An ambient AI scribe on a telehealth visit captures the conversation, not the call. It listens to what you and the patient say over video, then drafts a structured note. It does not plug into your video platform, record the screen, or examine the patient. The words become a note; the objective exam stays yours.
That’s the honest version, and it’s the version most product pages skip. A video visit is missing the thing an in-person visit gives you for free: your hands, your stethoscope, the smell of the room. The scribe can’t recover any of that. What it can do is take the talking part of the visit off your keyboard so the thin objective section is the only thing you’re writing by hand.
Key takeaways
- The scribe listens to the spoken visit and drafts a note. It does not integrate with Zoom, Doxy, or any video platform, and it never records video.
- It captures the conversation: HPI, review of systems, your spoken assessment and plan. It cannot see, palpate, or auscultate, so the objective section is still yours to write.
- CMS names any visit where a physical exam would change your recommendation as less appropriate for telehealth. The scribe doesn’t change that math; it just handles the documentation.
- Primary-care physicians log a median of about 36 minutes of EHR time per 30-minute visit (JAMA Network Open). The talking part of a video visit is exactly the part a scribe removes from your evening.
of EHR time logged per 30-minute primary-care visit (JAMA Network Open, via AMA)
from visit-end to a drafted note you review and sign
Assist tier per clinician (US, annual); same scribe for in-person and virtual
If a video visit is where a lot of your week now happens, read on. What follows is exactly what the scribe hears, what it can’t, and where the line sits between the note it drafts and the record you sign.
How does an AI scribe work on a telehealth visit?
The same way it works in a room. You start a session on your phone or laptop before the visit, get the patient’s consent, and talk. The scribe captures the spoken conversation ambiently and drafts a note from it. There’s no screen recording, no video capture, no hook into your telehealth software.
That last point matters, so let’s be blunt about it. The scribe is not a telehealth feature. It doesn’t join your Zoom room or read your Doxy session. It listens to audio the way a person sitting quietly beside you would, and on a video visit that audio is just your side of the call plus whatever the patient’s mic sends back. If your platform pipes the patient’s voice through your device speakers or headset, the scribe hears both halves of the conversation. If it doesn’t, it hears you. Either way, it’s listening to speech, not tapping the call.
AI Medical Scribe is one module inside Practice Copilot, the platform Patient Square builds. 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. On a telehealth day that draft lands the same as it would after an in-person visit: subjective, objective, assessment, plan, with the objective section reflecting only what you said out loud.
Want the roll-out mechanics for a small practice going hybrid? Our 2-week implementation plan walks through it.
What can a scribe capture on a video visit, and what can’t it?
Here’s the split, because it’s the whole point of this page. A telehealth visit has a full conversation and a hollow exam. The scribe is excellent at the first and useless at the second, and pretending otherwise is how documentation goes wrong.
| The visit part | Who captures it on a video visit | In the note |
|---|---|---|
| Chief complaint, HPI | The scribe, from the conversation | Subjective, drafted |
| Review of systems (what the patient reports) | The scribe, from the patient’s answers | Subjective, drafted |
| What you observe on camera and say out loud | The scribe, from your spoken words | Objective, only if you narrate it |
| Palpation, auscultation, anything you touch | Nobody. You can’t do it over video | Objective — absent unless done in person |
| Vitals the patient reads from a home device | The scribe, if the patient or you say the numbers | Objective, drafted from speech |
| Your assessment and plan | The scribe, from your spoken reasoning | Assessment + Plan, drafted |
| ICD-10 suggestions, prescription draft | The scribe, from the above | Handed to you as drafts to confirm |
Read the fourth row twice. The objective section of a physical exam is the part telehealth already struggles with, and no scribe fixes that. A 2021 review in The Journal for Nurse Practitioners put it plainly: face-to-face encounters are necessary where auscultation or palpation is needed, and telehealth is best used to supplement in-person care, not replace it. An AHRQ patient-safety commentary the same year was blunter still, noting that the value of smell and touch in the exam is simply lost on a video visit, and that certain parts of the physical exam cannot be performed online at all.
So the scribe captures a rich subjective and a spoken assessment, and it leaves you a lean objective section to fill. If you did a “look at the camera and show me the rash” exam, narrate what you saw and the scribe writes it down. If you couldn’t examine at all, that gap is real, and it belongs in your clinical judgment, not the tool’s.
If you also see patients in the clinic, the same scribe covers both; here’s the 20-patient-day math for primary care when the visits are in person.
Why doesn’t the scribe just plug into my telehealth platform?
Because it’s built to listen, not to integrate, and that’s a deliberate line. Integrations into video platforms, EHRs, and scheduling systems are a different product surface with a different risk profile. What ships today is an ambient listener that turns spoken visits into drafted notes, in a room or over video, without touching your other software.
There’s an upside to that limit that’s easy to miss. Because the scribe isn’t wired into your telehealth stack, it doesn’t care which platform you use. Switch from one video vendor to another and nothing about your documentation changes; you’re still starting a session and talking. The scribe’s job starts and ends with the conversation.
And the audio doesn’t stick around. Visit audio is processed in memory and discarded the moment the note is drafted. There’s no audio archive, not of your video visits, not of anything, for us or for your practice. Notes are encrypted in transit (TLS 1.2+) and at rest (AES-256), they belong to your practice, and you can export or delete any visit whenever you want. The full posture, including what we sign and what we don’t, is on our security page.
Does telehealth documentation have rules a scribe should know about?
Yes, and the scribe drafts toward them without deciding them for you. CMS is clear that telehealth documentation follows the same discipline as any visit: complete the documentation at the time of service, record the time you spent, and code the place of service correctly. For Medicare real-time video, that includes the CPT modifier 95. Those are billing and compliance calls the clinician owns; the scribe just gives you a clean, structured draft to make them from.
CMS also draws a line the scribe can’t draw for you. Its provider toolkit lists, among the situations telehealth is less appropriate for, “any situation in which a physical exam would change your recommendation.” That’s a triage decision, not a documentation one. The scribe will faithfully write up a video visit you probably should have brought into the clinic; it won’t tell you that you should have. Reading the draft is where you catch it.
The safer way to think about it: the scribe lowers the cost of documenting a telehealth visit to almost nothing, which frees your attention for the judgment calls that actually carry risk. What to examine in person, when to escalate, what your thin objective section is missing. Those stay with you.
Where does an AI scribe fall short on telehealth visits?
Three places, and they’re worth naming before you trust the note.
The objective exam is the big one, and we’ve said it, but it bears repeating because it’s where harm hides. There’s a real AHRQ-reviewed case: a frail, non-ambulatory patient mentioned skin breakdown on her “backside” during a video visit, the clinician didn’t visualize the area, and a perirectal necrotizing infection went undetected until she returned in septic shock and died after surgery. The video visit wasn’t wrong to happen; the missed exam was the failure. A scribe would have captured every word of that conversation and still not seen the wound. The tool documents; it does not examine.
Audio quality is the second. Telehealth audio is messier than a quiet room. Dropped packets, a patient on speakerphone in a noisy kitchen, a relative talking over the line, a connection that cuts out mid-sentence. When the audio degrades, so does the transcript, and the scribe can only write what it heard cleanly. Read the draft closely on your worst-connection visits.
Third, drafts contain errors, on video as much as in person. Models mishear drug names, compress two complaints into one, occasionally write a plausible line that wasn’t said. That’s why the review-and-sign step is load-bearing. The prescription draft reflects what was said in the visit; it does not screen for interactions or dosing, so you read it against the chart, adjust the dose, and sign, exactly as you do today. A draft an LLM wrote is still a draft an LLM wrote. You’re the author of the record, and the safety check, in every setting.
When your connectivity is unreliable, capture also works offline with on-device AES-256-GCM encryption and syncs the note later, so a dropped internet connection doesn’t cost you the documentation.
What should a telehealth-first clinician ask before signing up?
Five questions, tuned for virtual practice.
- “Does this actually integrate with my video platform, or does it just listen?” The honest answer for us is: it listens. Be suspicious of any vendor selling “telehealth capture” as a deep integration, and ask exactly what it touches.
- “What happens to the visit audio, and when is it deleted?” One sentence, with a timeline. Ours: processed in memory, discarded at note draft, no archive.
- “How does the objective section get built on a video visit?” The right answer is: from what you narrate. Anyone implying the tool examines the patient is overselling.
- “Will you sign a BAA?” A vague answer is disqualifying, and so is anyone claiming to be “HIPAA certified,” because no such certificate exists. We map safeguards to the HIPAA Security Rule and offer a BAA to every US customer.
- “What’s the price, in full, after year one?” Get the whole ladder in writing. Ours is published, from $79/mo on the annual Assist plan, same scribe for in-person and virtual, no telehealth surcharge.
A scribe that answers those five cleanly will hold up across a hybrid week. Consent handling shifts a little between settings and states, so it’s worth reading our notes on patient consent by state and BAAs and consent before you roll out.
The honest bottom line for telehealth
An ambient scribe is a good fit for the talking-heavy visits telehealth does well: medication management, chronic-disease follow-up, counseling, results review. Those are conversation, and conversation is exactly what it captures. Where you truly need hands on the patient, telehealth is the wrong setting and a scribe won’t rescue it; that’s a triage call, not a software one.
The value is narrow and real. On a video visit, the scribe takes the conversation off your keyboard and leaves you one thing to write by hand: the objective section only you can judge. That’s less typing, faster sign-off, and more attention on the patient’s face instead of your charting. Book a demo to see the draft against your own visit types, then run the 7-day trial on a real telehealth week and read every note closely. That’s the evaluation that counts.