White paper · September 2026
Why is something called “ambient” actually so noisy?
Ambient AI scribes have failed to fix the endemic workflow problems plaguing private practice urology.
A Viscera white paper for private practice urologists
September 2026 · viscera.com
EXECUTIVE SUMMARY
AI scribes for medical practices are designed generically and described in near-identical language. They compete on irrelevant counts — specialties supported, languages transcribed — rather than on the number that matters to doctors in private practice:
how long it takes to get the right data from the patient into the EHR.
The largest controlled study to date measured about 16 minutes saved per clinic day and no change in after-hours EHR time for clinicians at 5 academic health systems who adopted AI scribes. Two-thirds of AI scribe adopters used the tool in fewer than half their visits.
Urology is among the 3 specialties with the lowest adoption in one of the largest reported deployments, and 38.1% of urologists have access to an AI scribe while 13.3% use one regularly. This gap signals a significant product-problem mismatch.
The EHR is a system of record, not a system of process. The urologist is forced to be the data-entry translator between the AI scribe and the EHR.
The current financial environment for private practices leaves little room for tools that do not deliver significant value to a practice. Urology faces a 0% net impact on total Medicare allowed charges for CY2026, a proposed 2% decrease for CY2027, and 2.7% projected cost growth.
Contents
I
AI scribes are generic by design
“Focus on people, not paperwork.” 1
“Practice medicine, not paperwork.” 2
“Less process. More practice.” 3
“You focus on your patients, Scribe will handle the paper work.” 4
Many of the available AI scribe products present themselves in nearly identical terms and show up in the marketplace with nearly identical proof points: adoption rates, satisfaction scores, KLAS badges. Differentiation occurs by way of questionable feature bloat. One company advertises “200+ specialties supported,”2 even though the American Board of Medical Specialties certifies physicians in 38 specialties and 90 subspecialties (128 specialty and subspecialty certificates in all).5 Another advertises transcription in 110+ languages.6
This is the sound of noise. But it’s not a bug in the system. The conversation around how to incorporate AI into private practice is noisy by design. The AI scribes bring chainsaws into the laparoscopy OR. Claims are optimized for spec sheet comparisons across every specialty at once, which means the underlying products are optimized for nobody in particular. Generic AI that solves a large quantity of problems might be useful for generic varieties of care (think: primary care practice), but specialty practices demand specialty tools solving specific problems.
Just because AI is good at something
doesn’t mean you need it.
Is lack of multilingual support across 110 languages really what keeps a team in the office after 5:15 pm? How many clinics need post-procedure instructions generated in Kazakh and Aleut?
What the AI scribe companies will not discuss is the one number that’s actually meaningful for improving the state of private practice: how long providers and their teams spend getting the right data from the patient, into the EHR, and moving it along the path to better care.
specialties
languages
Figure 1 · The noisy sameness of the category
Five vendors, anonymized, as they describe themselves. At the time of writing, no vendor in the review published an estimate of how long they think a physician should spend moving patient data from the AI-generated note into the EHR.
Source: vendor websites accessed August–September 2026; pricing and product claims reviewed across 8 ambient documentation vendors.7 Headlines as published.1–4,6
II
The failure of ambient scribes to improve private practice urology: what the evidence shows
In April 2026, JAMA published the largest controlled study of ambient AI scribes to date: 1,809 adopters against 6,772 controls, 8,581 clinicians across 5 academic health systems, tracked from June 2023 to August 2025.8 Here’s what they found for clinicians using ambient AI scribes.
16
minutes of documentation time saved per 8 hours of scheduled patient time
That is per clinic day. The figure is sometimes misreported as per encounter.
Before
After
No change
in after-hours EHR time
The hours that cost physicians their evenings did not move.
0.49
additional visits per week
One additional patient every 2 weeks, against roughly 58 seen in that time.
$167
marginal E/M revenue per clinician per month
Gross, across all specialties, at academic systems.
32%
of adopters used the AI scribe in half or more of their visits
Over two-thirds used it in less than half of their visits.
Used in visits
Did not use in visits
Figure 2 · What 2 years of measurement produced
Source: Rotenstein LS, Holmgren AJ, Thombley R, et al. JAMA. Published online April 1, 2026.8 Figures as reported; per clinic day unless stated.
A 2025 NEJM AI randomized trial of 66 clinicians at UW Health found a reduction in burnout alongside a documented savings of about 22 minutes of documentation time per day.9 The Peterson Health Technology Institute, drawing on 8 health systems in March 2025, concluded that the financial impact of ambient scribes “is unclear,” and that throughput gains, which most systems did not observe, remained anticipated rather than documented.10
Kaiser Permanente ran one of the largest reported deployments: 7,260 physicians across specialties, accounting for more than 2.5 million uses over roughly 15 months.11 Adoption was highest in mental health, emergency medicine, primary care, allergy, and cardiology. It was lowest in infectious disease, obstetrics and gynecology, and urology.12
The AUA’s own 2025 Census makes the same point from the other direction.13 Of US urologists, 38.1% have access to an AI scribe. Only 13.3% use one regularly, a 25-percentage-point gap between having access to the tool and using it regularly. Low adoption rates and low usage of a tool that has been brought into the clinic do not suggest a failure on the part of urologists. They speak to a product-problem design mismatch. Generic AI scribes are not designed to support the workflow that actually consumes the bulk of urologists’ administrative time.
One peer-reviewed study put these tools in front of urologists specifically. In Urology in February 2025, Moryousef et al. had 20 Canadian urology faculty and trainees evaluate 5 freely accessible AI scribes against reference notes for simulated stone, BPH, and PSA encounters.14 Three-quarters of respondents called documentation a significant contributor to burnout. Respondents were receptive (90% were open to using AI scribes), but even the top-rated scribe had a 28% critical-error composite score.14
Generic AI scribes are not designed to support the workflow that actually consumes the bulk of urologists’ administrative time.
The urology community wants this problem solved. The evidence suggests that it has not yet been solved by ambient AI scribes.
Access versus use · US urologists, 2025
Adoption by specialty · Kaiser Permanente deployment
Highest adoption
Lowest adoption
Figure 3 · The adoption gap
The low adoption rate among urologists speaks to the design mismatch between the tool and the problem it is attempting to solve. Generic AI scribes are not designed for the real problems specialty private practices face.
Sources: AUA 2025 Census13 (access and regular use); Tierney AA, Gayre G, Hoberman B, et al. NEJM Catalyst, April 202511 and Becker’s Hospital Review, June 202512 (specialty adoption).
III
The EHR was built as a system of record, not a system of process
Regardless of which EHR a practice uses, the EHR remains a system of record, not a system of process. The EHR is essentially a database with a minimally designed interface bolted on top of it. Its job is to hold discrete, retrievable, legally durable facts: diagnosis, order, result, signature, and timestamp. It was built to be queried and audited. Its use is mandated, but it was never intended to be the mechanism that moves a patient through a longitudinal course of care.
A urology visit is a process. It has prerequisites, sequence, thresholds, branch points, and downstream obligations. The note is a byproduct of that process. The process is what gets the patient treated and the practice paid.
Between the process and the record sits a translation layer. Right now, that translation layer is the individual urologist. Every copy-paste and click is a piece of that larger process translation.
117
clicks in the EHR during a single office visit, counted from physicians’ own audit logs across the practices taken live by Viscera.
27
variants of the same billing code surfaced in one practice at integration, accumulated through years of staff changes and office workarounds.15
Adding an ambient scribe into the mix adds a new layer of captured English prose. Sometimes the prose is quite good and reasonably accurate. But the EHR cannot make use of prose. The EHR contains discrete fields. When the prose is delivered from the scribe, the urologist still has to read it, correct it, decide the diagnosis specificity, select the level, place the orders, and move the actionable pieces of information into their EHR boxes. The scribe automated the typing. The translation process and the workflow itself remain manual and cumbersome.
The conversation
Captured, ambiently
AI SCRIBE SPEEDS THIS UP
A bucket of words
Delivered as unstructured prose
AI SCRIBE’S WORK ENDS HERE
The EHR
Structured data entered manually
Figure 4 · The bucket of words
AI scribes stop being useful at the point of the EHR, when manual data entry, often in the form of copy-pasting, takes over.
This is why peer-reviewed studies have found such minimal changes in providers’ administrative workload after the introduction of ambient scribes: typing visit notes was never the most time-consuming part of the process.
Scribe vendors concede this in the fine print. One vendor describes its output as “draft documentation.”3 One product diagram has 3 steps. The middle step for the provider is “Edit.”4 We could not find a single vendor that publishes how long they think the “Edit” step should take for a provider.
Steps in the encounter
AMBIENT AI SCRIBE
Referral arrives
Chart prepped
Patient seen
Decision made
Orders placed
Level selected
Claim built
Claim adjudicated
Translation and data entry
falls on the provider
Referral doc
Problem list
Encounter note
Diagnosis code
Orders table
E/M level
Claim line
Remittance
THE RECORD OF STRUCTURED DATA ACCEPTED BY THE EHR
Figure 5 · The patient journey vs. the EHR
The patient encounter breaks out into many discrete steps for the practice. These steps do not align with the structured data accepted by the EHR database. The provider is forced to be the mechanism of translation between the steps of the encounter and the EHR. This translation process happens for every patient encounter, every day.
Where does your practice sit on the EHR maturity curve?
Eight questions, 90 seconds, 4 stages from captive to autonomous.
IV
A knowing-urology problem
General-purpose AI tools have become reasonably competent at summarizing conversation. But summarizing conversation is not the constraint holding back private practice urology. The constraint is the shape of the conversation itself. An exam room exchange is a structured clinical argument with supporting details. The details that carry the weight are precisely where the generalist AI model breaks down.
A provider might say “hematuria” in the exam room, but is it gross or microscopic? This distinction carries different codes, different workup, and different defensibility.
A provider might say “aquablation.” But were the other options that were offered first documented and attached to that word? Assuming the AI captured “aquablation” and not “aggravation,” does it have any way of knowing if the documented prostate size clears the threshold that a particular payer is applying this quarter? An administrator put it plainly at the July 2026 Viscera Health Private Practice Urology Summit: “When a national payer changes a prostate size requirement, every practice should learn it at once, so providers aren’t all making the same mistakes over and over.”15
A patient’s PSA may appear in 4 notes from the same year. Which one matters today? When did ADT start? When did the PARP inhibitor start? Those questions require a trajectory, which will not be produced by a summary of 36 months of notes. The summarized reason for a visit might say “elevated PSA.” The patient has advanced prostate cancer. Or it says “trouble urinating,” but the patient came in for Cialis. No amount of language modeling repairs a trajectory that was muddled before the visit started.
Urologic depth is not a template.
AI scribe vendors provide specialty templates but then invite providers to build their own: “Choose from existing templates or build your own from a sample note. Create as many as you need.”1 At least one vendor’s library is explicitly a community library, populated by the clinicians who use the product.6
Stated plainly: Some of the specialty depth being advertised is authored by customers, during their own evenings, and offered back to the next customer as a feature. “You’re in control” is how configuration work gets marketed.16
A urologist should not have to teach software they are paying for what a cystoscopy is. Nor should they need to be sufficiently expert in billing and compliance regulations to build billing-compliant templates.
V
The invisible gates: leveling and coding
Since 2021, an office visit is leveled on medical decision making or on total time, at the practitioner’s election.17 A medically appropriate history and exam are still performed but no longer determine the level. MDM has 3 elements, and 2 of the 3 must be met or exceeded:
The number and complexity of problems addressed at the encounter
The amount and/or complexity of data to be reviewed and analyzed
The risk of complications and/or morbidity or mortality of patient management
When a visit is leveled on MDM, the difference between a 99213 and a 99214 comes down to low versus moderate in the table. One stable chronic illness versus 1 chronic illness with exacerbation, progression, or side effects of treatment, or 2 stable chronic illnesses, or 1 undiagnosed new problem with uncertain prognosis. Two data elements versus 3, or, for moderate, a single independent interpretation of a test performed by a different provider, or a documented discussion with an external clinician. Low risk versus moderate risk, where the canonical example of moderate is prescription drug management.
For urology, that last qualification is often the crux of it all. Starting, adjusting, stopping, or continuing an alpha blocker, a 5-ARI, an anticholinergic, a beta-3 agonist, or an antibiotic for recurrent UTI can satisfy the risk column, provided the record shows the provider evaluated the condition the drug is managing. Whether that appears as a documented decision, with the reasoning attached, is frequently the difference between 2 codes.
THE 3 GATES
Medical decision making is scored on 3 elements. Two of the 3 must be met or exceeded at a given level for the visit to qualify there.
Problems addressed
Low
2 or more self-limited or minor problems; 1 stable chronic illness; or 1 acute, uncomplicated illness or injury
Moderate
1 or more chronic illnesses with exacerbation, progression, or side effects of treatment; or 2 or more stable chronic illnesses
Data reviewed
Low
Any 2: review of prior external notes, review of each unique test result, ordering of each unique test; or assessment requiring an independent historian
Moderate
Any 1 of: 3 data elements from the Low list; independent interpretation of a test performed by another physician (not separately reported); or discussion of management with an external physician
Risk of management
Low
Low risk of morbidity from additional diagnostic testing or treatment
Moderate
Prescription drug management, or a decision about minor surgery with identified patient or procedure risk factors
The same visit, documented 2 ways
Each element is scored at the level the documentation itself supports. The visit lands where 2 of the 3 land.
BPH follow-up, written as narrative
Urinary symptoms discussed. Alpha blocker adjusted. Recent PSA noted in passing.
The same visit, documented against the elements
Chronic BPH with progression of symptoms. PCP’s note reviewed; PSA and creatinine results from the PCP’s lab reviewed. Tamsulosin dose increased — prescription drug management.
If nothing is attached to an element, the problems read as stable and the data review is not demonstrable in the record.
Figure 6 · The leveling gate
A and B describe the same medicine. Only one of them is defensible.
Source: AMA CPT E/M office or other outpatient (99202–99215) code and guideline changes, effective January 1, 2021.17
The rules look objective on a table, but the application of the rules is a judgment call made under time pressure, dozens of times a day, by people who are trained to treat disease and not to score tables. It is a consequence of system design. And leveling has always been contestable: a classic OIG review found that 26% of 2010 Medicare Part B E/M claims were upcoded and 15% downcoded.18 Errors run in both directions. Vendors selling “more codes captured” tend to mention only one direction of improvement.
Meanwhile, the other side of the room has stopped reading. As one urologist described payer review at the same Summit: “They are not reading your documentation, they are checking boxes, and they are more likely to deny than not.”15
Today, payers are pushing reviews through algorithmically. The AMA has documented “an increasing number of payers” downcoding claims “automatically using software algorithms, without first requesting and reviewing clinical records,” and some paying at the reduced rate without changing the billed code.19 One major national payer implemented automatic 1-level downcoding of level 4 and level 5 E/M claims effective October 1, 2025. Organized medicine won a partial, temporary pause in California, and in March 2026 Maryland regulators fined the insurer and ordered it to stop automatic downcoding in that state. The policy has not been withdrawn nationally.20
There is a version of this story where AI documentation actually makes the situation worse for private practice urologists, not better. In PwC’s June 2026 survey of actuaries at 27 health plans, 70% ranked AI-enabled documentation and coding tools among their top 3 cost inflators.21 The Blue Cross Blue Shield Association published research in March 2026 estimating that about 20% of a 9% per-member rise in inpatient costs from 2023 to 2024 was attributable to rising coding intensity.22 A tool that drives level distribution upward without a defensible reasoning trail underneath is actually selling audit exposure on a monthly subscription.
The durable position is narrower: capture the work that was actually performed and under-documented and attach the reasoning to it, so the level holds under audit scrutiny.
The level is supported by the MDM elements, not by the length of the note.
VI
The 5-minute patient visit extends over weeks
Ambient capture addresses the administrative burden of 5 minutes during the day. It begins when the recording starts and ends when the recording stops. But a great deal of what exhausts a urology clinic happens on either side of that brief window.
Every urologist we talk to describes some version of the same morning: 38 lab results in the inbox, 12 belonging to patients scheduled that day, each requiring its own click to discover that fact. Prior notes provided as grainy PDFs and unopened referral packets. Labs drawn at the primary care office under the PCP’s name, entirely legally, with results routed to the PCP, not the urologist, and the patient who just arrived saying, “I did the tests.” We know of some practices that employ a person whose entire job is chart prep because if the missing item in a chart is a lab, they actually needed to know 2 weeks earlier.
There is the after-hours problem as well: the 5:15 window when the answering service calls because an antibiotic never got phoned in, the controlled substance prescription rewritten because the patient did not reach the pharmacy in time, the refill request the pharmacy generated automatically for a drug the patient may not be taking.
The real workflow of urology unfolds amidst the messy reality of many individuals’ lives. While solving for the burden of a very specific 5 minutes within that messy reality is not nothing, it’s not the solution to the whole problem either.
Before the visit
CLINIC HOURS
After the visit
7 am
9 am
11 am
1 pm
3 pm
5 pm
38
results in the inbox
only 12 belong to patients scheduled today
Grainy referral PDFs, unopened packets
Labs routed to the PCP, not the urologist
Needed an order placed 2 weeks ago
A staff member whose entire job is chart prep
Will only ever call into the 5:15 answering service
Antibiotic was never phoned in
Needs controlled substance scripts rewritten
Auto-generated refill requests
Unsigned notes
Denials and rework
1 PROVIDER, 1 CLINIC DAY · 15 PATIENTS
Work with no visit on the schedule to attach it to
Figure 7 · Patient visits occur linearly. The work surrounding each visit does not.
An average day in a private practice urology clinic is full of tasks left unsupported by AI scribes. Many of these tasks are time-bound to visits from several weeks ago, or to visits that are coming soon.
Source: Viscera Health internal practice deployment observations and remarks recorded at the Viscera Health Private Practice Urology Summit, July 2026.15
VII
The math squeezing private practice urologists
Neither the administrative environment nor the financial environment for private practice urology is stable.
For CY2026, urology is projected to see a 0% net impact on total Medicare allowed charges.23,24 The conversion factor rose 3.26% outside qualifying APMs, and CMS applied a 2.5% efficiency adjustment to work RVUs across nearly everything urologists do procedurally. Cystoscopy fell from 1.53 to 1.49 wRVU, and TURP, which was also revalued, fell from 13.16 to 9.75. Ureteroscopy with lithotripsy fell from 8.00 to 7.80. Laparoscopic radical prostatectomy fell from 22.46 to 21.90.23,25
For CY2027, CMS projects a 2% decrease in total allowed charges for urology.26 Two proposals aim squarely at the office-based visit: paying second and subsequent same-day services at 50% when a separately identifiable office E/M accompanies a procedure in a 0-, 10-, or 90-day global, and deleting G2211 in favor of a modifier reimbursed at 16% of the E/M value.27 For a specialty where office cystoscopy, bladder instillations, and other 0-day procedures routinely share a visit with the evaluation, that is not a rounding error.
Against that, CMS projects practice cost growth of 2.7% for 2026.28 In MGMA polling in June 2026, 84% of medical groups reported year-to-date costs above 2025, while 47% reported higher revenue and 36% reported lower.29 Prior authorization consumes 13 hours per physician per week, including staff time, across roughly 40 requests, and 94% of physicians say it increases burnout.30
The results show in the AUA Census. In 2014, 62.8% of practicing urologists were in private practice. By 2025, it was 43.2%.13 Private practice went from the predominant setting where most urologists worked to an environment where a minority do. In the AMA’s benchmark survey, the third most highly rated reason (63.6%) physicians gave for selling their practice was to better manage payers’ regulatory and administrative requirements.31
Private practice urologists, share of the workforce
CY2026 work RVU reductions
2.5% efficiency adjustment applied to work RVUs
52000
Cystoscopy
1.53 → 1.49
52601
TURP
13.16 → 9.75
52356
Ureteroscopy with lithotripsy
8.00 → 7.80
55866
Laparoscopic radical prostatectomy
22.46 → 21.90
TURP was also revalued (13.16 → 10.00) before the 2.5% adjustment.
0%
Net impact for urology
CY2026 final rule
−2%
Projected change in total allowed charges for urology, CY2027 proposed
+2.7%
CMS-projected practice cost growth, 2026
Figure 8 · The squeeze on private practice urology
A flat update against rising costs, and specialty-specific procedural values moving down. CY2027 figures are proposed; the comment period closed September 14, 2026, and the final rule is pending.
Sources: AUA 2025 Census13 (private practice share, 2014 and 2025 endpoints; intervening years not plotted); CMS CY2026 final rule23 and LUGPA24 (net impact); CMS CY2026 final rule23 and AUANews, February 202625 (wRVU values); AUA CY2027 proposed rule highlights26 and CMS CY2027 proposed rule fact sheet27; AMA analysis of the 2026 final rule28 (practice cost growth).
The trap is familiar to anyone working in private practice: the answer to a shrinking margin per visit is more visits. More visits need more support. More support adds an additional cost burden, so practices buy technology to reduce the support costs. But the technology carries a per-provider fee, an implementation cost, and a template-building tax that gets paid for in off-time.
On the revenue side, the JAMA study measured $167 per clinician per month in estimated E/M revenue, gross, at academic systems, across all specialties.8 On the cost side, published independent-practice pricing runs roughly $39 to $130 per provider per month; 5 of the 8 major vendors publish no price at all, and enterprise deployments are widely reported at several times that.7 On the time side, the same study measured about 16 minutes per clinic day and no reduction in after-hours work, which is where physician quality of life takes the biggest hit.8
Run the arithmetic on your own practice.
The Viscera ROI calculator can help your clinic estimate what documentation time, missed codes, undercoded visits, and unworked denials cost each year.
VIII
Six questions worth asking any AI scribe vendor
For practices still considering an ambient AI scribe, the following questions can help separate meaningful quality-of-life improvement from ambient AI noise.
How many minutes does a physician spend editing the average note with the tool, measured rather than surveyed?
What happens to the output after it is approved? Does it populate discrete fields or deliver a bucket of words someone has to copy and paste by hand? Does “integration” mean an API or a browser extension that pastes?
What does the tool do with a detail that changes the code, for example gross versus microscopic hematuria, or prostate size against a payer threshold?
What is the reasoning trail behind a suggested level? If a payer downcodes a claim, what can the clinic return that is defensible?
What did the vendor team build, and what is the clinic expected to build? How many templates is the provider asked to author, and at what tier is configuration support made available?
Does the tool support anything that happens before the patient arrives, or after they leave?
What does EHR autonomy look like?
Our EHR autonomy brochure walks through the clinical, documentation, and revenue workflow that surrounds a urology visit.
IX
Viscera is designed for EHR autonomy
Viscera is not an AI scribe; it’s a workflow platform built specifically for urology.
Viscera provides an autonomy layer around the EHR and connects the full patient story to the clinic workflow and the structured data requirements of the EHR. The work arrives prepared, providers stay in command, and the clinical, documentation, and revenue workflows surrounding a visit run without being manually reassembled each time. Urologic depth is the product.
Viscera does not turn over a note generated without context; it populates information directly from the encounter into the EHR.
Providers do not need to spend time editing or copy-pasting into the EHR; they simply review and sign off.
Viscera preps the day’s notes to situate patient encounters within a longitudinal care journey, even when the EHR goes down.
Administrative time savings are measured in hours per day, not minutes per week.
Viscera relieves the cost squeeze on private practices by coding correctly and producing defensible documentation.
Figure 9 · Viscera’s urology platform
Administrative time savings are measured in hours per day, not minutes per week.
Fewer denials, less charting, providers in command of every decision.
We’re here to help urologists practice freedom. Viscera welcomes conversations with urology practices about how their clinics actually work, the roadblocks in their processes, and what those roadblocks cost.
See what EHR autonomy could mean for you.
—
References
- 1Freed. Focus on people, not paperwork. Accessed September 23, 2026. https://www.
getfreed. ai and https://www. getfreed. ai/ features - 2Ambience Healthcare. Practice medicine, not paperwork. Accessed September 23, 2026. https://www.
ambiencehealthcare. com and https://www. ambiencehealthcare. com/ clinicians - 3Abridge. Less process. More practice. Accessed September 23, 2026. https://www.
abridge. com/ clinicians - 4Commure. Commure Scribe: the AI medical scribe for every clinician. Accessed September 23, 2026. https://getscribe.
commure. com - 5American Board of Medical Specialties. Specialty and subspecialty certificates. Accessed September 23, 2026. https://www.
abms. org/ member-boards/ specialty-subspecialty-certificates/ - 6Heidi Health. Heidi scribe; Heidi template community. Accessed September 23, 2026. https://www.
heidihealth. com/ en-us/ scribe, https://www. heidihealth. com/ templates, and https://www. heidihealth. com/ en-us/ changelog/ template-community - 7Pricing and product claims reviewed across 8 ambient clinical documentation vendors: Abridge, Ambience Healthcare, Commure, DeepScribe, Freed, Heidi Health, Microsoft Dragon Copilot, and Suki. Vendor websites accessed August–September 2026.
- 8Rotenstein LS, Holmgren AJ, Thombley R, et al. Changes in clinician time expenditure and visit quantity with adoption of artificial intelligence–powered scribes: a multisite study. JAMA. Published online April 1, 2026. doi:10.1001/jama.2026.2253
- 9Afshar M, Baumann MR, Resnik F, et al. A pragmatic randomized controlled trial of ambient artificial intelligence to improve health practitioner well-being. NEJM AI. 2025;2(12). doi:10.1056/AIoa2500945
- 10Peterson Health Technology Institute. Adoption of Artificial Intelligence in Healthcare Delivery Systems: Early Applications and Impacts. March 25, 2025. Accessed August 26, 2026. https://phti.
org/ adoption-ai-healthcare-delivery/ - 11Tierney AA, Gayre G, Hoberman B, et al. Ambient artificial intelligence scribes: learnings after 1 year and over 2.5 million uses. NEJM Catal Innov Care Deliv. 2025;6(5). doi:10.1056/CAT.25.0040
- 12Bruce G. 16K hours saved: ambient AI scribes at Kaiser Permanente. Becker’s Hospital Review. June 13, 2025. Accessed August 26, 2026. https://www.
beckershospitalreview. com/ healthcare-information-technology/ ai/ 16k-hours-saved-ambient-ai-scribes-at-kaiser-permanente/ - 13American Urological Association Education and Research, Inc. The State of the Urology Workforce and Practice in the United States 2025. AUAER; May 2026. Accessed August 26, 2026. https://www.
auanet. org/ research-and-data/ aua-census/ census-results - 14Moryousef J, Nadesan P, Uy M, Matti D, Guo Y. Assessing the efficacy and clinical utility of artificial intelligence scribes in urology. Urology. 2025;196:12-17. doi:10.1016/j.urology.2024.11.061
- 15Viscera Health. Internal practice deployment observations and remarks recorded at the Viscera Health Private Practice Urology Summit; July 2026; Dallas, TX.
- 16DeepScribe. Customization studio. Accessed September 23, 2026. https://www.
deepscribe. ai/ customization - 17American Medical Association. CPT Evaluation and Management (E/M) Office or Other Outpatient (99202-99215) and Prolonged Services (99354, 99355, 99356, 99417) Code and Guideline Changes. Effective January 1, 2021. Accessed August 26, 2026. https://www.
ama-assn. org/ system/ files/ 2019-06/ cpt-office-prolonged-svs-code-changes. pdf - 18Office of Inspector General, US Department of Health and Human Services. Improper Payments for Evaluation and Management Services Cost Medicare Billions in 2010. Report OEI-04-10-00181. May 2014. Accessed August 26, 2026. https://oig.
hhs. gov/ oei/ reports/ oei-04-10-00181. pdf - 19American Medical Association. Payer evaluation and management (E/M) downcoding programs: what you need to know. 2022. Accessed August 26, 2026. https://www.
ama-assn. org/ system/ files/ payer-em-downcoding-resource. pdf - 20Cigna. Evaluation and Management Coding Accuracy, reimbursement policy R49. Effective October 1, 2025; applies to CPT 99204-99205, 99214-99215, and 99244-99245. The pause was partial, temporary, and limited to California: fully insured HMO plans were paused October 1, 2025 pending review by the California Department of Managed Health Care, and fully insured PPO plans on November 9, 2025 for 30 days pending review by the California Department of Insurance; self-insured plans were not paused, and Cigna described the policy as in effect as of November 18, 2025. See California Medical Association, October 1 and November 10, 2025. On March 13, 2026, the Maryland Insurance Administration fined Cigna $80,000 and ordered it to stop automatic downcoding in Maryland. https://providernewsroom.
com/ cigna-healthcare/ professional-claims-for-evaluation-and-management-services/ - 21PwC. Medical cost trend 2027: behind the numbers. June 11, 2026. Accessed August 26, 2026. https://www.
pwc. com/ us/ en/ industries/ health-industries/ library/ behind-the-numbers. html - 22Blue Cross Blue Shield Association. Rising Coding Intensity and Its Impact on Health Care Affordability. Issue brief. March 2026. Accessed August 26, 2026. https://www.
bcbs. com/ dA/ 70bb93b3a9/ fileAsset/ Rising-Coding-Intensity-and-Its-Impact-on-Health-Care-Affordability. pdf - 23Centers for Medicare & Medicaid Services. Calendar Year (CY) 2026 Medicare Physician Fee Schedule Final Rule (CMS‑1832‑F). Fact sheet. October 31, 2025. Final rule published at 90 FR 49266 (November 5, 2025). https://www.
cms. gov/ newsroom/ fact-sheets/ calendar-year-cy-2026-medicare-physician-fee-schedule-final-rule-cms-1832-f - 24Large Urology Group Practice Association. CMS Releases 2026 Medicare Physician Fee Schedule Final Rule. November 2025. Accessed August 26, 2026. https://www.
lugpa. org/ cms-releases-2026-medicare-pfs-final-rule - 25Rubenstein J. Coding tips & tricks: Current Procedural Terminology code updates for 2026 of interest to urology. AUANews. February 17, 2026. Accessed September 23, 2026. https://auanews.
net/ issues/ articles/ 2026/ february-2026/ coding-tips-and-tricks-current-procedural-terminology-code-updates-for-2026-of-interest-to-urology - 26American Urological Association. Medicare Physician Fee Schedule CY 2027 Proposed Rule Highlights. July 2026. Accessed August 26, 2026. https://www.
auanet. org/ advocacy/ get-involved/ comment-letters-and-resources/ physician-payment-and-coverage-issues/ medicare-physician-fee-schedule-cy-2027-proposed-rule-highlights - 27Centers for Medicare & Medicaid Services. Calendar Year (CY) 2027 Medicare Physician Fee Schedule Proposed Rule. Fact sheet. July 14, 2026. Accessed August 26, 2026. https://www.
cms. gov/ newsroom/ fact-sheets/ calendar-year-cy-2027-medicare-physician-fee-schedule-proposed-rule - 28American Medical Association. 2026 Medicare Physician Payment Schedule and Quality Payment Program Final Rule: Summary and Analysis. November 2025. Accessed August 26, 2026. https://www.
ama-assn. org/ system/ files/ 2026-mpfs-final-rule-summary-analysis. pdf - 29Harrop C. Revenue growth narrows as costs climb: the 2026 squeeze. MGMA Stat. Medical Group Management Association; July 2, 2026. Accessed August 26, 2026. https://www.
mgma. com/ mgma-stat/ revenue-growth-narrows-as-costs-climb-2026-squeeze - 30American Medical Association. 2025 AMA Prior Authorization Physician Survey. May 2026. Accessed August 26, 2026. https://www.
ama-assn. org/ system/ files/ prior-authorization-survey. pdf - 31Kane CK. Physician Practice Characteristics in 2024: Private Practices Account for Less Than Half of Physicians in Most Specialties. Policy Research Perspectives. American Medical Association; May 2025. Accessed August 26, 2026. https://www.
ama-assn. org/ system/ files/ 2024-prp-pp-characteristics. pdf
