Published 2026-11-11. Sources opened 2026-11-10. Every claim sourced; see the methodology.
One short, sourced email a week: what changed in AI scribe pricing, privacy terms, and features.
The Big Finding
Two rankings, two answers, and both are honest. Scribechoice, an independent comparison site that ranks therapy scribes from vendor documentation and re-checks its facts monthly, last verified October 2026, puts Mentalyc at the top of its ranking at 9.2 and Upheal next at 8.9 [1]. Our clinical pilot, which scored the notes the two tools actually produced from three standardized synthetic psychiatric encounters, flagged both vendors "Unsafe as written" on two of three encounters [2][4]. The contradiction is not an error in either ranking. It is a lesson in asking what a ranking measures before you buy on the strength of it.
What the pilot found is specific. Mentalyc scored 79.0 overall with two critical flags: a follow-up note documented Lexapro as citalopram and continued it as citalopram, and a suicide-risk note diagnosed major depressive disorder with psychotic features while its own mental status examination denied hallucinations [2][4]. The same notes also labeled a wife's sleeping pills with the invented drug name "Collided" when the audio named zolpidem, or Ambien, and invented tearful in-session examination language for the PTSD visit [2]. Upheal scored 63.0 overall with three critical flags, the most in the six-vendor cohort: a follow-up note never documented vitamin D or occasional ibuprofen, both predefined critical items for that encounter, and a suicide-risk note grouped zolpidem with the patient's own bupropion without attributing the zolpidem to his wife [2][4]. Fifty-four checklist items were non-full across Upheal's three notes, which is why its overall score sits well below the rest of the cohort even apart from the critical flags [2].
The documentation picture points the other way, and it is real too. On our research scorecard, Mentalyc is 75/100, ranked fifth among vendors with a documented BAA, with a perfect 20/20 on privacy and security and the lowest published paid entry price in its group, at $19.99 per month for Mini or $14.99 billed annually [3]. Upheal is 71/100, ranked sixth among vendors with a documented BAA, with a permanent free plan and AI Notes at $39 per month [3]. A free plan and a strong feature list are real advantages. They do not answer whether a given draft note is safe to sign.
Caveats, stated plainly: the pilot is a pilot. It ran three synthetic encounters, one scored run per vendor per scenario, with text-to-speech audio, and the evaluations were adjudicated by a single psychiatrist scorer, with no second independent psychiatrist review yet [4]. Vendor settings differed: Mentalyc was tested as an intake note in longer form online, and Upheal on default settings [4]. The rest of the cohort shows the two kinds of evidence can agree: Heidi scored 96.0 with zero critical flags, Freed 91.0, JotPsych 90.2, and Blueprint 83.2, all with zero critical flags [4]. When documentation strength and note accuracy point the same way, a buying decision is easy. Mentalyc and Upheal are the cases where they point in different directions, and those are the cases where the choice of evidence matters most.
How to use two rankings that disagree: use a documentation ranking to build the shortlist, and use note-accuracy evidence to test the shortlist. Run your own encounters through each finalist, and read the drafts the way the pilot reads them: medication names and ownership, diagnosis against the documented examination, and anything the note asserts that the session did not support [2]. Check the failure mode, not just the score. A wrong SSRI identity and an omitted interacting medication are different risks from a missing scale score, and they call for different safeguards in your review routine. And price the review step, not just the subscription: if a vendor's drafts need heavier correction, that correction time is part of its cost. If you are weighing a standalone scribe against an EHR-bundled note tool, our psychiatric EHR comparison prices the bundled alternative across 13 EHRs [2]. Read the full post: clinicalaiweekly.com/blog/two-rankings-two-answers
Product Watch
- Same-day closure gains did not extend to interpreted sessions. A cross-sectional study published in JAMA Network Open on November 6, 2026, evaluated 209,009 in-person encounters among 117,121 patients seen by 286 clinicians across 14 adult primary care clinics at a Boston academic medical center, from May 2025 to May 2026 [5]. In adjusted analyses, AI scribe use was associated with higher same-day note closure for language-concordant encounters without an interpreter: up 10.2 percentage points for non-English language preference encounters and up 1.7 percentage points for English language-concordant encounters. For encounters with an interpreter, and for nonconcordant encounters without one, the findings were inconclusive given wide confidence intervals [5]. Caveats: retrospective and single-system, with two scribe platforms in use and clinicians self-enrolling for access; only one of the two tools supported non-English, non-Spanish documentation without an interpreter [5]. For therapy practices that work through interpreters, the closure benefit the vendors advertise may not arrive with the interpreter in the room.
- Sunoh.ai reports a clinic case study, with no independent numbers. A November 5 press release says Conway OB-GYN Clinic in Arkansas uses Sunoh.ai to generate notes, accelerate chart completion, and save hours of documentation time daily, and notes multilingual capabilities for encounters involving interpreters [6]. Caveats: this is a vendor-reported case study with no measured figures and no comparison group. Sunoh.ai is 35/100, ranked 17th on our research scorecard [3].
- Cerbo takes its AI scribe to virtual visits. Cerbo announced November 3 that its AI Scribe, launched in August as a native tool inside the Cerbo EHR, now drafts the structured note inside the chart for Cerbo Telemedicine and Zoom visits, included at no additional cost for AI Scribe users with those integrations [7]. Cerbo is an EHR for cash-pay practices and is not currently tracked on our scorecard.
- No new scorecard vendor this week. The scorecard still covers 19 tracked vendors, 17 ranked, with scores as of 2026-10-23 [3]. One pricing change from our November 4 audit: Twofold Health's $21 per month annual promotion has ended, and the scorecard now lists Personal at $49 per month billed annually or $69 per month monthly [3]. Note the November 6 flag: our recheck could not re-verify the $49 figure because Twofold's pricing page now redirects to its homepage, so treat that figure as listed, not freshly confirmed [14]. Scores and ranks are unchanged.
Clinical Reality Check
Should residents use AI scribes, or should they write their own notes first? A New England Journal of Medicine Clinical Decisions feature published November 4, 2026, puts that question to two essays: a vignette accompanied by one essay recommending implementing an AI scribe for residents' clinical notes and one recommending that it not be implemented [8]. The case against rests on what note-writing teaches. Writing a note requires a young clinician to interpret symptoms, distinguish important information from irrelevant details, consider possible diagnoses, and develop a treatment strategy, and routine reliance on a drafting tool could erode exactly that training [8]. There is also the accuracy problem and automation bias, the tendency to trust computer-generated text because it looks authoritative, when a polished note can omit important information, misattribute statements, or introduce details that were never discussed [8]. The case for is time: less typing means more meaningful patient conversations and more time for education [8].
A commentary on the feature argues for responsible integration rather than either adoption or refusal: medical students and residents should first develop fundamental skills in interviewing, clinical reasoning, and documentation, and AI should then be introduced as a supervised assistant, with every AI-generated note reviewed, corrected, and approved by a qualified physician, and residents continuing to write notes independently so their skills do not deteriorate [9]. For psychiatry training, the same reasoning applies to the mental status examination. A resident who never writes the examination from observation may never learn to see it, and the examination is the part of the note the microphone never hears. The feature does not settle the question. It frames the supervision decision programs have to make now, before the habit forms.
States are writing the notice rules while HIPAA stays silent. A November 4 report on state AI-scribe rules notes that Louisiana requires a spoken disclosure before recording, Rhode Island requires patient notice and a review of the note, and California requires a disclaimer on AI messages no clinician has read [10]. HIPAA has no specific patient-notice rule for AI scribes [10]. Nearly half of 3,151 physicians surveyed by Doximity (47%) said the AI decision-making process at their institution is "still evolving" [10]. The scale behind the rulemaking: Abridge supports upward of 100 million clinical conversations a year, according to a June Fortune report cited by PYMNTS, and Cleveland Clinic used Ambience Healthcare's scribe in more than 3.5 million patient encounters in its first year, according to a paper its clinical leaders published in npj Health Systems, as cited by PYMNTS [10]. For a behavioral health practice, the practical point is that consent and notice are state-by-state. Check your state's rule before you record, and document the notice the way you document consent.
Head-to-Head
Two real-world studies disagree on after-hours work, and the difference is in what they measured. The Beth Israel Lahey study published in Annals of Internal Medicine on October 26, 2026, followed 902 outpatient clinicians after they enrolled in Heidi's ambient scribe. Median note-active time fell from 21 to 18.6 minutes per encounter, while median note length rose from 4,854 to 5,187 characters, but the study found no significant change in same-day note closure, after-hours note closure, or work relative value units [12]. A longitudinal quasi-experimental study from Washington University School of Medicine, posted as a preprint, followed 220 primary care clinicians across 36 clinics and 314,845 encounters, and found a different pattern at 150 days after scribe initiation: note writing time down 15%, pajama time (5:30 PM to 7 AM) down 18%, time outside scheduled hours down 13%, and total work relative value units up 2% [11]. But its early gains faded: note length was up 5% and note closures within 24 hours were up 31% at day zero, with no evidence for changes from the pre-period in either by day 150 [11].
Read the two together, with their limits attached. The Beth Israel Lahey study is peer-reviewed but used an interrupted time series around each clinician's enrollment, so it shows an association, not proof that Heidi alone caused the changes; time spent editing a draft inside Heidi before it reached the EHR was not captured, and note accuracy was not assessed [12]. The Washington University study is a preprint, not peer-reviewed, and one author (Kannampallil) is an unpaid advisor on Abridge's research steering committee [11]. The two studies also differ in vendor (Heidi at Beth Israel Lahey; Abridge, named in the WashU methods, inside Epic), setting, and follow-up window. The after-hours question is not settled. The way to settle it for your practice is the same either way: log your own after-hours EHR time for the weeks before a trial and during it, and report adoption per clinician, not per practice.
One Action This Week
Run the five-check pre-signing routine before every signature. Our November 9 post builds it from the exact errors our clinical pilot caught [13]: check 1, medication names and identity, including dose and start, stop, or continue instructions; check 2, medication ownership, whose drug each line belongs to; check 3, diagnosis against the examination, read together to see whether they agree; check 4, only findings observed in the session, anything the note claims you saw that surprises you probably did not happen; check 5, what the note left out, by checking the plan against the decisions and counseling you remember from the session [13]. Sign each note per encounter, close to the session, not in an end-of-week batch, when recall of any session is weakest [13]. And before you record the next session, check your state's notice rule, as this issue's Clinical Reality Check describes [10].
Also new on the site
- New blog post, November 9: "The Pre-Signing Routine: Five Checks Before You Sign an AI-Scribed Note," the full version of this issue's One Action, including the five pilot errors it is built on [13]. Its opening scale figure: from October 2025 to October 2026, Centerstone generated about 510,000 notes with ambient documentation and saved roughly 85,000 hours of clinician time, according to data the organization shared with Becker's Behavioral Health [13].
- New blog post, November 5: "Two Rankings, Two Answers: What Scribechoice's Top Two and Our Clinical Pilot Disagree About," the full version of this issue's Big Finding [2].
- Corrections, November 4, stated plainly: vendor fact fixes from the full research audit of all 19 scored vendors. Twofold Health's $21 per month annual promo has ended (now $49 per month billed annually); PMHScribe no longer publishes annual prices; Sunoh.ai's EHR Sync browser extension added to its EHR line; Suki's directory trial label is now Ask vendor; Nabla and Microsoft Dragon Copilot source links updated; Doximity Scribe EHR wording clarified; AutoNotes FALL39 promo, JotPsych JotAudit price discrepancy, and Blueprint claim fee wording noted [14]. Scores and ranks unchanged.
- Corrections, November 6, stated plainly: weekly vendor-fact recheck of live official pages, plus wording and mobile fixes. Berries scorecard pricing now leads with $79 per month billed annually or $99 per month month to month; Tali Pro updated to US$75 per month billed annually (was listed as US$100 per month) [14]. Flagged without changing scores or listed prices: Twofold's pricing page redirects to the homepage and the $49 per month annual figure was not visible; Dragon Copilot's $0.25 Flex figure was not visible on the licensing page; Abridge Trust Center returned 403; Nabla's EHR page still documents bidirectional chart access (still scored 0/5) [14]. Scores and ranks unchanged.
- Scorecard unchanged: 19 tracked vendors, 17 ranked, scores as of 2026-10-23 [3]. Research scores rate public vendor documentation, not note accuracy or clinical outcomes.
The bottom line: Two rankings can both be honest and still disagree, the after-hours evidence is mixed, and the states are writing notice rules faster than HIPAA is, so shortlist on documentation, decide on the drafts, run the five checks before you sign, and check your state's rule before you record.
Sources
- Scribechoice, "AI therapy scribes, ranked" (Mentalyc 9.2, Upheal 8.9; last verified October 2026; facts re-checked monthly; rankings editorial and independent), https://www.scribechoice.com/ (opened 2026-11-10)
- ClinicalAIWeekly, "Two Rankings, Two Answers: Scribechoice vs Our Pilot" (blog, published 2026-11-05), https://www.clinicalaiweekly.com/blog/two-rankings-two-answers (opened 2026-11-10)
- ClinicalAIWeekly, "AI scribe research scorecard" (19 tracked, 17 ranked; scores as of 2026-10-23), https://www.clinicalaiweekly.com/scorecard (opened 2026-11-10)
- ClinicalAIWeekly, "Psychiatric Note Accuracy Pilot: Six Vendors Scored" (Heidi 96.0, Freed 91.0, JotPsych 90.2, Blueprint 83.2, Mentalyc 79.0 with 2 critical flags, Upheal 63.0 with 3 critical flags; single psychiatrist scorer, no second independent review), https://www.clinicalaiweekly.com/clinical-benchmarks/results (opened 2026-11-10)
- JAMA Network Open, "Artificial Intelligence Scribes and Same-Day Note Closure by Patient Language Preference and Interpreter Modality" (published November 6, 2026), https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2856327 (opened 2026-11-10)
- Sunoh.ai, "Sunoh.ai Assists Conway OB-GYN Reduce Documentation Burden, Improve Patient Engagement, and Reclaim Hours Each Day" (press release, November 5, 2026), https://sunoh.ai/press-releases/sunoh-ai-assists-conway-ob-gyn-reduce-documentation-burden-improve-patient-engagement-and-reclaim-hours-each-day/ (opened 2026-11-10)
- Cerbo, "Cerbo Expands AI Scribe to Virtual Visits, Bringing Native AI Charting to Telehealth" (press release, November 3, 2026), https://www.wcia.com/business/press-releases/cision/20261103CL88024/cerbo-expands-ai-scribe-to-virtual-visits-bringing-native-ai-charting-to-telehealth/ (opened 2026-11-10)
- Campos A, You JG, Iverson C, "Use of an Artificial Intelligence Scribe for Clinical Notes," New England Journal of Medicine Clinical Decisions (published November 4, 2026), N Engl J Med 2026;395:1864-1866, https://www.nejm.org/doi/full/10.1056/NEJMclde2600746 (opened 2026-11-10)
- BiomedGlobal, "When Artificial Intelligence Starts Writing the Doctor's Notes: Progress or a Threat to Medical Learning?" (commentary on the NEJM Clinical Decisions feature), https://www.biomedglobal.org/en/blog/when-artificial-intelligence-starts-writing-the-doctors-notes-progress-or-a-threat-to-medical-learn (opened 2026-11-10)
- PYMNTS, "As AI Listens In on Doctor Visits, States Are Writing the Rules" (November 4, 2026), https://www.pymnts.com/healthcare/2026/as-ai-listens-in-on-doctor-visits-states-are-writing-the-rules/ (opened 2026-11-10)
- Waken RJ, et al., "Longitudinal effects ambient AI scribe use on documentation burden and financial productivity: A quasi-experimental study" (medRxiv preprint), https://www.medrxiv.org/content/10.64898/2026.01.12.26343538v2 (opened 2026-11-10)
- Booth SE, et al., "Impact of Unrestricted Artificial Intelligence Scribe Deployment in a Single Health System on Clinician Workload and Productivity: A Quality Improvement Study," Annals of Internal Medicine (published online October 26, 2026), https://doi.org/10.7326/ANNALS-25-05294 (opened 2026-11-03)
- ClinicalAIWeekly, "The Pre-Signing Routine: Five Checks Before You Sign an AI-Scribed Note" (blog, published 2026-11-09), https://www.clinicalaiweekly.com/blog/ai-scribe-pre-signing-routine (opened 2026-11-10)
- ClinicalAIWeekly, "Corrections policy and changelog" (November 4 and November 6 entries), (opened 2026-11-10)
- ClinicalAIWeekly, "Psychiatric EHR comparison 2026" (13 EHRs), https://www.clinicalaiweekly.com/practice-stack/ehr-comparison/ (opened 2026-11-10)