A wRVU (work relative value unit) is the standard unit Medicare uses to measure the physician work in a service: the time, skill, effort, and stress it takes to perform it. A quick medication check is worth about 1 wRVU. A moderately complex office visit is worth about 2. Major surgery can be worth 20 or more.

If you’re an employed physician, there’s a good chance wRVUs matter more to your income than the dollars your visits collect, because most productivity-based compensation models pay a fixed rate per wRVU you generate. The wRVUs you get credit for are set by the codes you bill, and the codes you can defensibly bill are set by what’s in your note. Your documentation, not your effort, is what gets counted.

How wRVUs fit into the RVU formula

Medicare has priced physician services using the resource-based relative value scale (RBRVS) since 1992. Every CPT code carries three components:

• Work RVU (wRVU): the physician’s own effort
• Practice expense RVU: overhead, meaning staff, space, and equipment
• Malpractice RVU: liability insurance cost

Each component is adjusted for your geographic area, summed, and multiplied by the annual conversion factor to produce the Medicare payment. For 2026 the conversion factor is $33.40 for most practices ($33.57 for qualifying alternative-payment-model participants, a wrinkle your billing team will know). Exact values for any code are in the CMS Physician Fee Schedule lookup tool.

The wRVU is the piece employers use for compensation because it isolates your work from overhead and geography. A 99214 earns the same wRVU credit in Birmingham as in Boston.

How wRVUs become your paycheck

Most employed-physician productivity models work like this: your contract sets a dollar rate per wRVU, or a salary with a wRVU target and a bonus above it. Rates vary by specialty and market, commonly in the $40s to $80s per wRVU (MGMA and AMGA publish the benchmarks; your contract states yours).

The arithmetic is unforgiving in both directions. At an illustrative $50 per wRVU:

• A visit credited at 1.30 wRVUs pays you $65.
• The same visit credited at 1.92 wRVUs pays you $96.

Those aren’t hypothetical numbers. They’re the wRVU values of the two most common established-patient office visit codes.

wRVU values for common office visits (2026)

Code	Visit	wRVU
99212	Established patient, straightforward	0.70
99213	Established patient, low complexity	1.30
99214	Established patient, moderate complexity	1.92
99215	Established patient, high complexity	2.80
99203	New patient, low complexity	1.60
99204	New patient, moderate complexity	2.60
99205	New patient, high complexity	3.50
G2211	Add-on: visit complexity for longitudinal care	0.33
99417 / G2212	Add-on: each additional 15 minutes of prolonged visit time	0.61

Notice the step from 99213 to 99214: 0.62 wRVUs, a 48% increase in work credit for the same slot on your schedule. The two add-on codes at the bottom of the table are the other lever most physicians leave unpulled; more on them below.

Why your documentation sets your wRVUs

Since the 2021 AMA E/M revisions, office visit levels are chosen by medical decision making (MDM) or by total time, and both live or die in the note. If you managed two chronic conditions, one of them not at goal, reviewed outside records, and adjusted a prescription, that’s the substance of a moderate-complexity visit. But if the note doesn’t capture the outside records you reviewed or the second condition you addressed, and what remains no longer reaches two of the three moderate columns, then unless your qualifying time supports a higher level, you are down to whatever lower level the note does support, and that is what you can defensibly bill. The work happened; the credit didn’t.

Doing level-4 work and writing level-3 notes is common enough to have a name: undercoding. A physician whose documentation under-supports one visit per clinic day is forgoing roughly 155 wRVUs a year (0.62 × ~250 clinic days), which at $50 per wRVU is about $7,750 in personal compensation. That loss comes from documentation gaps, not from seeing fewer patients.

How to document visits so your notes support the wRVUs you earn

None of what follows is about billing for work you didn’t do. It’s about making sure the note contains the evidence for the work you did. Under the 2021 rules, an office visit level is determined by the higher of two paths, MDM or total time, and you only need to satisfy one. Most physicians default to MDM, so start there.

Understand the MDM table: you need two of three columns

MDM has three elements. To reach a given level you must meet or exceed that level in two of the three:

1. Number and complexity of problems addressed
2. Amount and complexity of data reviewed and analyzed
3. Risk of complications from patient management

The practical consequence: you don’t have to be exhaustive everywhere. If your problems column and your risk column both clearly reach “moderate,” the visit is a 99214 regardless of how much data you reviewed. Knowing which two columns you’ll rely on for a given visit tells you what has to be in the note.

Column 1: Document every problem you addressed, and its status

“Addressed” means you evaluated or managed it at this visit. The visit was about the knee, but you also refilled the lisinopril and looked at the home blood pressure log. That’s two problems addressed, not one, and problems addressed are the first column of the MDM table.

Two habits that move this column:

• List each problem separately in the assessment, even the ones that took thirty seconds. A single line reading “HTN, controlled, continue lisinopril” is enough to count.
• State the status. A chronic illness that is stable counts as low complexity. A chronic illness that is not at treatment goal, or is worsening, counts as moderate. “Type 2 diabetes, A1c 8.4, above goal, increasing metformin” is a moderate-complexity problem; “diabetes, continue meds” is not, even if the clinical situation is identical. Two stable chronic conditions also reach moderate on their own.

Column 2: Capture the data you reviewed, ordered, or interpreted

Reviewed-but-undocumented data does not exist for coding purposes. The data column reaches “moderate” if you hit any one of three categories:

• Category 1: any combination of three of the following: reviewing prior external notes from each unique source (the cardiologist’s letter, the ED visit), reviewing the result of each unique test, ordering each unique test, or obtaining history from an independent historian (a spouse, a caregiver) where the assessment requires it.
• Category 2: independent interpretation of a test performed by another physician, such as reading the actual ECG tracing or looking at the imaging yourself rather than just the report. It does not count if you separately bill for that interpretation.
• Category 3: discussion of management with an external physician or appropriate source, provided that discussion is not part of a service you report separately.

The documentation failure here is almost always omission. You looked at the outside echo report, you ordered a BMP, and you ordered a lipid panel. That’s three Category 1 items, but if the note says only “labs reviewed,” a coder can’t count them. Name each test and each external record. Ordering a test and reviewing that same result count once between them, not twice.

Column 3: Document the management decision, not just the plan

The risk column is about the risk of the management options you considered and chose, and it’s the column physicians most often under-document because the decision felt routine.

• Prescription drug management is moderate risk. That includes starting, stopping, adjusting, or deciding to continue a prescription. “Continue amlodipine 5 mg; BP at goal, no changes” is a documented management decision. “Continue meds” is arguably not.
• Decisions about hospitalization, elective major surgery, or escalation of care are moderate or high risk. If you considered admitting and decided against it, say so. The decision not to escalate is a decision.
• Social determinants of health that significantly limit diagnosis or treatment (a patient who can’t afford the medication, can’t get to the imaging center) count as moderate risk when they change your management. Document the constraint and what you did about it.

The time path: when to use it and what to write

On some days MDM undersells the visit. A 45-minute counseling visit with an established patient for a newly diagnosed condition may be low-complexity MDM but is a 99215 on 45 qualifying minutes. For a new patient the same 45 minutes supports a 99204. Time-based coding counts your own qualifying time on the date of service: pre-visit chart review, the face-to-face visit, ordering, documentation, and care coordination that day. Time spent on services you report separately does not count toward the E/M. The 2026 thresholds for established patients are 20–29 minutes (99213), 30–39 (99214), and 40–54 (99215); for new patients, 30–44 (99203), 45–59 (99204), and 60–74 (99205).

To use it, the note must state the total time and, ideally, what it covered. One sentence at the bottom of the note does it: “Total time on date of service: 42 minutes, including review of outside records, counseling on treatment options, and documentation.”

Two add-on codes that add wRVUs to visits you’re already doing

G2211 (visit complexity, 0.33 wRVU). Medicare’s add-on for visits where you are the continuing focal point of the patient’s care, or are managing a single serious or complex condition over time. It attaches to any office visit level, so a 99214 with G2211 credits 2.25 wRVUs instead of 1.92. It requires no separate note, but the visit documentation should make the longitudinal relationship visible: reference the prior visit, the ongoing plan, and the follow-up interval. One important exclusion: it generally cannot be billed when the visit carries modifier 25 for a same-day procedure. Medicare does allow it alongside same-day vaccine administration and Part B preventive services such as the annual wellness visit. Commercial payer coverage varies, and whether your employer credits the wRVU depends on your contract, so ask.

Prolonged services (99417 for most commercial payers, G2212 for Medicare, 0.61 wRVU per 15 minutes). When total time runs well past the 99215 or 99205 threshold, each additional 15-minute block is separately creditable. The two codes have different starting thresholds by payer, so confirm with your coder, but the documentation requirement is the same as the time path above: a stated total time.

Before and after: the same visit, two notes

A synthetic example (no real patient data). Established patient, 20-minute slot, hypertension and hyperlipidemia follow-up, brought a home BP log and a cardiology consult letter.

Note A (thin documentation):

Assessment/Plan: HTN, HLD. Meds adjusted. Labs reviewed. RTC 3 months.

Note B (supports 99214):

Assessment/Plan:

1. Hypertension, not at goal. Home BP log reviewed, average 148/88 over 2 weeks. Increase lisinopril from 10 mg to 20 mg daily. BMP in 2 weeks.

2. Hyperlipidemia, stable on atorvastatin 40 mg. Reviewed lipid panel from 6/2026 (LDL 72, at goal). Continue.

3. Reviewed cardiology consult letter from Dr. Smith dated 7/2026; agree with recommendation for annual echo.

RTC 3 months.

Same encounter, same physician work. Note B reaches moderate in the problems column (one chronic condition not at goal, plus a second stable chronic condition), moderate in the data column (external note reviewed, unique test reviewed, unique test ordered: three Category 1 items), and moderate in the risk column (prescription drug management). Note A describes the same encounter and the same plan, but a coder reading it cannot tell which problem was not at goal, what data you reviewed, or which medication changed. The difference can be up to 0.62 wRVUs, and the physician’s effort was identical.

The gaps that cost a level most often

• The second and third problems addressed never make it into the assessment.
• “Labs reviewed” instead of naming each test and each outside record.
• A prescription continued without the word “continue” and the reason.
• A hospitalization considered and rejected, undocumented.
• Total time not recorded on long visits.
• No mention of the ongoing care relationship on visits that would support G2211.

Every one of these is a documentation habit, not a clinical one. Fixing them costs little or no additional patient time.

What it adds up to

Illustrative, at $50 per wRVU and ~250 clinic days:

• One visit per day documented to the 99214 it supported instead of the 99213 it was coded at: ~155 wRVUs, ~$7,750 a year.
• G2211 captured on five eligible longitudinal visits per day, if your contract credits it: ~410 wRVUs, ~$20,600 a year.
• Both together come to about 568 wRVUs, which for many specialties is the equivalent of several weeks of clinic.

Run it with your own rate, your own visit mix, and your own contract terms. The point isn’t the specific number; it’s that documentation is a productivity lever that doesn’t require seeing another patient.

Where NovaScribe fits

This is the problem NovaScribe™ was built around. Because notes are generated from the full encounter conversation, the second problem you addressed, the outside letter you discussed, and the medication you decided to continue are in the draft without you reconstructing them from memory at 6 p.m. Every note on the Professional tier comes with a suggested CPT code and its wRVU estimate, the ICD-10 diagnoses, and an analysis of what the documented MDM supports, including where the documentation could support a higher level. You review, you decide, you sign. The tool’s job is to make sure the note reflects the visit that actually happened.

Frequently asked questions

Is a wRVU the same as an RVU?

No. The RVU is the total (work + practice expense + malpractice). The wRVU is only the physician-work component, which is the piece compensation formulas use.

Who sets wRVU values?

CMS, with heavy input from the AMA’s RUC (Relative Value Scale Update Committee), which surveys physicians about the work involved in each service. Values are published in the Physician Fee Schedule and updated annually.

Can better documentation alone raise my wRVUs?

Only to the extent the work was actually done. Documentation doesn’t create work; it records it. The gains described here come from notes finally reflecting problems addressed, data reviewed, and decisions made that were previously left out. Coding above what the visit supports is fraud, and no documentation tool changes that.

Do wRVUs matter if I’m in private practice and paid on collections?

Less directly, but the same documentation drives your E/M level, and the E/M level drives the payment. In 2026, the Medicare gap between a 99213 and 99214 is about $40 per visit. The note carries the same weight either way; only the denominator changes.

A note from NovaScribe

wRVUs are how the system counts your work, and the note is the only evidence the system accepts. You already do the work. Make sure the documentation keeps up with it. If you’d like to see wRVU estimates and MDM analysis on your own notes, the free tier includes the full billing analysis on 10 notes a month, no credit card.

See wRVU estimates on every note, free

An AI medical scribe listens to your patient encounter and drafts the clinical note for you. You review it, edit it, sign it, and move to the next room. Done well, it returns one to two hours of your day. The market now has more than a dozen options, priced anywhere from free to more than $1,000 per provider per month, and the differences between them are not obvious from their homepages.

Full disclosure before anything else: I’m Patrick Proctor, a practicing cardiologist, and I built NovaScribe™, one of the products in this guide. I use it in clinic every day. You should read my product’s entry with that in mind, and I’ve tried to be as fair to the others as I’d want them to be to me. Where a competitor does something better than we do, I say so.

What is an AI medical scribe?

An AI medical scribe is software that records the conversation between you and your patient, transcribes it, and generates a structured clinical note, typically a SOAP note, H&P, or progress note. A smaller number of tools, NovaScribe among them, also generate operative notes from a dictated procedure summary. You don’t dictate the note itself. You just talk to your patient the way you normally would.

These tools exist because documentation has swallowed the workday. A time-and-motion study in Annals of Internal Medicine found that ambulatory physicians spend nearly two hours on EHR and desk work for every hour of direct patient care. A 2024 analysis in Annals of Internal Medicine found primary care physicians now average 2.7 hours of after-hours EHR work per day. That’s the problem every product below is trying to solve.

How we evaluated these tools

Five criteria, weighted toward what matters for an independent or small-group practice:

• Note quality and editing burden. How much do you have to fix before you can sign? That is what matters most. A scribe that saves you typing but costs you ten minutes of editing per note hasn’t saved you anything.
• Speed. How long between ending the encounter and having a draft in front of you?
• Billing support. Does the tool surface CPT codes, ICD-10 diagnoses, and E/M level analysis, or does it stop at the narrative note? For physicians paid on productivity, whether direct revenue or wRVUs, this is where a scribe pays for itself. (More on that math in our AI medical scribe pricing guide.)
• EHR fit. Some tools write directly into Epic or athenahealth. Others work with any EHR by copy-paste. Deep integration sounds better until you see the price tag and the implementation timeline attached to it.
• Total cost. Monthly price, plus implementation fees, contract commitments, and minimum seat counts.

One thing we deliberately did not score: clinical accuracy claims. Every vendor makes them and few publish head-to-head data, so judge note quality yourself, on your own patients, during a free trial. Nearly every tool here offers one.

The best AI medical scribes in 2026

1. NovaScribe: Best for independent physicians paid on productivity

Our product, so apply skepticism accordingly. NovaScribe generates a formatted note in under 20 seconds and works with any EHR by copy-paste, no integration project, no IT department required. You customize your notes with plain-language instructions (“keep my assessments in bullet points,” “always include a lipid trend”) rather than rigid templates. It also generates operative notes for cardiology and orthopedic procedures, which most tools in this price range don’t attempt.

There are two self-serve paid tiers. Basic ($99/month, or $999/year) is unlimited notes with dual-pass transcription, prior-visit patient context and plain-language note customization, without the billing suite; if you’re on a straight salary and just want a note you don’t have to rewrite, that’s the tier to compare against Freed and Heidi. Professional ($229/month) is the differentiator. Every note comes with suggested CPT codes with wRVU estimates, ICD-10 diagnoses, E/M level analysis, and suggestions for where your documentation could support a higher level of medical decision making. If you’re paid per wRVU or on collections, that layer is the reason to pick us over cheaper options.

The free tier includes 10 notes per month with all Professional features and no credit card. Limitations: no direct EHR write-back (copy-paste is the workflow, by design), and we are a small company, so you get founder-level support but not a 24/7 enterprise help desk.

2. Freed: Best-known budget option

Freed built its reputation on simplicity and price, currently $39 to $119 per month depending on tier, with a 7-day free trial. Setup takes minutes and clinicians generally describe the notes as solid for routine visits. Its Premier tier includes ICD-10 and CPT codes, and it has added specialty templates over time. If you want an inexpensive scribe and do not need wRVU dollar value or E/M level analysis, Freed is a reasonable place to start. See our full NovaScribe vs Freed comparison.

3. Doximity Scribe: Best free option

Free for verified U.S. physicians, NPs, PAs, and medical students, with no paid tier at all. It lives inside the Doximity app many clinicians already use, works by copy-paste with any EHR, and can append billing codes to a note on request. For low-volume users, medical students, and anyone who wants to try ambient documentation at zero cost, it is hard to beat.

The tradeoffs are the flip side of free: it’s one feature inside a network and advertising platform rather than a dedicated documentation product, and its customization and billing analysis are lighter than the paid tools here. We wrote a full breakdown in NovaScribe vs Doximity Scribe.

4. DeepScribe: Best for specialty tuning with human review

DeepScribe pairs its AI with specialty-specific tuning and offers human-in-the-loop review options. Pricing is quote-based, generally reported in the $350–$500 per provider per month range, with a sales process rather than self-serve signup. It suits practices that want a managed, higher-touch deployment. Our NovaScribe vs DeepScribe comparison covers the details.

5. Abridge: Best for large health systems on Epic

Abridge sells to health systems, not individual physicians. It integrates deeply with Epic, has major enterprise deployments, and is priced through enterprise contracts (pricing is not published; third-party estimates run from about $200 to $1,200+ per provider per month). If you’re a CMIO evaluating a system-wide rollout, Abridge belongs on your shortlist. If you’re a solo or small-group physician, you can’t really buy it. Comparison: NovaScribe vs Abridge.

6. Microsoft Dragon Copilot (formerly Nuance DAX Copilot): The enterprise incumbent

Microsoft merged DAX Copilot and Dragon Medical One into Dragon Copilot in March 2025. It has the deepest Epic integration in the market and the largest installed base. It’s also enterprise-only: no self-serve trial, reseller pricing from roughly $369 per provider per month with a $700 implementation fee and 12-month commitment, and reported quotes running to $600+ for small groups. A NEJM AI study at Atrium Health found time savings per note were real but didn’t translate into system-level productivity gains.

7. Suki: Best voice-command assistant

Suki is more than a scribe; it’s a voice assistant that can stage orders, answer chart questions, and write notes directly into Epic, Oracle Health, athenahealth, and MEDITECH. Industry-reported pricing runs $299 per month for Suki Compose and $399 for Suki Assistant, on annual contracts with setup fees. If hands-free EHR control is what you want and the budget supports it, Suki is the most mature option.

Also worth knowing

Heidi Health offers a free plan with paid tiers from about $30/month. Nabla has free and paid tiers and a strong reputation in primary care. Twofold Health is a budget option around $49/month. All three are credible; none currently offers the CPT/wRVU/E/M analysis layer that defines the productivity-pay use case.

Comparison table

Product	Typical price/provider/mo	Free option	Billing codes	EHR approach	Buying process
NovaScribe	$99–$229	10 notes/mo, full features	CPT + wRVU, ICD-10, E/M, MDM	Any EHR, copy-paste	Self-serve
Freed	$39–$119	Trial	ICD-10 + CPT (Premier tier)	Copy-paste + integrations	Self-serve
Doximity Scribe	Free	Fully free	On request	Any EHR, copy-paste	Self-serve
DeepScribe	Not published (est. $350–$500+)	Demo	Yes, varies by plan	Integrations	Sales process
Abridge	Not published (est. $200–$1,200+)	No	Enterprise coding tools	Epic-deep	Enterprise sales
Dragon Copilot (DAX)	~$369–$600+	No	Voice-query coding	Epic-deep	Enterprise sales
Suki	~$299–$399	Trial	Ambient CPT/E/M	Epic, Oracle, athena, MEDITECH	Sales process

---

Try NovaScribe For Free. Ten notes a month, every Professional feature switched on, no credit card.

Pricing reflects published and reseller-reported figures as of mid-2026; confirm current pricing with each vendor.

How to choose

Start with two questions.

Who’s buying? If the answer is a health system with an IT department and an Epic contract, look at Abridge, Dragon Copilot, and Suki. If the answer is you, look at the self-serve tools: NovaScribe, Freed, Doximity, Heidi, Nabla.

How are you paid? If your compensation runs through wRVUs or collections, a scribe that only writes the narrative solves half your problem. The note determines the code, and the code determines what you’re paid. That’s the case for NovaScribe’s Professional tier, and it’s a fair case for Suki’s coding features at the enterprise level.

Then trial two or three on real patients for a week each. Note quality is personal. It depends on your specialty and on how you talk in the room. No review, including this one, substitutes for reading a draft of your own note.

Frequently asked questions

Are AI medical scribes HIPAA compliant?

The established vendors, including everyone named above, operate as HIPAA business associates and will sign a BAA. Ask two sharper questions instead: is your audio used to train AI models, and how long is it retained? (NovaScribe’s answers: never, and audio is deleted within 6 hours.)

Do AI scribes work for my specialty?

Most cover the common outpatient specialties well. Ask about yours specifically. NovaScribe supports 15 specialties, including operative notes for cardiology and orthopedics.

Will an AI scribe increase my revenue?

Peer-reviewed evidence suggests documentation tools can affect coding: one 2026 policy brief reports a health system that saw an 11% rise in wRVUs after ambient scribe deployment. Your results depend on whether your current documentation under-supports the codes you could defensibly bill. Our wRVU explainer walks through that math.

A note from NovaScribe

You don’t need us to tell you documentation is broken; you live it every clinic day. Any tool on this list will help. The right one depends on who’s paying for it, how you’re paid, and how much editing you’ll tolerate. Trial the free options first, including ours, and let your own signed notes make the decision.

---

Try NovaScribe For Free. Ten notes a month, every Professional feature switched on, no credit card.

An AI medical scribe costs anywhere from $0 to more than $1,000 per provider per month in 2026. That range is so wide it’s almost useless, so this guide breaks it into bands, explains what each band actually buys, and walks through the fees that don’t appear on pricing pages. It ends with the arithmetic for deciding whether a scribe pays for itself in your practice.

We sell one of these products (NovaScribe™), so treat this the way you’d treat any vendor’s market guide: check the sources, which we’ve linked throughout.

The short answer

Band	Monthly cost per provider	Representative products
Free	$0	Doximity Scribe, Heidi (free plan), Nabla (free tier), NovaScribe free tier
Budget	$30–$119	Twofold (~$49), Freed ($39–$119), Heidi paid (from ~$30), NovaScribe Basic ($99)
Mid-market	$229–$500	NovaScribe Professional ($229), Suki ($299–$399), DeepScribe ($350–$500, quoted)
Enterprise	~$200–$1,200+, contracted	Dragon Copilot/DAX, Abridge

For context on what sits inside each product, our full comparison of the best AI medical scribes covers them one by one.

What the price bands buy

Free ($0)

Several tools offer permanent free access, not just trials. Doximity Scribe is free for verified U.S. clinicians with no paid tier. Heidi and Nabla maintain free plans with limits. NovaScribe’s free tier includes 10 notes per month with all Professional features and no credit card.

Free tiers make sense for low visit volumes, trainees, and evaluation. The limits are volume caps or lighter feature sets; a clinician documenting 15–20 encounters a day will outgrow most of them in the first week, which is what they’re designed for.

Budget ($30–$119)

This band buys unlimited (or high-cap) note generation with solid quality for routine visits. Freed, the best-known name here, runs $39 to $119 per month depending on tier. NovaScribe’s Basic tier ($99/month, or $999/year) sits at the top of this band: unlimited notes, dual-pass transcription, prior-visit patient context, and plain-language note customization, without the billing suite. What this band generally does not include is billing analysis: you get the narrative note, and coding remains your job. If that’s all you need, Basic and Freed’s upper tiers are the right comparison; the billing layer is what the next band is charging for.

Mid-market ($229–$500)

Here the products differentiate. NovaScribe’s Professional tier ($229/month, or $2,299/year) takes everything in Basic and adds CPT codes with wRVU estimates, ICD-10, E/M level analysis, and MDM suggestions to every note. Suki, reported at $299–$399 per user per month, adds voice commands and direct EHR write-back for Epic, Oracle Health, athenahealth, and MEDITECH. DeepScribe quotes per practice, generally reported at $350–$500, with specialty tuning and human-review options. At this level the price covers what happens after the transcript, not the transcription itself.

Enterprise (contracted)

Dragon Copilot (formerly Nuance DAX Copilot) and Abridge sell through enterprise sales, not self-serve checkout. Reseller listings put DAX-line pricing at $369 per provider per month with a $700 implementation fee and 12-month commitment, with small-group quotes reported at $399–$600+ and large-system blended pricing lower. Abridge does not publish pricing; third-party estimates run from about $200 to $1,200+ per provider per month, contracted through the health system. The product you’re buying in this band is deep EHR integration plus organizational deployment, and the price assumes an organization is paying for it.

The costs that aren’t on the pricing page

Four to ask about before you sign anything:

• Implementation fees. Enterprise tools carry one-time fees from $700 per provider to five figures per organization. Self-serve tools have none.
• Contract commitments. Twelve-month terms are standard in the enterprise band and common at Suki and DeepScribe. Month-to-month is the norm for self-serve tools.
• Annual vs monthly billing. Most self-serve tools discount annual prepay 15–20%. NovaScribe’s Professional annual plan works out to about 17% off monthly.
• Your time. Enterprise deployments run weeks to months and involve your IT resources. Self-serve tools work the day you sign up. That difference rarely appears in a cost comparison, and it’s real money.

How AI scribes compare to human and virtual scribes

The traditional alternatives set the ceiling. An in-person human scribe runs roughly $32,000–$42,000 per year per provider, call it $2,700 to $3,500 per month, before benefits and turnover. Virtual scribe services typically run $1,200–$4,000 per month per provider. Against either, even the most expensive AI scribe is a fraction of the cost. The counterpoint: a good human scribe does things software doesn’t, like managing your inbox or queueing orders. The question is whether those extras are worth a 5–20x price difference for your practice.

Does an AI scribe pay for itself?

Two ways to run the math. Use your own numbers, not a vendor’s.

Time. The Sinsky time-and-motion study in Annals of Internal Medicine found ambulatory physicians spend nearly two hours on EHR and desk work for every hour of patient care, and later AMA-supported research put primary care after-hours EHR work at 2.7 hours per day. If a scribe returns even 45 minutes a day, price that hour however you value it: added visits, earlier sign-out, or simply getting your evening back.

Revenue. This one is more concrete. At 2026 Medicare national rates, a 99214 reimburses about $40 more than a 99213. If more complete documentation lets you defensibly support one additional level-4 visit per clinic day that you’re currently coding at level 3, that’s roughly $800 per month at Medicare rates alone, several times the cost of any mid-market subscription. Whether that applies to you depends on whether you’re currently undercoding, which is more common than most physicians assume. Our wRVU explainer covers the numbers. (If wRVUs drive your compensation instead of collections, the same math runs through your wRVU rate.)

How NovaScribe prices

For transparency, our full price list: Free ($0, 10 notes/month, all Professional features), Basic ($99/month or $999/year, unlimited notes without the billing suite), Professional ($229/month or $2,299/year, unlimited notes plus CPT with wRVU estimates, ICD-10, E/M and MDM analysis), and Enterprise (contact sales for multi-user administration and volume pricing). No implementation fees, no contracts on monthly plans, works with any EHR by copy-paste.

Frequently asked questions

Why do prices vary this much for similar-sounding products?

Because “AI scribe” describes the transcription layer, and transcription is becoming a commodity. The price differences reflect what’s built on top: billing analysis, EHR write-back, voice commands, human review, enterprise deployment.

Is the free tier of anything actually usable?

Yes, within its limits. Doximity Scribe is fully free. Free tiers are the cheapest possible way to find out whether ambient documentation fits how you practice.

Do any of these charge per note?

A few tools meter usage on lower tiers (that’s what free-tier caps are). Established paid plans in every band above are flat-rate unlimited.

A note from NovaScribe

The number on the pricing page is the smallest part of this decision. What matters is the total: fees, contracts, deployment time, and, on the other side of the ledger, the hours and billing capture a good scribe returns. Run the math with your own visit volume and payer mix. If you want to see what our billing analysis surfaces on your actual notes, the free tier exists so you can check before paying anything.

View pricing