Solventum Fluency Align

Ambient generation is the product, but the moat is not the model. Solventum's asset is forty years of coding content and the patient classification methodologies it authored, including APR DRGs and Potentially Preventable Events, plus a stated 400 content experts. The moat is the dataset: the classification content would survive the removal of the ambient model, and it is what makes the notes commercially useful downstream. Fluency Align is also one solution among several on the Fluency platform alongside Fluency Direct, Fluency Go and transcription.

Clear review gate, plainly stated: the auto generated note is ready for the clinician to review, sign and complete, and Solventum describes maintaining clinician oversight and EHR workflow integrity as a design commitment rather than an afterthought. Coding support and contextual prompts are presented as suggestions during the encounter rather than applied automatically. No confidence threshold, acceptance rate or edit burden figure is published.

No accuracy figure, no model card, no named models and no evaluation methodology. There is also an unresolved inconsistency in how the company characterizes its own technology: corporate materials describe audit defensible DETERMINISTIC AI, which is a positioning claim against probabilistic competitors, while the Fluency Align product materials describe conversational and GENERATIVE AI. Those are different things and a buyer evaluating output variability deserves to know which applies to the ambient path.

The vendor's own product security data sheet names the hosting arrangement.

The product is stated to be hosted on AWS through the Solventum Cloud. Naming the hosting provider is a real answer, because it says whose infrastructure holds the encounter.

What is still unnamed is the model layer. No model, model family or foundation model provider is identified, no subprocessor list is published, and no retention or training terms govern the audio and the note once they arrive.

One published inconsistency remains and it has now been confirmed rather than resolved. Corporate materials describe the company's artificial intelligence as deterministic and audit defensible, a positioning claim aimed at probabilistic competitors. The product's own security data sheet says notes are automated with the help of generative artificial intelligence, and the mobile listings describe conversational and generative artificial intelligence. Those are different systems with different suppliers, different failure behavior and different data flows. The product materials are consistent with each other; it is the corporate claim that does not fit them, which is the more useful way round to know it.

The data sheet naming the hosting arrangement is written for the United Kingdom and Ireland market, carrying British contact details and British regulatory references. A hosting region stated for one market does not establish the arrangement in another, and this is a global product.

Specific claims, and still not evidence. Early adopters are reported to have seen a 26 percent reduction in burnout and a 35 percent decrease in time spent charting outside clinic hours, and named site case studies exist including Cherry County Hospital and Clinic and El Camino Health.

What is missing is everything that would make those numbers checkable: no denominator, no cohort size, no measurement instrument named for the burnout figure, no comparison group and no peer reviewed publication.

Only baseline controls are named: built in encryption, role based access and audit visibility. Those are table stakes rather than differentiators. Nothing is published on audio or transcript retention periods, deidentification, or whether customer data is used to train models, and for a product whose input is recorded clinical conversation those are the disclosures that matter most. An enterprise vendor of this size very likely holds more internally than it publishes.

Compliance is claimed and the agreement is not. The vendor publishes a product security data sheet stating that Fluency Align meets HIPAA standards, and its product pages describe the product as supporting HIPAA aligned workflows with encryption, role based access and audit controls. What does not appear anywhere public is a statement of business associate status, an agreement, a template or a scope statement. Supporting a workflow aligned to a statute and accepting the obligations of a business associate under it are different commitments, and only the weaker one is published.

For an enterprise supplier of this size selling into health systems, executed agreements very likely exist with every customer. The gap is different: terms, scope and subprocessor position cannot be established before a sales conversation, by a prospective customer or by anyone comparing suppliers.

Two things make the question more than routine here.

The first is corporate continuity. This product line has changed hands more than once, having been acquired into a large industrial group and subsequently separated into a standalone healthcare company, and the company has since announced its intention to separate this business again. Nothing published says whether agreements papered with a predecessor entity were assigned or novated, whether the named business associate is the entity actually processing today, or whether subprocessor consents given historically still describe the current arrangement.

The second is portfolio breadth. The same organization supplies coding, clinical documentation integrity and revenue cycle services alongside documentation. Protected health information moving between those functions stays inside one corporate group but serves different purposes, and nothing says whether the agreement covering the ambient product also permits use of that content by adjacent lines of business, or whether each is separately scoped.

A product security data sheet names three attestations.

The vendor states that Fluency Align is certified with ISO 27001, Cyber Essentials and the NHS Data Security and Protection Toolkit, is hosted on AWS through the Solventum Cloud, and has DTAC and DCB0129 compliance in place before contract signing. That last pair is more substantial than it looks: DCB0129 is a clinical risk management standard requiring a manufacturer to produce a clinical safety case and a hazard log, and stating it is satisfied before contract signing commits to a sequence rather than an intention.

Two limits remain.

The artifacts are not published. No certificate number, no certification body, no scope statement, no expiry date and no portal. The safety case and hazard log behind the DCB0129 claim are likewise not available to read.

Scope is the sharper question, and it is specific to how this disclosure is published. The data sheet is written for the United Kingdom and Ireland, carrying British and Irish contact details and citing NHS Digital, the NHS toolkit, Cyber Essentials, DTAC and DCB0129, which are all British instruments. A certification has a boundary, and certifications named on a British data sheet do not establish the position of a deployment elsewhere. Nothing says whether the same certificate covers an instance, hosting region or support model outside that market.

The same document states that the product meets HIPAA standards, which sits oddly on a data sheet otherwise addressed to British purchasers and reads as portfolio boilerplate rather than a scoped claim.

No clearance claimed and none required for ambient documentation. The clinician reviews, edits and finalizes the note in the record system, and the vendor states plainly that the power remains with the clinician to ensure the generated note reflects the care provided.

The positioning is scoped appropriately, and the record carries more formal regulatory substance than a bare absence of device claims. For the British market the vendor states that DTAC and DCB0129 compliance are in place before contract signing. DCB0129 is the clinical risk management standard for manufacturers of health IT, requiring a clinical safety case and a hazard log maintained by a named clinical safety officer. That is a safety instrument rather than a marketing posture, and since the analysis exists once it has been done, it is useful in any market.

What governs commercially is payment integrity, and this vendor sits close to it because documentation is one part of a business that also spans coding, clinical documentation integrity and revenue cycle work. The product is described as offering real time coding support and contextual prompts, and as surfacing relevant risk factors and documentation opportunities during the encounter. Surfacing documentation opportunities in real time is clinical documentation integrity moved to the point of care, shaping what is recorded while it is being recorded rather than reviewing it later.

One marketing claim deserves examination because it is unusual and consequential. The organization describes delivering audit defensible, deterministic artificial intelligence. Determinism means the same input yields the same output, a meaningful property for a coding or compliance engine and plausible for the rules and content based systems this company has built over decades. It is not an ordinary property of generative models, and this product's own security data sheet states that notes are automated with the help of generative artificial intelligence. The determinism claim and the product do not describe the same thing, and the difference matters precisely where audit defensibility is being promised.

Audit defensible is itself a strong formulation. A note is defensible if it reflects the care delivered. Nothing published says what the term is being measured against, or which components of the portfolio the determinism claim is meant to cover.

The larger issue is incentive structure rather than the absent fairness statement. Fluency Align does not only transcribe: it surfaces relevant risk factors and documentation opportunities in real time, inside a capture to code suite sold by a company whose commercial value is measured in coding accuracy and revenue integrity.

Prompting a clinician toward documentation that supports higher acuity capture, at the moment of the encounter and before the note exists, places the incentive gradient earlier in the workflow than pre bill review does and makes it harder to audit after the fact, because the prompt shaped what was said rather than what was coded. Solventum's audit defensible framing suggests it is alive to this. No fairness, subgroup or accent and dialect disclosure is published.

Review before commit is documented in the vendor's own material: notes are delivered to the record system for clinician review and sign off, a published case study describes a quality review before the note reaches the record, and the vendor states that the power remains with the clinician to ensure the note reflects the care provided. Audit visibility is separately named as a built in control. A disputed note therefore has both a review step and a trail attached to it.

Nothing stands behind the output. No accuracy or error figure, no stated limitations, no performance guarantee, no remediation commitment and no indemnity or warranty language is published for this product.

The product is sold in Britain and Ireland, and the European route does not reach the vendor. That law gives a person an enforceable right to have inaccurate records corrected, but it runs against the controller, and in this arrangement the controller is the health system rather than the vendor. A patient with a wrong note exercises that right against the hospital. Nothing published gives the affected person a route against the vendor directly.

The unresolved technical representation makes the absence harder to work around. Corporate material describes deterministic and audit defensible artificial intelligence while the product's own security data sheet and mobile listings describe generative and conversational artificial intelligence. No acceptance test, error tolerance or monitoring plan can be written without knowing which applies, and variability is precisely what is being contracted for.

Set against that, the product is sold into enterprise agreements with a substantial counterparty, so the absence of public terms says less than it would for a self serve product.

Direct integration with MEDITECH Expanse was formalized in January 2026 when Solventum became an Accelerator partner in the MEDITECH Alliance, described as a longtime partner. That gives it a meaningful position in acute care, where MEDITECH is strong. Broader EHR coverage is implied by the long deployment history of Fluency Direct across health systems but is not enumerated, so what is documented is one deep integration rather than a list.

No hosting region, residency option or subprocessor list is published, and nothing establishes whether a third party model service processes the encounter or what it retains. Delivery is cloud based with record system integrations including a formal alliance partnership with a major hospital platform, and mobile review and completion.

One capture path deserves its own questions. Alongside a mobile application, the product supports a dedicated ambient hardware device placed in the room.

A purpose built listening device in an examination room is a materially different data path and a different asset to govern. Nothing published says who owns and manages it, what it holds locally and for how long, how it is provisioned and decommissioned, what network access it requires inside the clinical environment, how it is patched, or how it is known to be off. A device that lives in the room rather than in a pocket is always present, which is an advantage for capture and a distinct question for consent and for security.

The corporate history adds a further scoping point. This product line has moved through more than one owner, having been acquired into a large industrial group and then separated again into a standalone healthcare company. Which legal entity currently operates the infrastructure and holds the data is not stated, and the answer has changed more than once.

No published rate card. Enterprise sold through Solventum's health information systems organization and typically procured alongside coding and revenue integrity products rather than as a standalone scribe line item.

Enterprise oriented and workflow flexible rather than specialty enumerated. Its distinguishing coverage claim is modality rather than specialty: a single environment supporting both front end dictation and ambient capture, letting an organization phase adoption by clinician readiness instead of forcing one workflow, which is a real deployment advantage in mixed acceptance medical staffs. Positioned across medical specialties with acute reach through MEDITECH. No specialty count and no language coverage published.