AutoScribe by Mutuo Health
AutoScribe is the ambient documentation product of Mutuo Health Solutions, a Toronto company founded by AI researchers and a family physician, and it is the only vendor in this category that began as a published research project. AutoScribe was developed through a collaboration between the University of Toronto and Saint Michael's Hospital, and the team published a peer reviewed case study of the technology that describes its drawbacks as well as its benefits, alongside a pilot validation study of the note generating model in primary care. It was pre qualified as a vendor under Canada Health Infoway's National AI Scribe Program, and in New Brunswick it reaches clinicians white labelled as Nexus AI. It handles 21 of the most commonly spoken languages in Canada and can generate the note in a different language from the conversation. Its published clinician guidance is unusually conservative, anchored to College of Physicians and Surgeons of Ontario and Canadian Medical Protective Association positions, and it cites independent safety literature about its own product category. Mutuo now operates as a WELLSTAR company, WELLSTAR being a majority owned subsidiary of WELL Health.
Capability Axes
The company exists to commercialise a research model. AutoScribe originated as a University of Toronto and Saint Michael's Hospital research project applying natural language processing to clinician patient dialogue, and the founding team combines machine learning researchers with a practising family physician. There is no non AI business underneath.
The most conservatively framed oversight position located in this category, though it is a posture rather than a mechanism. The vendor states plainly that an AI scribe produces a draft the physician reviews and signs, that the value is removing the blank page rather than removing the clinician, and that the clinician remains firmly in control of the final record. More unusually, its published clinician guidance is anchored to College of Physicians and Surgeons of Ontario and Canadian Medical Protective Association positions on AI use and medico legal risk, and it cites independent peer reviewed work evaluating the quality and safety of ambient digital scribe platforms generally. Citing external safety literature about your own product category is not something vendors normally do. Held at B rather than A because no acceptance rate, edit burden figure or confidence threshold is published, and no structural verification mechanism of the kind that earned Abridge, Speke, Corti and TORTUS an A is described.
Rare among vendors here in having a peer reviewed description of its own technology, and rarer still in what that paper contains. The published case study sets out the benefits AND THE DRAWBACKS of the approach, which is a vendor volunteering its own limitations in an indexed venue rather than a marketing page. A separate pilot validation study on the note generating model in primary care is documented through an academic research group. Held at B because no current accuracy figure, model card or evaluation methodology for the shipping product was located, and the published work predates the generative era, so it describes a lineage rather than today's system.
Genuine academic provenance rather than testimonial. The technology was developed through a University of Toronto and Saint Michael's Hospital collaboration, described in a peer reviewed publication, and subjected to a pilot validation study of the note generating model in primary care through an academic research group. Externally, AutoScribe was pre qualified as a vendor under Canada Health Infoway's National AI Scribe Program, a national body assessment, and in New Brunswick it is identified as the approved scribe underpinning the Nexus AI deployment. Held at B because no outcome study, controlled comparison or published accuracy result for the current product was located; what is evidenced is rigorous origin and external vetting rather than measured clinical benefit.
Operates under PHIPA and Canadian privacy regulation, with a named privacy officer contact rather than a generic form. Notably, the vendor states that third party security and privacy assessments exist and will be provided to organisations on request, which means external assessment has been performed even though the reports are not published. Held at B because the public privacy language is generic, referring to standard measures required by regulation, and no retention schedule, de identification practice or training use statement was located. Ask for the assessments; they exist.
Not applicable as assessed. Deployment is Canadian under PHIPA and provincial privacy regimes rather than HIPAA, and no United States commercial motion was evidenced in this pass.
Better than nothing and short of published. No named certification such as SOC 2 Type II or ISO 27001 was located and no trust centre exists, but the vendor confirms it holds third party security and privacy assessments and names a privacy officer address to request them. That is positive evidence that external review has happened, which distinguishes this from the vendors here graded Not Rated where no assessment is evidenced at all. Publishing the certification names and dates would move it immediately.
Not a regulated medical device and none claimed. The governing framework here is professional regulation rather than device regulation, and the vendor engages with it directly by publishing guidance referencing College of Physicians and Surgeons of Ontario expectations.
Engages seriously with the professional and medico legal governance of its own category, which is a different and in some ways more useful contribution than a fairness statement. Its published material explains to clinicians what regulators expect of them when using an AI scribe, distinguishes the tool from dictation in the terms the regulator uses, and cites Canadian Medical Protective Association guidance on medico legal exposure and independent literature on ambient scribe quality and safety. On language it is specific rather than expansive, supporting 21 of the most commonly spoken languages in Canada and disclosing that the note can be produced in a different language from the conversation, which is an honest description of a translation step that most vendors leave implicit. Held at B because no fairness statement, subgroup analysis or performance breakdown by accent or language was located.
Canadian EMR focused, with reach extended two ways. It is white labelled as the underlying technology for Nexus AI in New Brunswick, giving it distribution under another brand, and Mutuo now sits inside the WELLSTAR group, a majority owned subsidiary of WELL Health, whose ecosystem is described as serving more than 40 percent of Canadian practitioners. That ownership position is a plausible route to deeper native integration than a standalone vendor could negotiate. Held at B because named EMR integrations and their depth were not enumerated or verified in this pass.
Not assessed. No hosting region, residency commitment or sub processor detail was located, which is a notable gap for a product sold into Canadian provincial privacy regimes where residency expectations are explicit.
No rate card, but a real trial and a public funding route. A one month free trial is offered, and eligible Canadian primary care clinicians may access AutoScribe through Canada Health Infoway's National AI Scribe Program, where funding is offered on a first come first served basis, which makes the effective cost to a clinician a function of a public programme rather than a private negotiation. Held at C because no list price was located for buyers outside that programme.
Primary care centred, consistent with its research origin and with the Canada Health Infoway programme it was pre qualified for, which targets primary care providers. Supports 21 of the most commonly spoken languages in Canada with cross language note generation. An AutoForm capability to prepare standard forms such as insurance documentation from encounter content is described as forthcoming rather than shipped. No specialty count published and no acute or inpatient coverage evidenced.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | BAA Tier | Implementation | Source |
|---|---|---|---|---|
|
Not published. One month free trial; funded access via the Canada Health Infoway National AI Scribe Program
|
Not disclosed. Direct sale plus access through a publicly funded national programme, and white label distribution as Nexus AI in New Brunswick. | PHIPA and Canadian provincial privacy regimes rather than HIPAA. Third party security and privacy assessments available on request from a named privacy officer. | None published. | Vendor Published |
No list price located, but the practical cost question for the target buyer is not a list price. Eligible Canadian primary care clinicians can reach AutoScribe through Canada Health Infoway's National AI Scribe Program, where funding is allocated first come first served and selection happens through national or provincial registration portals, so the effective price depends on programme eligibility and remaining funding rather than negotiation. A one month free trial is offered outside that route. One item belongs in diligence rather than pricing: ownership has moved, with current materials describing Mutuo as a WELLSTAR company under WELL Health while an earlier post described it as a Healwell AI company, so confirm who holds the contract.