Sayvant

Ambient generation is the whole product and the company exists only to do it. Note for completeness rather than as a deduction: the underlying generative capability runs on Microsoft Azure OpenAI Service, so the differentiated work is the emergency medicine specific reasoning, note structure and clinical content layered on top rather than a foundation model built in house.

Draft and review, with the platform generating charts that providers review before signing, and vendor materials describing preservation of clinician control as a design goal. The HITLAB evaluation assessed error prevention and system visibility among its criteria, which is more external scrutiny of the interaction design than most vendors here invite.

Held at B because no acceptance rate, edit burden figure or confidence threshold is published, and because the oversight question is sharper here than usual given what the system is designed to construct.

Names its own infrastructure, which is rarer than it should be: Sayvant states it is built on Microsoft Azure OpenAI Service, letting a buyer reason about the underlying model family and its known behaviour rather than treating the system as a black box of proprietary AI. It also describes its design philosophy explicitly rather than in adjectives, stating that most ambient tools capture what was said while Sayvant generates what the note needs to reflect. Held at B because no accuracy figure, model card, hallucination rate or evaluation methodology was located.

One of the few vendors in this lane that names the model layer at all. The company states it is built on a named cloud provider's hosted model service, which identifies both where the workload runs and which model family sits behind it, so a buyer can reason about known behaviour, published limitations and the provider's own commitments rather than treating the system as proprietary artificial intelligence.

That is exactly the disclosure this axis asks for and most competitors decline to make. Held below the top grade because the naming stops at the service. No sub processor list was located, no model version or update policy is published, and nothing states which other parties touch encounter content. The setting makes one of those questions sharper than usual.

This is an acute care product supporting a clinician running several patients concurrently across multiple sessions, so the volume and sensitivity of content moving through the chain is high and the retention question at the model service layer is a live one rather than theoretical. Ask for the sub processor list, confirm which model deployment and region are used, and establish what the hosted service retains under the vendor's own configuration.

The most ambitious study design in this category, and held at B for who ran it rather than how it was built. Published June 2026, it analysed over 250,000 encounters across 50 emergency departments against a historical baseline of more than 20 million encounters spanning three years, at a scale intended to control for the confounders that drive variation in emergency medicine, and it is the first in this category to take patient throughput and professional fee capture as primary outcomes rather than charting time.

That is a genuinely better question than most of the literature asks. But the study was conducted and published by the vendor, released through a press distribution service rather than a journal, with no peer review located and with outcome measures chosen by the party selling the product. Independent work exists alongside it: a HITLAB heuristic evaluation white paper with named academic authors, and a KLAS Spotlight. Peer reviewed publication of the multicentre analysis would move this to A.

No statement on audio or transcript retention, de identification or training use was located, and those are the disclosures this axis measures. The security page is substantive on controls and silent on stewardship, which is a common split and worth separating: how well a vendor protects data is a different question from how much of it they keep and what else they do with it.

Two features of the acute care setting make the omissions more pointed here than for a clinic product.

The environment is one the vendor itself identifies as uniquely hard for ambient capture. An emergency department is loud, and a clinician may be running several patients at once, returning to each across multiple sessions that accumulate into one chart. The vendor supports that concurrency deliberately. It also means several patients' recorded speech is open simultaneously under one clinician's session, which raises an attribution question of the same kind seen in institutional care: what prevents content from one encounter reaching another patient's chart, and what does the clinician confirm about which chart they are adding to.

And emergency departments treat patients who cannot consent. Unconscious, intoxicated, confused or acutely distressed patients are recorded in the ordinary course of the work, and psychiatric presentations and intoxication are documented in the same note. Nothing published addresses how consent is handled where it cannot be obtained, or whether retention differs for those encounters.

A third party account states the platform runs on a named cloud and model provider. That was not confirmed on the vendor's own material and is not graded on, but it makes the subprocessor retention question concrete.

Ask for the retention schedule, the training position, and the consent approach for incapacitated patients.

No published business associate agreement posture, template or scope statement was located. The vendor states it meets or exceeds the administrative and technical controls required by the health privacy and health technology statutes, which describes safeguards rather than the contractual relationship this axis measures.

Agreements plainly exist. The product is live across dozens of hospitals and health systems and the vendor states it has been vetted and approved at some of the largest systems in the country. That vetting is meaningful context and is not a substitute: a health system's security review examined this vendor against that system's own requirements and produced a private conclusion, which tells a different prospective buyer nothing they can act on.

One scope question follows from the vendor's stated direction and is worth asking now rather than later. The product is described as expanding beyond acute care into hospice, wound care and nursing. Those are post acute settings governed by different documentation regimes and, in the case of hospice, by a distinct payment and eligibility structure. An agreement written for emergency department documentation may not anticipate them.

A second follows from the partnership model. The vendor describes working with revenue cycle and clinical documentation improvement partners to deliver recommendations at the point of care. Where a partner's system contributes to or receives content derived from an encounter, a buyer should establish whether that partner is a subprocessor under this vendor's agreement or a separate counterparty they are contracting with unknowingly.

Ask for the agreement, its scope across settings, and the position on revenue cycle partners.

The earlier assessment located no attestation. That is overturned, and this is one of the more substantive security disclosures in the lane.

The vendor publishes a dedicated privacy and security page stating an AICPA SOC 2 Type 2 certification alongside regular internal and third party audits, continuous automated security testing including static and dynamic analysis and vulnerability scanning, and third party penetration testing.

Two things lift this above a certification claim. Continuous testing and independent penetration testing are ongoing activities rather than a point in time examination, and naming them separately from the attestation shows the vendor understands they answer different questions. And the vendor states its security, privacy and risk programmes align with the United States national framework for AI risk management alongside the cybersecurity framework. That first reference is rare in this category and worth crediting specifically, because the AI risk framework addresses the failure modes an attestation does not reach: model behaviour, evaluation, monitoring for drift and the governance of systems whose output changes over time. A SOC 2 report examines whether controls operated; it does not ask whether a model still performs as it did at deployment.

Access control is also concrete rather than described. The platform integrates with hospital identity providers for single sign on, so authentication runs on the institution's own credentials and deprovisioning follows the institution's process.

What would move this to the top grade is specificity: the report date and audit period, a scope statement, and the alignment claim evidenced rather than asserted, since alignment with a framework is a self assessment unless someone has examined it.

No clearance claimed and none required. The earlier assessment framed the regulatory context correctly as billing rather than device, with the 2023 revisions to medical decision making documentation requirements as the stated driver. The second pass finds a described capability that is the sharpest version of a pattern this lane keeps circling, and it belongs on the record in full.

The company's chief executive describes the product as going beyond transcription to write out the actual thinking of the physician, including when the physician did not articulate it explicitly, naming differential diagnosis and why a course of treatment was chosen or rejected, and connecting this to charge capture accuracy.

Two vendors elsewhere in this lane generate the written justification alongside the billing code they select, which removes the justification's function as an independent check. This goes a step further. Here the system infers the clinical reasoning itself, records it as the physician's, and the value proposition is explicitly that documented reasoning supports the billing level, because medical decision making complexity is what determines that level.

The consequence sits with the clinician rather than the vendor and should be stated plainly. A physician signing that note is attesting to a differential they may not have formed and to alternatives they may not have weighed. If the reasoning is accurate it is a genuine improvement, since unarticulated thinking is real and goes undocumented constantly. If it is inferred incorrectly the record contains a fabricated account of the physician's judgement, attributed to them, in the document a payer audit or a malpractice claim will examine.

Ask what the inference is grounded in, whether inferred reasoning is visually distinguished from captured speech at review, and what the acceptance rate is on those passages specifically.

Graded on incentive structure, following the coding drift reasoning applied in this index to Charta Health, Candid Health and Solventum, and Sayvant states the position more plainly than any of them. Its clinical AI lead describes an emergency department note as a structured argument for medical necessity and says that where most ambient tools capture what was said, Sayvant generates what the note needs to reflect.

Its own headline study takes professional fee capture as a primary outcome, and its chief executive summarises the result as charges up and wait times down. A documentation system explicitly optimised to construct the justification for billing level, and marketed on the resulting revenue lift, carries an upcoding gradient that a compliance function must audit independently, because the tool and the clinician now share an interest in a more thoroughly justified chart.

Credit where it is due: saying this out loud is more candid than competitors who market the same capability as completeness. No fairness, subgroup or accent disclosure was located despite support for more than 30 languages.

Two passes located no accuracy or error figure, no hallucination rate, no published limitations and no warranty, indemnity or remediation commitment. Two features of the acute setting create exposures with no described control, and both belong on this axis rather than only on stewardship. The first is attribution under concurrency.

A clinician may be running several patients at once, returning to each across multiple sessions that accumulate into one chart, and the vendor supports that deliberately. Several patients' recorded speech is therefore open simultaneously under one clinician's session, and nothing published describes what prevents content from one encounter reaching another patient's chart, or what the clinician confirms about which chart they are adding to.

Documentation attached to the wrong patient in an emergency department is among the most consequential errors available. The second is consent where it cannot be given. Emergency departments treat patients who are unconscious, intoxicated, confused or acutely distressed, and those encounters are recorded in the ordinary course of the work, with psychiatric presentation and intoxication documented in the same note.

Nothing published addresses how consent is handled where it cannot be obtained, whether retention differs for those encounters, or what a patient is told afterwards. The recorded person in that situation is the least able of any in this index to object at the time and has no route to raise it later. Ask for the attribution control, the consent approach for incapacitated patients, and an accuracy figure.

The earlier assessment found no documented integration and flagged that as the first thing to establish for a product live at seventy sites. It is now answered, and the answer is among the better ones in this lane.

The vendor has a standards based integration with the major record system built on the open application interface specification for health data, recognised through that vendor's own partner showroom programme, and states integrations with three further major record systems. Access runs through hospital identity providers by single sign on.

That matters because of what it is not. Six vendors in this lane reach the chart by driving the user interface, through browser extensions or robotic process automation, which means their actions inherit a clinician's session and appear in the audit log as that clinician typing. A standards based integration is permissioned, carries its own identity, and is constrained by the access controls the institution already operates. It is also reviewable: an institution can see what scopes were granted.

Two things hold it short of the top grade. The depth of write back is not described per system, so whether the note lands in structured fields or as narrative text is unstated and is likely to differ across the four. And integration is described as configured at each clinical site, which suggests per site work rather than a uniform capability, so what a given hospital gets may not match what the partner listing implies.

Ask which systems have bidirectional write back today, whether structured data flows or only the note, and what the integration looks like at a site running one of the three unlisted systems.

Partial. The hosting substrate is disclosed, running on Microsoft Azure OpenAI Service, which is more than most vendors in this category reveal and lets a buyer inherit Azure's regional and compliance story as a starting point. Nothing further was located on data residency configuration, regional deployment or sub processors.

No published rate card. Sold to emergency department groups and hospital systems. Note that the business case Sayvant itself advances is revenue rather than cost: its published outcomes are charge capture and throughput, so the return argument rests on professional fee capture increasing rather than on documentation labour falling, which is a materially different justification to take to a compliance committee.

Deliberately narrow, and the narrowness is the product rather than a limitation. Coverage runs across the acute care spectrum from freestanding emergency departments to Level 1 trauma centres, with urgent care alongside, and support for more than 30 languages. Emergency medicine imposes constraints most scribes are not built for, including encounter pace, disposition decisions and a documentation standard tied to medical decision making complexity.

For an emergency department buyer this is the closest fit in the index; for anyone documenting scheduled outpatient visits it is not a candidate. Graded B on breadth, which is the axis, rather than on fit.