Speke

The model does real work, but the vendor's own pitch is that the model is not sufficient on its own. Speke is explicitly positioned as AI paired with ScribeAmerica scribes, and the parent's moat is a twenty year staffing and scribe services business spanning 3,000 sites, not a model.

This is not the services rejection this index applied to CORL or symplr, because Speke is a software product with a services layer rather than a services business wearing AI language, and the index already accommodates tiered human review through Commure. Graded B: the intelligence is genuine and central, but so is the human workforce, and the vendor says so rather than hiding it.

The strongest oversight architecture in this category, and it earns the grade structurally rather than by promise. Trained medical scribes review and edit generated notes for accuracy and completeness before the provider signs, and they actively alert the provider through the app when the encounter is missing information a note requires, which is a prompt no automated scribe produces because no automated scribe knows what it failed to hear.

Crucially the depth of that review is a purchasable variable across Express, Plus and Pro tiers, so oversight is a procurement decision the buyer controls rather than a fixed vendor design. Compare Abridge, graded A for Linked Evidence, which makes verification auditable; Speke makes it someone's job. Held short of a published edit rate or acceptance threshold, which would be the obvious next disclosure.

No accuracy figure, no model card, no named models and no evaluation methodology. The technical claims offered are credentials rather than disclosures, notably that product and machine learning staff were hired from Amazon and Microsoft, which says nothing about how the system performs. Language coverage is also stated inconsistently across the vendor's own properties, appearing as 25 or more languages in one place and 30 or more in another.

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. As with several vendors in this lane the chain includes people as well as systems, since a human workforce reads recorded clinical conversation and edits draft documentation as a standard step, and that workforce is not enumerated in any respect.

Where it sits, whether review is onshore, how reviewers are vetted, and whether they are employees or contracted are all unstated, and the parent operates a separate remote scribe service, which makes the location question concrete. A buyer evaluating this product is therefore unable to name a single party that handles its encounter content beyond the vendor itself. Ask for the list of entities and countries that receive recordings or drafts, and for a sub processor list covering the automated path.

Independent third party validation exists, which lifts this above the vendor reported claims typical of the category, but it should be read for what it is. KLAS published an Emerging Technology Spotlight on Speke, which validates the experiences of organisations actually using it rather than repeating vendor assertions.

An Emerging Technology Spotlight is an early stage validation based on a small interview base and is not equivalent to a Best in KLAS ranking, so it evidences real buyer experience rather than comparative performance. Vendor reported figures alongside it, up to 50 percent less documentation time and an average of 2.5 hours saved daily, carry no denominator or design. No peer reviewed publication located.

The mechanism that earns this product its autonomy A creates its distinctive privacy exposure, and buyers should evaluate both as one decision rather than separately. Human scribes reviewing and editing notes means recorded clinical conversation and draft documentation are read by a human workforce, which is a materially larger and more variable PHI exposure surface than a fully automated pipeline.

Nothing was located on where that workforce sits, how it is vetted, whether review is performed onshore, how audio and transcripts are retained, or whether customer data trains models. Given that ScribeAmerica also operates a TeleScribe service, the workforce location question is the specific thing to get answered in writing.

No product specific business associate agreement posture, template or scope statement was located.

The earlier assessment flagged that this matters more than usual given a human review workforce handling protected health information. The second pass confirms that and adds two complications.

First, the tiers differ in who touches the data. The entry tier is software led. At the higher tiers a remote scribe employed by the vendor reviews the encounter, produces the documentation and routes it into the customer's record system, which means vendor personnel hold working access to the record itself rather than only to an audio file. An agreement scoped to the software does not describe that arrangement, and an agreement written for one tier does not describe another.

Second, and unusually, this product is sold to veterinary practices as well as to human healthcare providers. Veterinary records fall outside the health privacy rule entirely. So a single vendor compliance posture is being applied across customers where the rule governs completely and customers where it does not apply at all. A healthcare buyer should establish that the terms offered are the healthcare terms rather than a general form written to cover both.

Scale raises the stakes rather than lowering them. This is an established organisation operating across a large client base and many care sites with a workforce built over two decades, so the questions concern a mature operation rather than a startup's gaps: who the workforce are, whether they are employees or subcontracted, where they work, and what their record access reaches beyond the encounter they are documenting.

No SOC 2 report of either type, no HITRUST certification, no ISO 27001 and no trust centre were located in a second pass.

The absence is more consequential here than for a software vendor of comparable size, for the reason established across this cluster. This organisation's model places its own trained personnel in the documentation path. At the higher tiers those people review encounters, author notes and write into the customer's record system. That is a standing internal access surface of substantial size, operated across a large client base and many care sites.

It is precisely what an independent examination addresses: how personnel are vetted before being given access to patient records, how that access is provisioned and revoked, what a scribe can reach beyond the encounter they are documenting, whether their activity is logged in a form the customer can audit, and what happens when someone leaves or a subcontract ends.

Two smaller points. Recognition in a healthcare research firm's technology spotlight appears in the vendor's material, with customers giving high performance grades and stating they would buy again. That is a customer satisfaction assessment and a genuine market signal, and it is not a security examination. The two are different objects and only the second speaks to this axis. Separately, the length of the organisation's operating history is a reason to expect documentation to exist rather than a substitute for having it published.

The grade reflects what a counterparty can verify before contracting rather than a judgement that controls are absent. Ask for the report and its scope, and separately for the access control and monitoring model covering the scribe workforce.

No clearance claimed and none required for ambient documentation.

Two things on this record deserve attention, and the first is a credit this index does not often get to record.

The vendor states that the platform processes audio in more than twenty five languages and generates a note in English, and then says explicitly that it cannot act as an interpreter and that a native speaker of the recorded language will need to corroborate the note's accuracy. That is a vendor publishing the limitation of its own capability alongside the capability, and telling the buyer what mitigation is required. Compare the same feature elsewhere in this category, where cross language note generation is offered with no such caveat at all. The point bears directly on what a clinician is signing: where an encounter took place in a language the reviewing clinician does not speak, the signature attests to a translation the signatory cannot check, and this vendor is the one saying so.

The second is the service structure. At the higher tiers a remote scribe employed by the vendor reviews the encounter, performs the documentation with the assistance of the software, and routes the completed note into the customer's record system. The clinician then reviews and approves. That is a different object from a clinician editing a machine draft. The documentation is authored by a third party working from audio, and the clinician's signature attests to another person's account of an encounter they themselves conducted.

One scope note. The product is marketed to veterinary practices alongside human healthcare. That does not change the device analysis, since neither is regulated as a device here, but it does mean one product spans two entirely separate regulatory worlds.

Contains the single best language scope disclosure located in this category, held at B only because it stands alone rather than sitting inside a formal governance document. Speke states that it processes audio in 25 or more languages and generates the note in English, then adds explicitly that it cannot act as an interpreter and that a native speaker of the recorded language will need to corroborate note accuracy.

That does two things almost no competitor does: it names the limitation, and it assigns the verification duty to a specific person rather than leaving it unowned. Every vendor marketing multilingual capture should be asked to match that sentence. What would move this to A is what Microsoft Dragon Copilot publishes alongside it, a formal transparency document covering accent and dialect performance, unsupported use cases and subgroup considerations. No fairness or subgroup analysis was located here.

Human review of generated notes is part of how this product works, and a person checking machine output before it reaches the clinician is a real correction mechanism, which keeps this off the floor. Beyond that the record is bare. No accuracy figure, error rate or evaluation methodology was located, and no warranty, indemnity or remediation commitment.

The performance claims offered are credentials rather than measurements, notably that product and machine learning staff were hired from two large technology companies, which describes a hiring record and says nothing about how the system performs. Language coverage is also stated inconsistently across the vendor's own properties, appearing as 25 or more in one place and 30 or more in another, which is a candour signal on the same set of published claims a buyer would otherwise rely on.

The review workforce is the substantive open question and it belongs on this axis as well as on stewardship, because the reviewer is both the correction mechanism and an additional party with access. Where that workforce sits, whether review is performed onshore, and whether every note is reviewed or a sample are all unstated, and the parent operates a separate remote scribe service which makes the location question concrete rather than hypothetical. Ask for review coverage, workforce location, and an accuracy figure with a definition.

Bidirectional EHR integration was announced in September 2024 with the claim that integration is available for all industry leading and midmarket EHR systems, positioning Speke as EHR agnostic. The breadth is asserted rather than enumerated: no named integrations, certification programmes or partner listings were verified in this pass, so this is graded on the same basis as other asserted breadth claims in this index rather than on demonstrated depth. Bidirectional write back is the meaningful part of the claim and is worth confirming per site.

The earlier assessment framed this correctly as a two part question covering both storage and workforce access. The second pass confirms that framing and fills in one half of it.

On storage nothing has changed. No hosting region, no residency option and no subprocessor list were located, and nothing establishes whether a third party model service processes the audio or what it retains.

On workforce, more is now known. The service runs in tiers and the tiers differ in who handles the encounter rather than only in turnaround. At the entry tier the software produces the note. At the higher tiers a remote scribe reviews the visit data, performs the documentation and routes it into the customer's record system, with bidirectional record integration available at those tiers specifically. So the answer to where the data goes depends on which tier is purchased, and a buyer selecting on price or turnaround is also selecting an access model without necessarily realising it.

Where that workforce sits is not published, and for a service of this kind it is a residency fact rather than an operational detail. A note reviewed and authored by a person in another jurisdiction has crossed a border whatever the servers do.

Language handling adds a further question. The platform processes audio in more than twenty five languages and the vendor states that a native speaker of the recorded language must corroborate accuracy, which implies a review capability spanning those languages. Establish who performs that corroboration and where they are located.

Ask for the hosting region, the model provider, the workforce location for the contracted tier, and what scribe access reaches inside the record.

No prices published, but more structural transparency than most: three named service tiers, Express, Plus and Pro, tell a buyer that the variable being priced is the depth of human review rather than seats or volume. That is genuinely useful for budgeting a comparison against an automated only scribe, because the cost difference is a known quantity of human labour rather than an opaque margin. Publishing what each tier includes without publishing what any of them costs is halfway to a rate card.

Among the broadest coverage in this category and unusually well spread across care settings rather than concentrated in outpatient. Roughly 80 specialties spanning ambulatory, acute and urgent care, with genuine emergency department depth inherited from ScribeAmerica's core scribe business, plus veterinary medicine in the parent's service footprint. Audio processing across 25 or more languages, with the honest caveat that output is English. Site specific templates, macros and dot phrases carry over per provider, which is the practical form specialty coverage takes for a clinician switching from a human scribe.