A medical scribe documents the clinical visit so the clinician doesn’t have to. The scribe captures the history, exam, assessment, and plan in real time while the doctor sees the patient, then the doctor reviews and signs. A scribe can be a person in the room, a person off-site on a call, or software that listens and drafts the note.
That’s the role in one sentence. The interesting part is that “medical scribe” now covers three very different things, and they don’t cost the same, don’t carry the same privacy footprint, and don’t scale the same way. This page sorts them out.
Key takeaways
- A medical scribe handles the documentation so the clinician can stay with the patient. The clinician still reviews and signs every note.
- There are three kinds today: an in-person human scribe, a remote (virtual) human scribe, and an ambient AI scribe.
- The two human options put a salary on your books and a second person in the visit. Most are pre-med or pre-PA students passing through.
- US physicians log a median of 36.2 minutes of EHR time per primary-care visit (JAMA Network Open, 2023). That’s the burden every scribe type exists to cut.
In-person human, remote human, and ambient AI scribe, all doing the same documentation job differently
Median EHR time logged per primary-care visit (JAMA Network Open, 2023)
To review an AI-drafted note after the visit ends
What does a medical scribe actually do?
A scribe writes down the visit. That’s the whole job, and it’s a bigger job than it sounds.
The American Association of Medical Colleges puts it plainly: a scribe “helps physicians provide more effective and efficient care by saving them time by documenting findings from patient exams and assessments.” In practice that means the chief complaint, the history, what the exam showed, the results you talked through, your assessment, and the plan, all captured while the visit happens. One scribe profiled by the AAMC “transcribed hundreds of physician-patient encounters, medical histories, and diagnostic test results” in an ENT clinic.
The point of all that is simple. When someone else is typing, you can look at the patient. You can think instead of chart. The note gets written at the speed of the conversation rather than waiting for you at 9pm.
One thing a scribe is not: the author of the record. The scribe drafts. You read it, fix it, and sign it. Whether the scribe is a person or a piece of software, the documentation responsibility stays with the clinician who signs. That doesn’t move.
Scribe versus transcriptionist: what’s the difference?
People mix these up, and they’re not the same job.
A transcriptionist works from your dictation. You finish the visit, you narrate what happened, and they type that narration into text. The visit gets documented twice: once when you live it, again when you describe it. A scribe skips the second pass by capturing the visit itself, structuring the note as the encounter unfolds.
So the test is timing. A transcriptionist turns your words about the visit into text afterward. A scribe turns the visit into a note while it’s happening. An ambient AI scribe does the live, structuring job too, just without a person. If you want the difference between live capture and after-the-fact dictation laid out in full, our ambient AI versus dictation breakdown covers it.
What are the three kinds of medical scribe?
Here’s where the word splits. The same job, three delivery models.
The in-person human scribe. A trained person, usually hired through a scribe staffing company, stands in the exam room with you and types as you work. The AAMC’s profile describes a scribe working “closely alongside” the physician, “typically work[ing] with a single physician on a given day,” learning each provider’s charting habits. Most of these scribes are pre-med, pre-PA, or pre-NP students earning clinical hours before professional school. Pay runs roughly $12 to $20 an hour. It’s the oldest model and the most flexible, because a person can do more than document.
The remote (virtual) human scribe. Same idea, but the person is off-site. They join the visit over a secure audio or video link and document in real time. ScribeAmerica, one of the larger vendors, connects its virtual scribes “via a tablet on a stand, mobile cart, or a phone,” and the scribe “documents [the] encounter directly into [the] EHR using their own account in real-time.” You get a live human without a third body in the room. The trade-off: a stranger is listening to the whole visit, which makes the Business Associate Agreement and the encryption their own thing to vet.
The ambient AI scribe. Software, not a person. It listens to the visit, transcribes it, and drafts the structured note itself, then hands it back for you to review and sign. Nobody else hears the conversation. We won’t re-explain the step-by-step here, because we already wrote that page: how an AI medical scribe works walks through capture, transcription, structuring, and review in detail.
The three scribe types, side by side
This is the table worth keeping. Same documentation job, very different shape.
| In-person human scribe | Remote (virtual) human scribe | Ambient AI scribe | |
|---|---|---|---|
| Who does the work | A person in the exam room | A person off-site, on audio or video | Software, during the visit |
| Cost shape | A salary, scaling per clinician (a US scribe averages near $41k/yr) | A salary or per-hour contract, usually below an in-room hire | A per-clinician subscription, no hiring |
| Patient privacy footprint | A third person physically in the room | A third person listening remotely, under a BAA | No extra person; audio handling depends on the vendor |
| Scales to a group practice | Linearly and painfully; one hire per clinician | Linearly, but easier to staff than in-room | Per seat, flat, no recruiting |
| Main trade-off | Most expensive; high turnover; a stranger in the room | Privacy of a remote listener; depends on the vendor’s security | Drafts only; you must read every note before signing |
A note on the cost cells: we’ve kept them to the shape, not the full math, on purpose. If you want the staffing economics in detail, salary, benefits, turnover, the per-note breakdown, that’s its own page: AI scribe versus human scribe cost.
Why do human scribes turn over so often?
Because for most of the people doing it, scribing isn’t the career. It’s the on-ramp.
The workforce is largely pre-med and pre-PA students collecting clinical hours before they apply to school. The AAMC features exactly that path. So a practice doesn’t hire a scribe once and keep them for a decade. You hire, you train for a few weeks while they shadow, you get a productive stretch, and then they leave for medical school. ProScribe, a national scribe company, says the average length of employment with them is 16 to 18 months. That’s one vendor’s number, not a law of nature, but the pattern holds across the field: the staff is transient by design.
The scale of that churn is easy to underestimate. ScribeAmerica, the largest US scribe company, says it currently employs more than 15,000 scribes across 3,500-plus healthcare facilities in all 50 states, and that it has hired and trained over 100,000 scribes across its history. Read those two numbers together and the model comes into focus: a standing workforce in the low tens of thousands, and a lifetime headcount six times larger, because the people cycle through. Every one of those seats is a hire, a training block, and eventually a goodbye. A subscription is none of those things.
That churn is the quiet tax on the human model. The salary line is visible. The recruiting and re-training cycle that comes around every year or two is the part nobody puts on the invoice. It’s also the part software doesn’t have: a subscription doesn’t graduate and move to Ohio.
The documentation burden a scribe is fighting
It’s big enough that the whole category exists to fix it.
In a 2023 JAMA Network Open study, primary-care physicians logged a median of 36.2 minutes of EHR time for every visit, with a quarter of doctors above 45 minutes. The study also measured “pajama time,” the EHR work that spills into evenings and weekends, at 6.2 minutes per visit. Stack that across a full clinic day and you get the after-hours charting that scribes, in every form, are meant to claw back.
- EHR time during the day83%
- After-hours pajama time17%
Whichever scribe you choose, the goal is the same: get that time back so the note stops outliving the appointment that created it. Our reduce charting time guide gets into the specifics for clinicians weighing the options.
Where AI Medical Scribe by Patient Square fits
It’s the AI option in that table, built so the only person in the visit is you.
The AI Medical Scribe is one module inside Practice Copilot, Patient Square’s AI platform for the whole practice. 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. No second person in the room, no off-site listener on a call. The prescription is a draft built from the plan you stated; you review the drugs, doses, and instructions and sign. It doesn’t run an automated interaction check, and it never goes to a pharmacy on its own. The safety check is you, the same as when you write the script yourself.
On privacy, the footprint is the smallest of the three: visit audio is processed in memory and discarded the moment the note is drafted, so there’s no recording and no human transcriber who heard the conversation. Notes are encrypted in transit and at rest, they belong to your practice, and you can export or delete any visit anytime. For US practices the floor is a signed BAA, which we offer every customer, with safeguards mapped to the HIPAA Security Rule and a SOC 2 Type II audit underway. The receipts live on our security page.
On cost, it’s a per-clinician subscription, not a salary: the Assist plan starts from $79 per clinician per month on annual billing in the US, with a 7-day free trial, and moving up to Copilot bundles an AI EHR and messaging. The full ladder, with no asterisks, is on the pricing page, and we break down what drives the number in AI medical scribe pricing.
When is a human scribe still the better fit?
Honest answer: more often than a software pitch will admit. Here’s the verdict, no hedging.
Keep a human scribe, in the room or remote, when you need a person who does more than write the note. A scribe who can take a verbal order to staff, chase a pending result, queue the next patient, or handle the unscripted parts of a visit is doing a job no AI touches. If “scribe” in your clinic really means “flexible clinical assistant,” software replaces one slice of that role, not the whole thing. And a teaching practice that wants to give pre-med students real clinical hours has a second, legitimate reason to keep humans on: the turnover you’d otherwise grumble about is the entire point of the arrangement.
Go with the AI scribe when the job is the note. If your scribe’s day is mostly typing the visit into the chart, you’re paying salary-and-turnover for documentation that software now drafts in about two minutes, with no second person hearing the visit. That describes a lot of practices, honestly. Many of them hired scribes for one reason only: to stop charting at midnight.
So it comes down to scope. A human can do more than document. An AI does the documentation far cheaper and never leaves for medical school. Want to see the AI version against your own visit type? Book a short demo and run the 7-day trial on a real clinic week before you decide.