What should a clinic ask about any documentation option?

Run the same six questions with your physicians and with every option on the table: an in-house scribe, an AI tool built into the EHR, a third-party AI scribe, an outside human scribe service, or a combination. 1. How many minutes each day does each physician actually receive when the time is measured rather than estimated? 2. Who reviews the note before the physician signs it, and how long does that review take? 3. How many of our physicians use it in most of their visits? 4. What happens on a day it is not available? 5. How well does it handle our specialty and our most complex visits? 6. How much does it cost per physician when you take into account the time it takes to set it up and review it? A short pilot involving one or two doctors will typically answer them within a few weeks.

How much actual time do ambient AI scribes save?

In a randomized trial carried out at UW Health, ambient AI was associated with about 30 fewer minutes of documentation each day per provider. A multisite study published in JAMA in April 2026 compared more than 1,800 clinicians who used AI scribes with 6,770 clinicians who did not. The use of AI scribes was linked to 13 fewer minutes of total EHR time and 16 fewer minutes of documentation time over an eight-hour period of scheduled care, reductions of 3% and 10% respectively. There was no significant difference in after-hours EHR time between the two groups.

Who checks the note before the physician signs it?

Every option ends the same way: the physician reviews and signs the note. What changes is how much the physician has to fix. In a UC Davis pilot published in JMIR Medical Informatics, physicians assessed 356 AI-generated notes: 18% had accidental omissions, 11.5% had hallucinations, and 9.3% had accidental inclusions. Approximately 95% of the notes had no significant errors, although 5.3% contained errors that would pose a serious risk if not corrected. Review time is the hidden line item in every AI scribe budget.

Are your doctors really employing the tool?

Usage determines the results. In the JAMA multisite study, clinicians who used AI scribes in more than half of their visits saw about a 2x reduction in EHR time and a 3x reduction in documentation time, but only 32% of users reached that level. Pull usage by physician for the last 30 days; if use is uneven, it is worth asking why: specialty fit, note format, editing time, or patients who prefer not to be recorded.

What happens when the scribe or software isn't there?

In-house scribes go on vacation, fall ill, and leave the job, and software can experience outages, struggle in noisy environments, and depends on patients agreeing to be recorded. If either the scribe or the software is not available, the doctor completes the charting. Count the clinic days in the most recent quarter on which a doctor documented without the usual support, and ask about coverage for each option before you need it.

Does this align with your area of specialty and your most difficult visits?

With clear-cut visits, many AI programs generate clean drafts; complexity is where the options begin to differ. In the JAMA multisite study, primary care clinicians achieved the greatest time savings, while the differences for surgical specialties were not statistically significant. Pick the five visit types that take your physicians the longest to document and ask each option to show how it handles those. Specialty-trained human scribes learn the terminology, templates, and preferences of a specific physician, which matters most in the visits AI finds hardest.

How much does each option actually cost per physician?

AI subscription prices vary from low-cost self-serve tools to enterprise products priced by quote with implementation fees, while human scribe services are usually charged hourly for the hours the office schedules. A fair comparison also counts the physician's time spent reviewing and correcting drafts and unused licenses. Prepare a one-page comparison per physician listing the monthly cost, minutes saved each day (measured), review minutes each day, and the proportion of visits covered. The lowest subscription price does not always mean the lowest cost per completed note.

Is it possible for a human scribe to work with the AI tool that you currently have?

Yes. In a paired model, the AI records the encounter and prepares the first draft of the note. A trained scribe then reviews that draft, fills in any omissions, corrects any errors, applies the physician's template and specialty guidelines, and enters the note into the EHR. The physician then reviews and signs a completed note rather than editing the original draft. The practice keeps its AI investment, physicians spend less time on review, and a human check sits between the draft and the signature.

Don't Ghost Human Medical Scribes

Q: Where does this leave your practice?

Ambient AI is a useful tool, and it keeps improving. Human medical scribe services for clinics remain a strong option because they cover what AI still finds hard: careful review, complete notes, specialty nuance, and consistent use from one physician to the next. For many practices, the best result comes from using both. Running the six questions with one or two physicians over a few weeks gives the practice its own numbers, not vendor estimates.

Don’t Ghost Human Medical Scribes: What AI Alone Still Leaves Behind

By Nathaniel Smathers

Posted October 27, 2026

In AI, Medical Practice Tips, Scribe

Page Contents

This is for practice administrators and the doctors they support. We’ll look at why trained human scribes still matter in documentation, how remote scribing can be secure, and how scribes can work alongside your current dictation and AI tools.

Many practices have tried using an AI scribe, liked some features, and stopped talking about human scribes. This makes sense. AI subscriptions are easy to price, demos look great, and hiring is a hassle. But after the draft is done, someone still needs to catch missing exam findings, fix notes for complex visits, and handle cleanup. Often, that’s still the physician. Before moving on from human scribes, check if AI alone is really finishing the note.

Key Takeaways

Measure what AI alone actually finishes. In the largest multisite study to date, AI scribes saved about 16 minutes of documentation time per eight hours of patient care, and after-hours EHR time did not change significantly.
Ask who reviews and corrects every record before the physician signs it, and count that time as part of the cost of any option.
Require a business associate agreement and clear, revocable EHR access from any remote scribe partner, the same standard you would apply to an AI vendor or billing company.
Match the scribe to the specialty. Ophthalmology, orthopedics, dermatology, and other exam-heavy specialties document in ways general tools find harder, and ophthalmic scribes can hold a dedicated JCAHPO certification.
Pair instead of choosing. A trained scribe can work from the physician’s dictation, join the visit live, or finish the AI draft, then hand back a completed note for review and signature.
Check five recent notes for the details payers look for, such as the reason for each order, functional impact, and prior treatments with dates, before deciding what to change.
Start with one physician and the visit types that carry the most documentation risk.

What should you check before you ghost your human scribe?

Call it the “Five-Note Fright Check.” Pick one physician and five recent notes from visits that led to something a payer will review: advanced imaging, a surgery, therapy, durable medical equipment, or a specialty referral. Then run the same questions on whatever produced those notes today: the physician alone, dictation, an in-house scribe, an AI tool, an outside scribe, or a combination. Same questions, every option.

The Five-Note Fright Check

Was the chart closed the same day, or did it stay open into the evening?
Does each ordered service have a clinical reason stated in the note and tied to a diagnosis?
Is functional impact documented, meaning how the condition affects work, daily activities, or mobility?
Where the order depends on it, are prior treatments listed with dates, duration, and the patient’s response?
Is anything copied forward from a prior visit that does not match today’s encounter?
Who corrected the note before signature, and how long did that take?

If you can’t answer some of these questions right now, that’s helpful to find out early. Reviewing five notes honestly often reveals more than any vendor demo.

Run the same five-note check on every documentation option, and the comparison becomes fair.

Why do practices still need human medical scribes when AI is already in the room?

Ambient AI is a real step forward. Many physicians like it, and for simple visits, it quickly creates a usable draft. It’s important to say that up front.

The useful question is how much of the note AI finishes on its own. In a multisite study of 8,581 ambulatory clinicians at five academic health systems, published in JAMA in April 2026, AI scribe adoption was associated with 16 fewer minutes of documentation time per eight hours of scheduled patient care, a 10% reduction. Time in the EHR outside of work hours did not change significantly. The largest gains went to primary care clinicians, advanced practice clinicians, and frequent users, and only 32% of adopters used the tool in more than half of their visits.

The draft also needs a careful read. A UC Davis pilot published in JMIR Medical Informatics had physicians review 356 AI-generated notes and found accidental omissions in 18% and hallucinations in 11.5%. Most errors were minor, but 5.3% of notes contained errors that could pose serious risk if left uncorrected. Human scribes have their own record here: in the first randomized trial of medical scribes, published in the Annals of Family Medicine, physicians rated chart quality and accuracy higher with scribe support, and more charts were completed within 48 hours.

A useful test covers questions 1 and 6 from the check. For two weeks, have each physician note when the last chart of the day closes. If it closes before they leave the building, the current workflow is doing its job. If it closes at 9 p.m., that is the number worth improving. For physicians using AI, ask them to time their review of five drafts and jot down what they changed. One minute of light edits means the tool is serving them well. Five minutes and a missing finding means those five minutes belong in the cost comparison.

This is the gap with using only AI. AI can capture a conversation well, but it isn’t responsible for the finished note and doesn’t remember a physician’s preferences from visit to visit like a trained person does. If no one else handles the cleanup, it falls to the physician. For a closer comparison, check out our six-question review of human and AI scribes.

If physicians are still finishing notes at night, the documentation problem is still there.

Can a remote human medical scribe be safe for your practice?

The idea of “remote” can feel unsettling when protected health information is involved. Having a scribe in the room feels safer because you can see them, and that’s a reasonable feeling. Being physically present is an advantage, and the practice has direct control over the hire.

Remote scribing is safe when it is set up like any other HIPAA business relationship, which most practices already manage for their billing company, transcription service, cloud EHR, and AI vendor. HIPAA generally requires a covered entity to have a written business associate agreement with any outside party that creates, receives, maintains, or transmits protected health information on its behalf, and the same requirement carries down to that partner’s subcontractors. Business associates are also directly liable under the HIPAA Rules, so the obligation does not sit with the practice alone.

In practice, a good remote arrangement should feel familiar. The scribe uses your EHR with a unique login and role-specific access that your administrator can view and turn off if needed. Audio is sent through a secure session, usually with Zoom or a dictation app, not a personal phone. Nothing is stored outside your systems unless the agreement allows it.

Thought

Ask your current EHR administrator how quickly they could shut off an outside user’s access immediately. If the answer is minutes, the same control applies to a remote scribe.

Useful to ask any remote scribe partner (and any AI vendor)

Will you sign our business associate agreement, and do your subcontractors sign one with you?
Does each scribe use a unique login in our EHR, and can we see and revoke that access ourselves?
Where is visit audio stored, for how long, and who can hear it?
How are scribes trained on HIPAA, how often is it renewed, and how fast would we hear about a security incident?

Any option that listens to the visit, whether a live remote scribe or an ambient AI tool, raises the same practical question about how patients are informed. Recording-consent rules vary by state, so it is worth confirming your notice language with counsel once, for every listening option.

“Safe” also means the physician isn’t left alone. If a direct-hire scribe is out sick, the physician has to do all the charting that day. A remote partner with a backup scribe can fill in. Our comparison of direct hire and remote scribes explains cost and coverage in more detail. The real question isn’t “in the room or remote.” It’s “what controls are in place, and can we verify them in writing?”

Why do specialty practices benefit most from trained human scribes?

General AI tools do well when a visit follows a familiar conversational pattern. Many specialty visits do not. Much of the important information comes from the exam itself, not from what was said out loud, and it often belongs in specific structured fields.

The research points the same way. In the JAMA multisite study, time savings were concentrated in primary care. A Veterans Health Administration evaluation published in the Annals of Internal Medicine in April 2026 found clinician-written notes scored higher than notes from 11 AI scribe tools across quality measures, using test visits that deliberately included accented speech, background noise, and masked speakers, the conditions real clinics deal with every day.

Ophthalmology shows why specialty training matters. Clinicians record visual acuity, intraocular pressure, slit lamp and fundus findings, and test results as findings, often in dedicated EHR fields, and rarely say them as sentences an ambient tool can capture. A JAMA Ophthalmology study of 27 ophthalmologists found that 27% of exam time went to the EHR, and total EHR use averaged 10.8 minutes per encounter, or about 3.7 hours on a full clinic day, with 1.6 of those hours outside the clinic session. JCAHPO, the Joint Commission on Allied Health Personnel in Ophthalmology, offers a dedicated Ophthalmic Scribe Certification for this work, and DataMatrix’s JCAHPO-certified ophthalmic scribes document to that standard.

Other specialties have similar needs. Orthopedic notes rely on details like laterality, range of motion, special tests, imaging reviewed, and dates of conservative care. These are the same details required for imaging or surgery authorization. Dermatology visits may include several lesions, each having its own location, size, and appearance. Cardiology and neurology notes focus on symptom timelines, risk factors, and reviewed results. A note that looks good but misses the pressure reading or the correct side of the body isn’t helpful.

Here’s a helpful test

List the five types of visits that take your physicians the longest to document, like a new spine patient, a glaucoma follow-up, or a multi-lesion skin check. Ask each option to show how it handles those visits, not just a demo visit. An in-house scribe who knows a physician’s shorthand is a real advantage. Specialty documentation works best with someone who learns the physician’s templates, preferences, and the order for recording findings. That’s what our orthopedic scribes and other specialty-trained remote scribes are trained to do.

How do human scribes work with dictation and AI tools to support cleaner claims?

Many physicians already dictate, and many practices already pay for an AI tool. Neither has to go away. A trained scribe can fit around the workflow a physician already likes in three ways:

With dictation (offline): the physician dictates by phone, app, or recorder when it is convenient, and the scribe turns it into a structured EHR note, typically within 24 hours. This builds on the dictation and transcription habits many physicians already have.
Live: the scribe joins the visit remotely and documents in real time, so the note is close to finished when the patient leaves.
Paired with AI: the AI tool captures the encounter and produces a draft. The scribe reviews that draft against the physician’s template and specialty, fills omissions from the recording, corrects errors, and enters the note. The physician then reviews and signs a completed note instead of editing a rough draft.

Three ways a scribe fits the workflow you already have

With dictation (offline)

Physician dictates by phone, app, or recorder

→

Scribe builds a structured note

→

Note entered in your EHR, typically within 24 hours

→

Physician reviews and signs

Live

Scribe joins the visit remotely

→

Scribe documents in real time

→

Note close to finished when the patient leaves

→

Physician reviews and signs

Paired with AI

AI tool captures the encounter and drafts

→

Scribe checks the draft, fills omissions, corrects errors

→

Note entered in your EHR

→

Physician reviews and signs

Every path ends the same way: the physician reviews and signs the note.

Why this matters downstream: payers and auditors review the chart, not the intent behind it. In its Fiscal Year 2025 improper payment report, CMS estimated a Medicare fee-for-service improper payment rate of 6.55%, or $28.83 billion, and said most improper payments across its programs occurred when a reviewer could not determine whether a payment was proper because documentation was insufficient. In Medicaid, 77.17% of improper payments stemmed from insufficient documentation. CMS is clear that improper payments are not a measure of fraud. They are often a measure of what the record failed to show.

We see a similar pattern in our own work. In a DataMatrix review of 1,192 prior authorization denials across our book of business, 59% cited clinical criteria not met, 26% cited lack of medical necessity, and 15% cited both. This is internal data across all specialties, not an industry benchmark, but every one of those reasons depended on what the documentation supported, not on how quickly the request was submitted.

This is where questions 2 through 5 from the check matter. A trained scribe makes sure what the physician says and finds is included in the note in a way a reviewer can follow: the specific diagnosis, exam findings, functional impact, prior treatments with dates, and the reason for each order. Clinicians call this MEAT: Monitor, Evaluate, Assess, and Treat. A scribe also keeps old, copied text from earlier visits out of today’s note. When a scribe documents today’s visit, they write a new note for today.

There is a firm boundary here. A scribe never adds what the physician did not say or find. Scribes do not diagnose, select codes, or determine medical necessity; documentation must reflect what actually happened in the encounter. When something a payer typically expects is missing, the scribe flags it for the physician, and the physician decides whether it belongs.

Better documentation does not guarantee an approval. Payer policy, plan limits, and clinical criteria still decide. A complete note gives the physician’s decision the best chance to stand on its own and makes an appeal faster when needed.

A scribe can’t add anything the physician didn’t find, but they can make sure everything the physician did find is included in the note.

What if you didn’t have to choose between human and AI?

For many practices, the question is not whether to replace the AI tool or the in-house scribe. It is whether to add trained human capacity behind them. This has nothing to do with an in-house scribe’s quality or the value of the AI investment. It is about coverage and about who owns the finished note.

That capacity can be narrow and specific: one physician, the busiest clinic days, the visit types with the most documentation risk, or the AI drafts that take longest to fix. The practice keeps its tool, keeps its physicians’ dictation habits, and keeps the physician’s authority over every note. The scribe documents; the physician makes every clinical decision and reviews and signs the note.

Not sure what’s lurking in your notes? Let’s start with one physician.

Talk to DataMatrix Medical

Where does this leave your practice?

There’s no reason to stop using human medical scribes just because AI is available. Try the Five-Note Fright Check with one physician, time how long it takes to review a few AI drafts, ask your security questions in writing, and pay attention to the visit types that take the longest to document. After a few weeks, you’ll have your own data instead of relying on vendor estimates, which helps you make a better decision. For many practices, the best solution is using both AI and a trained human medical scribe together.

DataMatrix Medical has supported medical practices for over 25 years (300+ practices across 35+ specialties). Our trained remote medical scribes can work live or offline, enter notes directly into any EHR, use dictation and AI tools like DeepScribe and Ambula’s native AI tool, and include a backup. Plus, all scribes are JCAHPO certified. No long term contracts, 30 day cancellation notice.

References

Rotenstein LS, Mishuris RG, et al. “Changes in Clinician Time Expenditure and Visit Quantity With Adoption of Artificial Intelligence-Powered Scribes: A Multisite Study.” JAMA, April 2026. DOI 10.1001/jama.2026.2253. Summary via Mass General Brigham: https://www.massgeneralbrigham.org/en/about/newsroom/press-releases/ai-scribes-linked-to-modest-reductions-in-ehr-documentation-time
Taylor SL, et al. “Quality of Clinical Notes Created by Ambient Listening Generative AI: Pragmatic Prospective Pilot Study.” JMIR Medical Informatics, 2026. DOI 10.2196/86474. https://escholarship.org/uc/item/9r89g3sp
Veterans Health Administration evaluation of 11 ambient AI scribes vs. clinician-written notes. Annals of Internal Medicine, April 2026. Summary via Becker’s Hospital Review: https://www.beckershospitalreview.com/healthcare-information-technology/innovation/clinicians-produce-higher-quality-notes-than-ai-scribes-study/
Gidwani R, et al. “Impact of Scribes on Physician Satisfaction, Patient Satisfaction, and Charting Efficiency: A Randomized Controlled Trial.” Annals of Family Medicine, September 2017. DOI 10.1370/afm.2122. https://www.annfammed.org/content/15/5/427
Read-Brown S, Hribar MR, Reznick LG, et al. “Time Requirements for Electronic Health Record Use in an Academic Ophthalmology Center.” JAMA Ophthalmology, 2017;135(11):1250-1257. https://jamanetwork.com/journals/jamaophthalmology/fullarticle/2656337
Joint Commission on Allied Health Personnel in Ophthalmology (JCAHPO) and American Society of Retina Specialists. Press release on the Ophthalmic Scribe Certification (OSC), March 26, 2015. https://documents.jcahpo.org/documents/PressReleases/ASRS_JCAHPO_Announce_Initiative_CMS_Rule.pdf
U.S. Department of Health and Human Services, Office for Civil Rights. “Business Associates” guidance. https://www.hhs.gov/hipaa/for-professionals/privacy/guidance/business-associates/index.html
U.S. Department of Health and Human Services, Office for Civil Rights. “Sample Business Associate Agreement Provisions.” https://hhs.gov/ocr/privacy/hipaa/understanding/coveredentities/contractprov.html
Centers for Medicare & Medicaid Services. “Fiscal Year 2025 Improper Payments Fact Sheet.” January 15, 2026. https://www.cms.gov/newsroom/fact-sheets/fiscal-year-2025-improper-payments-fact-sheet

Nathaniel Smathers

Nathaniel Smathers is the VP of Client Education and Marketing. He is also a long time contributor of the DataMatrix Medical blog and has a background in healthcare content creation for over a decade. Nathaniel is passionate about exploring the intersections of healthcare, data analysis, and digital innovation.