Commure AI Scribe: Workflow, Privacy, Pricing and Evidence (2026)

Commure AI scribe is documentation software for turning encounters into draft clinical notes. The buying decision starts with the offering: self-service Scribe, an enterprise Ambient deployment, or a service that adds human documentation support. Those routes differ in staffing, EHR delivery and purchasing terms—not just price. Commure’s Ambient offerings.

Our assessment: shortlist Commure when your requirement includes an EHR-connected documentation workflow or a choice between automated and staffed support. For a small practice, evaluate the self-service trial separately from an enterprise proposal. The strongest decision evidence is a corrected note that reaches the right encounter with its unfinished work intact; a polished draft alone is not enough.

Published by Vero, a competing healthcare documentation software provider. This public-source guide covers documented capabilities, buying considerations and proposed workflow tests. Sources checked October 8, 2026.

Which Commure offering are you evaluating?

Commure’s public Ambient page describes four configurations: Ambient AI for automated documentation; Ambient Assist with medical documentation specialists refining drafts; Ambient Live with dedicated pre-, during- and post-visit support; and Ambient+ combining ambient documentation, coding and dictation. Pricing for these service levels is separate from the self-service plans below. Current Ambient overview, checked October 8, 2026.

Choose the workflow before requesting a demonstration:

• A clinician who wants to prepare and review their own notes: start with the self-service route and one representative template. Measure the work remaining after generation.
• A practice needing notes placed into an existing EHR: evaluate the exact integration and account entitlement, including who configures field mapping.
• A service needing staff to prepare or quality-check documentation: compare the staffed options, delivery hours, turnaround and replacement coverage. Ask which tasks people perform and which remain with the clinic.
• A buyer seeking coding automation: specify whether the requirement is suggested codes, reviewed coding or charge submission. A product family name cannot define that boundary.

The self-service features page documents specialty templates, SOAP and consultation formats, customization, multi-speaker recognition, coding suggestions, telehealth support, 60-plus languages, and iOS, Android and desktop access. Evaluate language performance and compatibility with your clinic’s devices, specialty and destination.

For example, a template can place the history and plan in separate sections, but it cannot supply an examination that never occurred. A speaker label can help distinguish the patient from a companion, but the reviewer still needs to check whose statement became whose history. Test the functions that could change meaning, not only the appearance of the finished note.

Commure Scribe pricing, trial and exit terms

The live pricing toggle, checked October 8, 2026, shows $89 with monthly billing and $708 billed annually. The annual amount equals $59 per month, paid annually. At unchanged published rates, 12 monthly payments total $1,068: $360 more than the displayed annual offer, before taxes or promotions. Monthly billing reduces the prepaid commitment; the annual offer costs less over a full year.

Separately, the Scribe landing page advertises $45 for the first month, with $89 shown as the reference price. Keep this introductory offer separate from ongoing monthly and annual costs, and save the checkout terms or written quote. Landing-page offer checked October 8, 2026.

The pricing FAQ says paid access continues through the billing period after cancellation and subscription payments, including annual payments, are nonrefundable. It recommends cancelling several days before the next charge. Record the renewal date and export route before committing; enterprise contracts may differ.

What a usable quote should contain

For an enterprise deployment, request one schedule listing the named offering, clinician licenses, minimum commitment, integration and setup charges, human-service hours, training, support, and renewal terms. Add the cost and format of retrieving records when leaving. Keep each quoted unit visible: a provider seat, booked staff hour and location license cannot be compared as if they buy the same thing.

A sample clinic budget shows why the subscription is only one component. Suppose a clinic budgets 60 currency units per month for software, 240 for setup spread over six months, and 200 encounters monthly. At two minutes of review and transfer per encounter and 120 units per clinician hour, active work costs 800. Total monthly cost is 60 + 40 + 800 = 900, or 4.50 per encounter. At four minutes, it becomes 1,700, or 8.50 per encounter.

Replace the sample inputs with your quote and measured workload. Use the same currency and volume for each option, include failed sessions and staff effort, and compare with the clinic’s baseline documentation work. Evaluate the complete workflow, including review and delivery. Our scribe pricing guide develops the wider cost model.

Workflow: from capture to the saved EHR note

The operational sequence is identify encounter → capture → generate → correct → transfer → inspect and authenticate. Human-supported services insert preparation and quality-assurance work into that chain; the final acceptance still belongs to an accountable clinician.

1. Configure the encounter and template

Use an approved account and confirm the patient, visit and selected template. Decide which sources the system may use. Prior notes, today’s conversation and a companion’s statement have different provenance. During evaluation, deliberately leave one expected document unavailable and check whether the output preserves that gap.

2. Capture the actual speakers

For an in-person test, record the device, microphone, speaker positions and background conditions. For telehealth, identify how remote audio reaches capture, particularly when a headset is in use. Include a test in which one channel is missing. A note must not imply that an unheard participant supplied information.

Commure documents multi-speaker and telehealth functionality. Check how stop, resume, closing the application and deletion work in the selected version. The ambient scribe guide covers these recording boundaries in more detail.

3. Review meaning before formatting

Read the source and draft side by side. Check omissions as well as changed statements: a note can contain only accurate sentences yet lose the person responsible for requesting an outside report. Inspect laterality, numerical units, temporal context, uncertainty and whether an action is planned or completed. Review coding separately from prose; a readable paragraph is not approval of a billing decision.

4. Verify the integration route

Commure’s eClinicalWorks page describes appointment-linked background sync, official API integration and browser-based writing into an open eCW session. It says notes map to 14-plus Progress Note fields and identifies Scribe-It as a required add-on for the commonly described onboarding route. Documentation checked October 8, 2026.

That distinction changes the acceptance test. Background sync must select the right appointment without relying on the currently open chart. Browser-based delivery must handle the active patient, field focus and interrupted session correctly. For either route, have the implementation team demonstrate a corrected plan arriving in its intended field, an explicit signature state and recovery after an uncertain transfer. Record the route enabled in your tenant; do not treat the three descriptions as interchangeable entitlements.

5. Reopen the destination

End the test in the receiving EHR. Compare the saved text with the approved draft, including unfinished requests and corrections. Check that a retry did not duplicate the note or overwrite a newer version. Separate successful text delivery from an order being placed, a message being sent or a clinician authenticating the record. The EHR integration acceptance guide provides a broader destination checklist.

Privacy, retention and human access

Start with the legal entity, specific service and agreement incorporated into the order. For U.S. HIPAA-regulated use, HHS explains that an appropriate business associate agreement and the customer’s safeguards are both necessary when a cloud service handles ePHI. Encryption does not, by itself, remove the cloud provider’s business-associate role. HHS cloud guidance, checked October 8, 2026.

Retention by data type

The current public Commure privacy policy, displaying an effective date of November 18, 2025, describes retaining personal data during an account’s lifetime and up to six years afterward, depending on data type, contract and law. It also describes indefinite anonymized-data retention and deletion limitations. Its scope excludes certain processing performed for third parties under their policies. Confirm the audio lifecycle for your specific Scribe deployment in its agreement and configuration. Checked October 8, 2026.

Obtain a copy inventory for the contracted workflow:

• Audio and transcripts: creation, access, retention trigger and deletion process.
• Drafts and corrected notes: version history, export format and post-cancellation access.
• Human-service workspaces: staff locations, permitted access and any additional copies.
• Integration logs, support attachments and backups: content captured, recipients and expiry.
• Receiving EHR: the clinic’s record-retention and amendment procedures, separately from the scribe account.

Ask for the configured values and contract references, not simply a statement that information is secure.

De-identified data and exit obligations

Commure’s public BAA permits de-identification under the cited HIPAA standard and use or disclosure of de-identified data for lawful purposes. On termination, it provides for returning or destroying PHI at the customer’s expense, with continued protections where that is infeasible. The general agreement separately addresses rights in de-identified data. Match these documents to the signed service agreement; ask specifically about model improvement, opt-outs and surviving rights. Both checked October 8, 2026.

For staffed support, include who can hear recordings or read drafts, subcontractor access, auditability and the response to a patient declining recording. For a Canadian deployment, have the clinic’s privacy lead check the applicable provincial or federal requirements, cross-border processing terms and patient explanation separately from the U.S. BAA.

What the evidence establishes

Product-specific customer experience

The August 2025 KLAS First Look interviewed seven individuals from seven organizations, drawn from a vendor-supplied list of 12 eligible organizations. Respondents described reduced documentation burden and valued support and customization; the report also recorded concerns about delivery timelines. This is a small customer-experience study, not a blinded note-accuracy assessment or randomized comparison. Some report content is explicitly vendor-provided. KLAS report, pages 2–3.

Use that finding to sharpen reference calls: speak with a similar specialty using the same integration, ask how unresolved template changes are handled, and compare promised with actual implementation dates. Match each reference’s product and service level to the offering you are evaluating.

Marketing metrics need their measurement method

Commure’s public pages describe the 90% metric differently. As checked October 8, 2026, the features page labels it “90% less time spent on documentation,” while the eClinicalWorks integration page pairs 90% with “Providers spend less time on documentation.” A reduction in time and a proportion of providers reporting a reduction are not equivalent findings.

Commure also reports 43-second chart closure and 91% reporting less fatigue on both pages. For these vendor-reported metrics, request the definition, product configuration, population, baseline, denominator and measurement period. Ask whether review, corrections, failed sessions and EHR transfer are included, and obtain clarification of the conflicting 90% descriptions.

Findings from a DAX Copilot and Nabla trial

A randomized study of 238 outpatient physicians across 14 specialties compared DAX Copilot, Nabla and usual care. Nabla reduced time-in-note by 9.5% relative to control; DAX’s difference was not statistically significant. Clinicians reported occasional inaccuracies. The study tested neither Commure nor Vero. Its relevance is methodological: adoption, time and quality need separate measurement, and benefits cannot simply be transferred between products. Lukac and colleagues, 2025 randomized trial.

A reproducible Commure evaluation pack

Use these fictional scripts for a repeatable sandbox evaluation. Arrange permission for testing and publication, then record the source, output and review effort for each attempt.

Record the date, offering, app version, template, enabled modules, device, microphone, OS, browser and EHR route. If human documentation support is included, identify its role and turnaround separately. Keep raw source recordings, untouched drafts and corrections together, and mark unavailable artifacts in the record.

Clinician: What happened to the outside report?
Patient: I requested it yesterday, but it has not arrived.
Companion: My report arrived this morning.
Clinician: Your companion’s report is not yours. No examination was performed today. Our records coordinator will request your report today. I will review it after receipt; no review has happened yet.

Clinician: The supplied historical measurement is 1.5 centimetres on 2 October 2026.
Patient: My discomfort was on the right—sorry, on the left.
Clinician: Left is the corrected side. Today is 8 October 2026. We did not repeat the measurement today.

Before stopping capture, say: “The report remains unavailable.”
Stop using the documented control. While capture is stopped, say: “OFF-RECORD MARKER: purple umbrella.”
Resume only if the configuration supports it. Say: “The records coordinator will request the report. Review remains pending.”
If resuming is unsupported, record that limitation and use the approved new-recording route.

Use the CM-01 source. Save a draft that retains the pending report and named request owner. Make a visible test edit: “Review pending receipt.” Transfer through the configured route to the designated sandbox encounter. Reopen the destination. Inspect the plan field, patient, encounter and signature state; then retry only using the approved recovery procedure.

STAFFED-SERVICE REHEARSAL — fictional source and sample timing.
Use an authorized Assist or Live evaluation with a documentation specialist and sandbox encounter. If no staffed service is available, mark this case not run.

SETUP
Name the specialist, supervising service lead, clinic coordinator, backup coordinator and approving clinician. Set a sample draft-delivery target of 30 working minutes from source handoff, pausing only while a logged clinician clarification is pending. Preserve total elapsed time as well. Clinical escalation follows the clinic’s separate protocol.

T+0 — SOURCE HANDOFF
Clinician: “The outside report has not arrived. Arrange follow-up on Friday. I have not reviewed the report.”
The packet deliberately omits who should act and the exact deadline. Keep the draft unsigned.

T+5 — CLARIFICATION
Prompt the specialist to send: “Who owns the request, what is the deadline, and does follow-up mean requesting the report or reviewing it?” Log receipt, recipient and pause start.

T+20 — SUPPLIED CLINICIAN RESPONSE
“Records coordinator A will request the report by noon on 9 October 2026. I will review it after receipt. Neither the request nor my review is complete.” Log the response and resume the delivery clock.

DELIVERY AND CORRECTION
Record the actual first-draft delivery time. If still unavailable at T+45, alert the service lead and clinic backup coordinator; keep the delivery status pending. On receipt, the clinician supplies a new correction: “Backup coordinator B now owns the request; keep the same deadline and pending status.” Ask the specialist to return a revised draft and record its version and return time.

FINAL HANDOFF
The clinician checks that coordinator B, the exact deadline and pending review survive in the saved sandbox note, then performs the authorized approval/authentication step. The specialist prepares and revises; the clinician owns final approval. Preserve the approved version against stale overwrites. The records request stays open until execution is separately recorded.

COMMURE WORKFLOW EVALUATION — blank record
Offering / plan / licensed unit:
Test date / tester / clinical reviewer:
App version / browser / OS / device / microphone:
Room / background audio / speaker positions / language:
Template / enabled modules / human service:
EHR edition / sandbox encounter / integration route:
Authorization and recording approval reference:

ONE ROW PER ATTEMPT (including failures)
Case / repeat / condition:
Source recording and expected-fact reference:
Raw transcript / raw draft reference (or unavailable):
Facts preserved / omitted / altered / unsupported:
Speaker, numeric and action-state defects:
Setup / active review / transfer / recovery minutes:
Generation wait / human-service turnaround (separate):
Staffed case: source handoff / clarification sent / reply received / pause reason:
Draft delivery / correction requested / revised draft returned / total elapsed time:
Agreed delivery target / counted service time / overdue alert and recipient:
Specialist / service lead / clinic backup / final-note approving clinician:
Correction made / corrected draft reference:
Saved record / signature state / duplicate check:
Screenshot references: control state, raw draft, corrected draft, destination:
Pass / hold / fail / not applicable with reason:
Remediation owner / retest date / approving reviewer:

DECISION
Scope accepted / exclusions / unresolved conditions:
Included attempts / failed attempts / unavailable artifacts:
Approver / date / triggers for recheck:
Keep evidence and findings in the clinic’s approved workspace.

Run and score the same configuration

Run the four capture-to-record cases, CM-01 through CM-04, twice in a quiet room and twice with the same documented background audio: 16 planned attempts per configuration. Retain the background file, playback setting and measured sound level when available; without measurement, report that limitation rather than naming an assumed decibel level. Use the same speakers and placement. Keep telehealth and language conditions separate instead of pooling them into one average.

For every attempt, count expected facts preserved, omitted or altered, and unsupported assertions added. Record active setup, correction, transfer and recovery time. Keep generation wait and human-service turnaround separate from active work. Report failed captures and transfers in the denominator, even when no note is produced. These four cases are boundary tests, not a representative estimate of clinical accuracy.

Capture dated screenshots of the recording state, untouched draft, correction and reopened destination using the synthetic source. Retain them alongside the attempt’s findings and timing record.

Evaluate the staffed handoff separately

CM-05 is an additional proposed case for Ambient Assist or Live, run separately from the 16-attempt audio comparison. Commure’s public overview describes specialist-refined drafts for Assist and dedicated visit support for Live. The rehearsal uses a sample delivery target; agree the actual commitment with the service team. Service descriptions checked October 8, 2026.

Run the case once with the assigned specialist and again using the agreed replacement or absence-coverage route. Record each attempt separately. Keep the original source, clarification exchange, first and revised drafts, clinician correction, destination version and approval log together. Report total elapsed turnaround, any excluded clarification wait, and active clinician correction time separately. A fast first draft is not a completed handoff when clarification or final review remains open. If the service or required audit trail cannot be evaluated, record that limitation rather than a pass.

Evaluation example: a convincing draft with an incomplete handoff

Fictional CM-01/CM-04 outcome: the draft correctly says the patient’s report has not arrived and clinician review is pending. It omits the records coordinator’s request task. The reviewer restores that sentence. After transfer, the reopened note again lacks the sentence, although the interface reports success.

The correct finding has two parts: draft omission and saved-record omission. The draft correction did not resolve the receiving-record failure. Mark the handoff hold, retain the source and both versions, assign investigation to the integration owner, and retest with the latest approved text. Do not record a pass because the clinician could manually repair the final note; repair and reliable transfer are different acceptance questions.

An example row with placeholder evidence references would read: “CM-04; configuration TEST-A; source CM-01; finding: request owner absent after sync; evidence: synthetic source, corrected draft and destination snapshots EX-A/B/C; decision: hold; owner: integration lead; next action: verify plan-field mapping and rerun; approver: pending.”

Strengths, tradeoffs and who should shortlist it

Reasons to shortlist: the documented choice of automated and staffed workflows, concrete EHR delivery routes, and a public self-service entry point make Commure relevant to several different documentation problems. The product-specific customer interviews add useful implementation context.

Tradeoffs to evaluate: enterprise cost and operational commitments require a quote, while configuration and retention depend on the contracted service. Include specialty-specific review effort in the trial: extensive correction or manual relocation can outweigh a low subscription price.

For a solo clinician, prioritize usable output, review effort and exit access. For an integrated practice, prioritize identity matching, field mapping and failure recovery. For a service buying human support, prioritize coverage, quality-control responsibilities and how clarification reaches the clinician. For all three, define the unacceptable failure before the demonstration begins.

Where Vero fits in the comparison

Vero’s current note-creation instructions document recording, typed context and uploaded files, followed by template selection, editing and export through copy, PDF, Word, print or email. That is a useful comparison when the requirement is a flexible documentation workspace. Evaluate native EHR writeback and staffed scribing as separate purchasing requirements. Source checked October 8, 2026.

Use the same source case and acceptance criteria for either product: what material goes in, what the clinician must correct, how the approved version reaches its destination and what work remains. Select the least burdensome complete workflow that meets the clinic’s quality, privacy and delivery requirements.