ScribeMD

Ambient capture and note generation are the entire product, delivered self serve with an API alongside. No services layer or platform business underneath.

The open question on this record is now resolved as far as published material allows, and the resolution matters for how third party sources should be used.

The earlier assessment recorded that a third party review described this product as operating a hybrid model combining machine learning with human oversight, which if accurate would place it in the human review cluster alongside Speke, ScribeRyte and Scribble and would materially change the record. A second search found that claim repeated on a further aggregator listing, and found nothing supporting it anywhere in the vendor's own materials.

What the vendor describes is the opposite shape. Its own account of the workflow is that the clinician starts a session, records the visit, and the software generates a draft note which the clinician reviews, edits and exports. No intermediate reviewer appears anywhere in that description. Vendors in this index that do operate a human review layer market it prominently, because it is the expensive part and the differentiator.

Treat the hybrid claim as unsupported. Aggregator listings in this category copy one another, and a claim appearing on two such sites is one source rather than two.

That leaves the ordinary position for this axis. The clinician is the only reviewer and the signature is the control point, which is the category norm. What is missing is any published detail about how well that control performs: no confidence threshold, no acceptance or edit rate, no description of what the system does when uncertain, and no statement of whether anything is flagged for closer attention. Ask for an edit rate and for the behaviour on low confidence output.

The figures circulating for this product, 98 percent transcription accuracy and 0.2 second processing on a proprietary large language model, come from a third party software directory rather than the vendor's own materials, and carry no methodology, reference standard or date. Nothing was located on the vendor's site to substantiate them. The site does describe itself as the number one AI medical scribe, which is unfalsifiable and joins the other superlatives this index tracks. No model card or evaluation protocol located.

Nothing identifies any party in the chain: no model or model family, no foundation model provider, no hosting arrangement and no sub processor list was located in two passes. The one claim that touches this axis, that the product runs on a proprietary large language model, appears in a third party software directory rather than in the vendor's own material, so the only statement a buyer can find about what sits behind the product was not made by the company that sells it.

A proprietary claim carries weight only when its author is accountable for it. Retention configurability being reserved to the top commercial tier compounds the position, because a customer on a lower plan cannot establish either who holds their content or for how long. Ask the vendor directly whether any third party model provider is invoked, for a sub processor list, and for the default retention position on the plan you are actually buying.

Adoption is claimed at more than 3,000 physicians, which is scale rather than benefit. No study, controlled evaluation, accuracy benchmark, third party performance rating or named institutional customer was located. Graded C on the standing precedent that adoption does not substitute for evidence.

Graded on a disclosure that comes from the pricing page rather than a privacy statement, which is itself the finding. Customising data retention and deployment is listed as a custom enterprise tier feature, which tells a buyer two things: retention is configurable, and that control is unavailable on the free and standard plans.

Default retention is not published anywhere located, so a clinician on the 99.99 dollar plan cannot determine how long their patient audio is held. No de identification practice or training use statement was located.

HIPAA compliance stated consistently across product and pricing materials. Business associate agreement terms are not published for inspection, and signup is self serve with a free tier, so use can begin before any agreement conversation.

SOC 2 certification is named directly, which places it above the near attestations tracked elsewhere in this index such as SOC 2 aligned or certified data centres. Held at B rather than A because the report TYPE is not stated and Type I and Type II are materially different assurances, nor is an audit date or scope given. Ask which and when.

No clearance claimed and none required for ambient documentation. The clinician reviews, edits and exports the draft, and that signature is the control point.

Two scope items sit around the documentation core. The product is marketed to dentists as well as physicians, which does not change the device analysis but does change the professional and payer context, since dental documentation runs through different coding and different payer rules.

More substantively, the vendor markets AI driven pre authentication aimed at faster claim approvals. Prior authorisation is a payer facing function rather than a documentation one. It sits on the reimbursement side of the boundary this index tracks across the category, and it raises a question note generation does not: what the system submits to a payer on the practice's behalf, and who is accountable for the accuracy of that submission. Establish whether the feature prepares a request for a human to send or transmits one itself.

Beyond that the product is narrow. No coding engine, no decision support module and no patient facing product were located, which keeps the core regulatory position simple.

Markets are wider than the United States. The vendor's site is localised into eight languages, and ambient documentation is treated as software as a medical device in the United Kingdom and European Union. No registration or classification under that regime was located, so a buyer in those markets should ask for it directly rather than assume the United States position travels.

A second search confirms the absence. No fairness statement, no subgroup analysis, no accuracy breakdown by accent, dialect or speaker population, and no evaluation methodology were located.

The absence is more conspicuous here than for a domestic only vendor, and the reason is the product's own positioning. The site is localised into eight languages and the vendor markets to multilingual practices. The range of accents, dialects and language backgrounds this system encounters is therefore wider by design than most of its category, which makes missing evidence about performance across that range more important rather than less. A vendor that chooses breadth has taken on the harder version of the problem.

The imbalance familiar across this lane is present too. An accuracy figure circulates for this product in the high nineties, presented as a single headline number with no denominator, no methodology, no specialty or language breakdown and no independent validation. A single number spanning eight languages is not a measurement a buyer can act on, because it cannot distinguish strong performance in one language from weak performance in another. The average conceals exactly the variation this axis exists to surface.

None of this is difficult to ask about. Peers in this category have submitted note quality to published evaluation against validated documentation quality instruments, and at least one has entered independent comparative benchmarking. The methods exist and are already in use in this segment, so the gap is a choice rather than a limitation.

Two passes located no accuracy or error figure from the vendor, no published limitations, and no warranty, indemnity or remediation commitment. The figures circulating for this product, a transcription accuracy percentage and a sub second processing time on a proprietary large language model, come from a third party software directory rather than the vendor's own materials and carry no methodology, reference standard or date.

Nothing on the vendor's site substantiates them, and a number a buyer finds in a directory is not a representation the vendor has made and cannot be relied on as one. The site does describe the product as the number one artificial intelligence medical scribe, which is unfalsifiable and joins the other superlatives this index tracks.

One commercial structure has a direct effect on recourse and is worth naming, because it is the clearest instance in the lane of control over your own data being sold rather than provided. Customising data retention and deployment is listed as a feature of the custom enterprise tier, which tells a buyer two things at once: retention is configurable, and that control is unavailable on the free and standard plans.

Default retention is not published anywhere located, so a clinician on a paid standard plan cannot determine how long their patients' audio is held, and cannot change it. Ask for the default retention period for each tier in writing, and for any accuracy claim the vendor is willing to make itself.

An integrations page and a public API both exist, which is positive evidence of an integration surface rather than a copy and paste only product, and a third party review describes notes syncing across devices and record systems. But no named electronic health record integration, architecture or write back mechanism was verified in this pass, so the depth cannot be assessed. For a self serve product at this price point, confirm what your specific system actually gets before assuming sync.

Partial and inferred from the commercial ladder rather than from documentation. Deployment is listed as customisable on the enterprise tier, which establishes that options exist, and a branded mobile application is offered at that level. No hosting region, residency commitment or sub processor detail was located for any tier.

A complete published ladder with the limits stated rather than implied. The free tier is 10 conversations a month with nine speaker support and a 48 hour support response commitment. The paid tier is 99.99 US dollars a month for unlimited conversations, adding analytics, custom note formatting and a 24 hour support response. The enterprise tier is custom quoted and names what it adds: a branded mobile application, customisable data retention and deployment, one hour support response and team management.

Publishing support response times per tier is unusual and useful, since support quality is a real differentiator that almost nobody commits to contractually.

Broad and unusually specific in two directions.

It states support for up to nine speakers, which is the most concrete multi speaker capacity published by any vendor in this index where competitors say only multi speaker, and that capability sits in the free tier rather than behind an upgrade. Nine speakers covers family consultations, paediatric visits with both parents, interpreter present encounters and small case conferences, which are exactly the situations two voice models handle badly.

Coverage spans a claimed 50 or more specialties including dentistry, a note structure almost nothing else attempts, alongside psychiatry and cardiology. The product is also genuinely international, with the site localised into English, Spanish, French, Portuguese, Korean, Japanese, Arabic and Greek.