Vero Scribe

Ambient capture, note generation, evidence search and code lookup are the entire product, with no services or platform business underneath.

No review gate, sign off step, confidence threshold, acceptance rate or edit burden figure is described. The product's positioning is that a clinician finishes the note before the patient leaves the room, and the interface offers commands and assisted editing, but nothing states what the clinician is required to confirm.

The second pass finds the gap is wider than the note alone. The platform produces the note, generates diagnosis codes in two coding systems, drafts follow up tasks, and fills forms. Each is a different artefact with different consequences, and a single sign off on a note does not obviously cover a code, a task or a completed form. Establish what confirmation applies to each.

One governance layer is present and worth crediting on this axis, because it is the institutional half of oversight rather than the individual half. The vendor supports shared patient context across a team with permissions, organisation scoped templates, and administrator and member roles. That gives a practice the ability to standardise documentation and see who did what, which is more than a purely individual tool provides and is the layer most direct to clinician products in this lane lack entirely.

One deployment decision made by a hospital using the product is instructive and belongs here. That institution has excluded trainees from using the tool. It is a considered call: a clinician learning to document is building a skill the product substitutes for, and the oversight question for a trainee differs from that for an experienced physician. No vendor in this lane addresses trainee use, and a teaching institution should ask about it.

Ask for an edit rate, the behaviour on low confidence output, and what a supervisor can see.

No accuracy figure, model card, named models or evaluation methodology located. Applying the standing question for this lane, whose model is it: nothing was found disclosing whether the underlying models are proprietary, licensed, or a general purpose foundation model with a healthcare layer on top, and for a vendor competing on price against enterprise incumbents that distinction is exactly what a buyer would want to know.

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 standing question for this lane applies directly and goes unanswered, which is whether the underlying models are proprietary, licensed, or a general purpose foundation model with a healthcare layer on top.

For a vendor competing on price against enterprise incumbents that distinction is exactly what a buyer would want to know, because it determines where the cost advantage comes from and which third party's policy changes could affect the product.

Configurable retention from one day to three hundred and sixty five with automatic deletion is published, which bounds how long content persists without saying who holds it during that window, and uploaded documents such as prior notes and consult letters may follow a different schedule from encounter audio, which is unstated. Ask whose models these are, for a sub processor list, and whether uploaded documents follow the same retention schedule as recordings.

No study, controlled evaluation, accuracy benchmark, third party rating or named health system deployment located. Evidence is testimonial. Note the sourcing problem specific to this vendor: because it publishes a large volume of comparison content about the whole category, search results about Vero and about its competitors are heavily populated by Vero's own writing, which is not independent assessment in either direction.

The retention position is now located and it is among the better ones in this category. The vendor publishes configurable retention windows from one day to three hundred and sixty five, with automatic deletion, and frames the choice as the customer's.

That combination answers more than a fixed default does. A stated range tells a buyer the shortest setting available, which is the number that matters for a privacy impact assessment, and configurability lets a practice with a strict regional requirement set it accordingly rather than accept a vendor's judgement. It is a better disclosure than the vague promise of prompt deletion that is the norm here, and better than configurable without a stated range, which leaves a buyer unable to tell what configuring achieves.

One caution belongs with it. Where retention is configurable the default matters most, because most customers never change it. Establish what the setting is on signup.

What holds this short of the top grade is the training question, which is unanswered in either direction. No statement was located on whether customer audio or transcripts are used to train or improve models, and no de identification practice is described. Peers now address this plainly, some committing never to train on clinical content and at least two operating an explicit permission gate. Either formulation would close it.

File upload sharpens the question rather than softening it. Prior notes, consult letters and other documents uploaded into the platform are not encounter audio and may carry a different retention treatment.

Ask for the default retention setting, the training position in contract language, and whether uploaded documents follow the same schedule.

Claims compliance across two regimes, HIPAA and PIPEDA, consistent with a company operating from Canada into the United States market. Business associate agreement terms are not published for inspection, and the product is self serve with a permanent free tier, so use can begin before any agreement conversation.

The earlier assessment found no attestation. That is overturned. The vendor states SOC 2 Type II certification on its own material, alongside end to end encryption, role based access controls, an invitation and permissions system, and administrator and member roles.

SOC 2 Type II is the stronger report type, testing whether controls operated across a period rather than whether they were designed at a point in time.

One formulation alongside it should be read precisely rather than added to the total. The vendor describes itself as fully compliant with the applicable health privacy security rule and with controls drawn from a health sector certification framework. Alignment to a framework's controls is not certification against it, and the two are easy to conflate. A vendor that has mapped its controls to that framework has done useful work and has not been examined against it. The SOC 2 report is the examined item here; the rest is self assessment.

What holds this short of the top grade is specificity. No report date, audit period or scope statement was located, and scope matters because the platform spans documentation, evidence retrieval, coding and form filling.

The buyer context is worth noting. This vendor sells into Canadian clinics where regional health authorities have raised expectations on residency, third party audit and retention controls, and where provincial privacy regulators have now issued specific guidance on scribes. That is a market where the report will be requested as a matter of course.

Ask for the report, its period, and the systems in scope.

No clearance claimed and none required for documentation. Evidence retrieval is a reference capability rather than a recommendation engine on the material located. This axis does not read as an absence, for two reasons.

The first is the regime. Ambient scribes in Canada are governed by provincial health privacy law rather than device regulation, and regulators in Ontario, British Columbia and Alberta have each issued specific guidance. The standards differ and the strictest is instructive: in Alberta a custodian may collect health information only where expressly authorised, so recording a full encounter must be essential to care or another permitted purpose rather than merely more efficient. Efficiency, which is every vendor's pitch in this category, is explicitly not sufficient justification there. Consent requirements apply across all three.

The second is evidence a buyer can act on, and it is the strongest of its kind in this lane. A Canadian hospital deploying this product publishes a patient facing page naming it, stating that the provider will seek verbal permission before recording, that a patient may decline or withdraw at any time, and that care will not be affected if they refuse. It supplies the wording a clinician should use, and records a deliberate scoping decision that trainees will not use the tool.

That material belongs to the institution rather than the vendor and the distinction should be kept. It does not evidence anything about the vendor's own governance. It does establish that the product has passed a hospital's privacy and clinical governance review in a jurisdiction with published regulatory guidance, and that the deploying institution documented the outcome publicly.

Ask to see the vendor's own consent material and its position under each provincial regime.

A correction first, because the earlier assessment recorded a point in the vendor's favour that rested on a misreading. It noted multilingual support stated modestly at up to three languages and credited the restraint. Three is the number of languages the product can capture simultaneously within a single recording. The total claimed is sixty. So the modesty credit is withdrawn, and this record sits alongside the other large unverified language claims in this category rather than apart from them.

On the substance, no fairness statement, subgroup analysis, accuracy breakdown by accent or dialect, or evaluation methodology was located.

Sixty languages is a substantial claim and nothing accompanies it: no list, no per language accuracy, no indication of which are well supported and which are nominal. Simultaneous capture of three languages in one encounter is a harder problem again, since it requires separating speakers and languages at once, and it is precisely the capability a bilingual consultation with a family member interpreting would rely on. No evidence for it was located.

The deployment context makes this more pointed rather than less. The product is used in Canadian urban practice, where multilingual consultations are routine rather than exceptional, and one of its marketing illustrations shows a Spanish language exchange rendered into an English note. A translation step inside the documentation pipeline changes what the signature attests to, because a clinician who does not speak the source language cannot verify the rendering.

Ask for the language list, per language accuracy, and what the product does when confidence is low in a language the clinician does not speak.

Two passes located no accuracy or error figure, no published limitations and no warranty, indemnity or remediation commitment for the output. The vendor does publish a genuinely good control on the data side, with configurable retention from one day to three hundred and sixty five and automatic deletion, framed as the customer's choice, and a stated range is a better disclosure than the vague promise of prompt deletion that is the norm here because it tells a buyer the shortest setting available.

That governs how long content is held rather than whether it is right, so it does not reach this axis. Nothing here does. There is no published characterisation of where the system degrades, no correction mechanism described beyond the ordinary ability of a clinician to edit a draft, and no commitment of any kind attached to the note.

One consequence of configurability belongs in diligence and applies wherever a vendor offers it: where retention is a setting, the default is what most customers will live with, so establish what it is on signup rather than what the range permits. Ask for an accuracy figure with a definition, the default retention setting, and what the vendor commits to when a note is wrong.

Integration is asserted in general terms as seamless EMR integration, with no named systems, architecture or write back mechanism verified in this pass. A third party marketplace listing names Epic, Cerner and MEDITECH, but the surrounding text on that listing is boilerplate that appears identically on other vendors' entries, so it carries little weight.

The strongest signal is the pricing structure itself: custom EMR integrations are listed as an Enterprise tier feature, which implies the standard product does not ship deep integrations and that most buyers at the published price are transferring notes rather than writing to structured fields.

The earlier assessment found no residency detail. That is overturned and the answer is one of the clearer ones in this lane. The vendor states that Canadian customers' data is stored in Canada.

That commitment is worth more here than a general statement about encryption, because it answers the question Canadian buyers are specifically told to ask. Provincial privacy regulators in Ontario, British Columbia and Alberta have each issued guidance on ambient scribes, and where information may be held is among the first items a custodian must resolve. A vendor that states residency plainly has removed the obstacle that stops most procurement conversations in that market.

The platform is delivered across web, mobile and desktop operating systems, with recording, typing, dictation and file upload as input paths.

Three things hold it short of the top grade, and they are gaps rather than doubts. The commitment names one jurisdiction, so it establishes nothing about where a United States customer's data sits, and this index has seen the same shape elsewhere: an onshore promise for the home market that says nothing about the others. No subprocessor list was located. And no model provider is named, so whether an external service processes the encounter, and whether that service is also in region, is unresolved. A residency commitment binds the vendor and not necessarily its suppliers.

Ask whether residency is mirrored per jurisdiction, for the subprocessor list, for the model provider, and specifically whether the Canadian commitment extends to any external processing the pipeline depends on.

A complete and self serve ladder, published rather than quoted. A seven day unlimited trial converts to a PERMANENT free tier of around ten encounters a month, the paid tier is roughly 69 US dollars a month billed annually or 79 monthly with unlimited notes, custom templates and AI commands, and an Enterprise tier is offered with custom integrations, white labelling and analytics. A permanent free tier rather than a time limited trial is genuinely rare here and lets a low volume clinician use the product indefinitely without a purchase decision.

Note types are enumerated properly rather than implied, covering SOAP, progress notes, consults, referrals and discharge summaries, and platform coverage is unusually wide across web, mobile and every major desktop operating system, which matters for clinicians on managed or non standard machines. Held at B because the headline claim of more than 150 specialties is asserted without enumeration or any specialty tuning detail, and language support is limited to around three languages.