHR time saved per eight hours of patient care, and called the effect “modest.”
So we’ll build the model on conservative numbers. If your notes track that 41-second figure and you write 20 a day, that’s roughly 14 minutes back daily. Over 20 working days, about 4.6 hours a month. Some clinicians save more, the studies note primary care and high-note-volume practices benefit most, but 4.6 hours is a figure you can defend.
The per-note math, worked out
Here’s the whole calculation, no black box.
Say you see 20 patients a day, 20 days a month, 400 visits. At $79 a month on the annual Assist plan, the subscription cost per visit is:
| Input | Value |
|---|---|
| Monthly subscription (US, annual Assist) | $79 |
| Visits per month | 400 |
| Cost per visit | $0.20 |
| Minutes saved per note (conservative, per UCLA) | ~0.7 |
| Total minutes saved per month | ~280 (about 4.6 hours) |
| Break-even time saving | recover ~1 hour/month and it has paid for itself at most clinician rates |
The break-even is the headline. You do not need to believe the optimistic time numbers. Even if you save a quarter of what the studies measured, the US subscription costs less per month than one recovered hour is worth to almost any clinician. That’s why the ROI question is usually settled before you reach the patient-revenue argument. If you want the full price ladder rather than this single launch figure, it’s on the pricing page, with the US rate card explained in our AI medical scribe pricing breakdown.
There’s a revenue angle too, and we’ll state it carefully. The 2026 JAMA study found the time freed up allowed for about half an extra patient visit a week, which one estimate valued at roughly $167 a month per clinician. Treat that as a possible upside, not a promise, it depends entirely on whether you fill the recovered time with patients or with leaving on time.
When does an AI scribe not pay off?
We’ll say it plainly, because an honest ROI page has to.
Sometimes charting just isn’t your bottleneck. If you already have a human scribe you’re happy with, or your notes are short and fast, the time saved is small and the subscription is only a cost. Sometimes the problem is the opposite, that you won’t actually read the drafts. The savings come from editing a near-complete note, not from skipping review, and any vendor implying you can sign unread is selling past the evidence. A wrong note you didn’t catch costs far more than $79. And sometimes the note quality just fails on your visits. A scribe that drafts badly on your accents, your specialty, or your noisy room adds cleanup time instead of removing it. That’s the one input the calculator can’t fake, which is why you test before you trust. Grade it with our SOAP-note quality rubric, and read why we won’t quote you a clean accuracy number in how accurate are AI medical scribes.
The AI Medical Scribe is one module inside Practice Copilot: it listens during the visit and hands back a structured SOAP note, ICD-10 suggestions, and a prescription draft, ready to review and sign about two minutes after the visit. The draft is a starting point you check and own. That review step is exactly why the ROI depends on quality, not just price.
Run the calculator on your own numbers
The model above uses defensible figures, but yours are the ones that matter. Plug in your real note count, your real per-note time, and your own hourly value, then decide.
The fastest way to get a true per-note time figure is to measure it. Book a demo to see a finished note appear about two minutes after a sample visit, then run the 7-day free trial on a real clinic week and time both the saving and the cleanup. The number you get from your own visits is the only ROI number worth trusting. If you’re comparing tools first, how to evaluate an AI medical scribe and the documentation-burden pillar on cutting charting time are the next two reads, and the best AI medical scribes roundup lays out the field.