QiiQ Scribe

Ambient capture, speaker separation and structured note generation are the entire product, with a stated intention to expand into a wider clinical assistant. No services layer or platform business underneath.

The review gate is built into the workflow rather than described in prose. Notes carry explicit In Progress, Complete and Archive states, and the vendor states plainly that the note may be edited by the physician before it is marked as complete, so completion is a deliberate clinician action rather than a system event. That state machine is a clear mechanism. What is not published is any acceptance rate, edit burden figure or confidence threshold.

The mechanics are described and nothing is measured. The description names voice separation between physician and patient, background noise suppression tuned for the emergency department, and prompts refined for nonlinear workflows and clinical anomalies rather than generic clinical language. The vendor names no models and publishes no accuracy figure, model card or evaluation methodology.

Nothing identifies any party in the chain: no model or model family, no foundation model provider, no hosting arrangement and no subprocessor list is published, and the governing terms and conditions name none of them either.

The product's mechanics fare better. Voice separation between physician and patient, background noise suppression tuned for the emergency department, and prompts refined for non-linear workflows and clinical anomalies are all described. That is an account of what the product does rather than of what sits between a patient record and an answer, and the two are easy to mistake for one another when a feature list is detailed enough.

Residency is the specific gap and it is not a technicality here. The vendor is Canadian, sells into Canadian emergency departments, and cites the United States privacy rule, which suggests the infrastructure and the posture may have been built for a different market. Some provincial statutes constrain where health information may be held or require notice when it leaves the province, so the answer matters before deployment rather than after.

The vendor publishes no study, controlled evaluation, accuracy benchmark, third party rating, named customer or deployment count. Time saving is claimed as hours per shift without denominator or method. App store presence across iOS and Android shows an actively maintained product with a visible release history, which is a modest reliability signal rather than evidence of benefit.

The only published statement is that the product meets the United States health privacy standard, which is a compliance assertion rather than a data lifecycle. The material carries no retention schedule for audio or transcripts, no deidentification practice and no statement on whether customer content trains models.

The assertion also suggests the compliance material may not have been written for the market being served. This is a Canadian vendor whose product is built for Canadian emergency departments. The United States rule is not what governs a Canadian hospital's health information. Provincial health information statutes do, and they differ meaningfully between provinces on consent, custodianship and where information may be held. A Canadian hospital reading this page is told the vendor meets a standard that does not apply to it, and nothing about the one that does.

That may reflect United States ambitions or a template. Either way, nothing published says which framework the vendor actually operates under or which provincial statute governs a deployment, so the claim on the page is not an answer.

The capture setting makes the missing retention schedule more pointed. Emergency departments record patients who are intoxicated, confused, in acute distress or unconscious, and psychiatric presentations are documented in the same system as everything else. Nothing published states how long that audio persists, whether it is deleted after the note is produced, or whether consent is addressed at all where it cannot be obtained.

Compliance with the United States health privacy standard is claimed in the product material. The governing instrument does not support the claim: it says nothing about health information at all.

The published terms and conditions govern the site and QiiQ Scribe by name, carry a January 2024 date, and are governed by the law of British Columbia. What they cover is the sale of goods, payment methods, subscription renewal, intellectual property and acceptable use. Protected health information, patient records, custodianship, retention, confidentiality of clinical content and business associate status appear nowhere in them. Parts read as unadapted template: the terms refer to goods displayed on the site including products listed as out of stock, which describes a retail catalog rather than an ambient scribe.

So the claim falls short twice. Compliance is claimed without the underlying document, and the published notice covers the website and its payment processing rather than the service that handles patients.

The jurisdictional gap sits on top of it. This is a British Columbia company selling into Canadian emergency departments. The cited standard does not govern a Canadian hospital's health information, and provincial health information legislation, which does, is named nowhere in the company's published material.

No security evidence appears in the vendor's material: no named or dated attestation, no report of either type, no penetration testing statement and no trust center.

The distribution model is what makes the absence consequential here. The product is downloaded directly by an individual physician, with no credit card required, a free first month, and marketing that emphasizes no hassle for the technology team and starting with zero delay. Nothing in that path involves the hospital.

So there is no procurement review, no privacy impact assessment, and no institutional security assessment anywhere between the vendor and a recording being made in an emergency department. The physician carries the professional obligation and has neither the tools nor the mandate to assess a vendor's controls. The hospital carries the custodial obligation for the health information and may not know the product is in use.

That is a real gap rather than a theoretical one, because emergency medicine is precisely where a physician might reasonably adopt a tool on their own initiative to survive a shift.

The vendor is small, and unpublished controls are not the same as missing ones. But the usual proportionality argument, that a small vendor selling to sophisticated buyers can defer certification while procurement asks the questions directly, does not apply when the model is designed so that no procurement happens at all. Publishing becomes the only mechanism.

Ambient documentation needs no clearance, and none is claimed. No device pathway attaches in either the Canadian or United States framework to a note the physician reviews and edits before use.

The scribe itself is narrow, and that is a strength. As far as the vendor's material shows, it has no coding engine, no risk adjustment scoring, no decision support layer, no patient facing component and no order generation. The vendor also describes speaker separation, distinguishing the physician's voice from the patient's, and states the note may be edited before use, which places the review gate where it belongs.

What governs in the Canadian setting is professional and medico legal rather than regulatory. Emergency department documentation is the subject of sustained attention from the national medical protective body, whose published analysis of medico legal cases identifies documentation deficiencies as a recurring factor in emergency medicine claims. So the standard an emergency physician's note is held to is set by the profession and tested in claims and college complaints, not by a device regulator. A tool that changes how those notes are produced sits inside that frame.

The vendor markets a further product described as an assistant that responds to commands, positioned as a departmental upgrade beyond the scribe, with capabilities the vendor states it intends to expand. Assistants that act rather than record move toward different territory. The assistant's full scope is not published, so the scribe's position cannot be assumed to carry across to it.

Markets appear to be Canada with the application also listed elsewhere.

The vendor's material states the product handles most languages, dialects and accents, and is unhindered by background noise. No evaluation, methodology, language list or subgroup data accompanies it. Alongside the claim the vendor invites the reader to see for themselves.

A claim without evidence is worse than silence, because silence leaves the question open while a claim invites people to stop asking it.

The invitation to verify personally is the part that most needs answering, because it sounds reasonable and cannot work. An individual clinician trying the product across their own encounters learns whether it works for them, with their voice, in their department, with the patients they happened to see. It cannot detect the failure that matters most here, because that failure is distributional. A system performing well on average and materially worse for one group of speakers looks fine to almost every individual tester, and the clinicians who would notice are those whose patients are affected, who are the least likely to be running the evaluation.

The setting makes it sharper still. Emergency departments serve unscheduled populations with the widest range of language, intoxication, distress and acuity of any care setting, which is the hardest acoustic and linguistic environment a speech model faces, and the one where a degraded transcript is least likely to be caught by the patient.

Rather than merely being silent, the vendor's own instrument runs against the customer.

The published terms exclude liability for any actions, claims, losses, damages, liabilities and expenses arising from use, without carve-out, and nothing in them attaches any commitment to the accuracy of a generated note. A consumer protection clause preserves mandatory statutory rights where that legislation applies, which is unlikely to reach a hospital or a physician group contracting in the course of business. No accuracy figure, warranty, indemnity or remediation commitment is published anywhere.

The statutory position is real, but it comes from the law rather than from the vendor. Canadian provincial health information legislation gives an individual an enforceable right to have inaccurate information corrected, with a regulator behind it, and the custodian hospital carries that obligation regardless of what its supplier publishes. The vendor has published no undertaking to support it.

The setting sharpens all of it. Emergency departments record patients who are intoxicated, confused, in acute distress or unconscious, and nothing published addresses consent where it cannot be obtained. The person least able to object at the time is the one with the least described route after the fact.

The default workflow is transfer: notes are viewed in the web application and copied into the record. EMR integration exists but is offered on request rather than shipped, with the vendor inviting interested buyers to make contact, which means integration is bespoke and its depth is unknowable in advance.

The pitch that a clinician can push notes into the record without bothering the IT department is appealing for a physician group with no informatics support, and it is also the description of a workflow that bypasses the usual integration governance.

No hosting region, residency option or subprocessor detail is published, and nothing names the model service that processes the encounter.

That is a specific gap rather than a generic one. This is a Canadian vendor whose buyers face provincial residency expectations that differ by province, and several provinces expect health information about their residents to be held in country or impose conditions on cross border transfer. A vendor serving those buyers who does not state where data sits has left them unable to answer the first question their own privacy office will ask.

One further fact changes how the integration picture reads. Record system integration is not a delivered capability. The vendor invites interested parties to get in touch about it, which places it somewhere between a roadmap and a bespoke engagement. So for a clinician downloading the application today, output leaves the product by hand.

That matters beyond convenience. Where notes are moved manually, the vendor's environment holds the only copy of the encounter for whatever interval elapses before transfer, and no institutional system has visibility of what was captured. It also means the deployment sits outside hospital information technology entirely, since the vendor markets that there is no hassle for the technology team and that a clinician can start with zero delay.

That is a real adoption advantage and it means the hospital whose patients are being recorded may not know the product is in use.

The trial is published plainly and the price is not. A first month free with no credit card required is advertised, and team discounts are promoted without figures. No rate card, tier structure or per clinician figure is published, and team pricing requires contact.

Self-service entry and a real trial are strengths rather than gaps, and they are what makes the posture legible. The unit of charge is the individual clinician, billing is monthly, and a buyer can start without a sales conversation. Only the number is missing.

The governing terms are more specific than the marketing. The trial runs thirty days from registration with unlimited use, and at its end the subscription renews automatically and the account is billed the monthly rate unless it is canceled first. Credit card, PayPal and debit are accepted.

Put together, the two facts are sharper than either half. The offer is an automatic recurring payment obligation, entered by self-service, at a rate the vendor publishes nowhere. A clinician can be billed a figure they were never shown.

Deliberately narrow and specialized, the same shape as Sayvant in the same specialty. The emergency department differs from outpatient care in ways this product addresses directly: acoustic clutter, and a non linear workflow where a physician holds several patients open at once, is interrupted, and returns to encounters out of order, which is why the note state machine matters more here than elsewhere.

Note structure covers history of present illness, review of systems, exam and assessment and plan. The stated roadmap runs from the emergency department to the wider hospital and then outpatient, so current coverage is one setting done properly, not several done thinly.