Blueprint

The AI Assistant is the product and the company is explicit that this is where the value concentrates: the 2026 repositioning made the EHR free and kept the AI layer paid, which is about as clear a statement of where the intelligence sits as a pricing model can make. No services layer underneath.

Draft-and-approve by design. Output is explicitly a draft the therapist must read, edit and approve before it enters the clinical record, and third party reviews are clear that this does not remove documentation responsibility. The 30 second note generation claim is about draft speed, not finality.

Graded B rather than A because unlike Eleos, which publishes an approximate 80 percent suggestion acceptance rate, Blueprint publishes no edit-burden figure, so how much correction a therapist actually performs is unquantified. In a category where the whole value claim is time saved, the post-generation edit rate is the number that matters and it is absent.

Little technical disclosure. Located material describes natural language processing generating notes and treatment plans but names no base model, no training data provenance, and no accuracy methodology. Accuracy is evidenced through clinician testimonials rather than measurement. Compare Eleos in the same category, which publishes its micro-dialogue evaluation approach in peer reviewed venues.

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, and no published position was found on whether session recordings or generated notes are used to train models. The absence of a training position is the material gap here rather than a routine omission, and it is a live question rather than a hypothetical one.

Therapists commenting publicly on this company's own marketing have asked directly whether their notes train models that could eventually displace them, and consent, presence and data use are recurring themes in that discussion. When the people who would be asked to adopt a product have already raised the question in public, publishing an answer is the minimum, and one peer in this same segment has published exactly that answer in one sentence.

The content makes it heavier still: a recorded psychotherapy session is among the most protected material in healthcare, and a client consenting to recording for their own care has not consented to becoming training data for a commercial product. Ask whether customer sessions or notes are used for training or improvement, in contract language, and ask for a sub processor list.

The central gap on this record, and the sharpest contrast with Eleos Health. No peer reviewed study, controlled trial or published accuracy evaluation was located. Evidence consists of adoption scale, more than 70,000 clinicians, and extensive user testimonial.

Adoption at that scale is a meaningful market signal and should not be dismissed, but it is not evidence of clinical benefit, and the index applies the same standard here it applied to Infervision: scale of use standing in for evidence of benefit earns a C. A measurement based care and AI progress tracking feature exists, which makes the absence of published outcome data more conspicuous rather than less, since the product is already computing symptom change.

The company states HIPAA compliance, but no published position was located on whether session recordings or generated notes are used to train models. This is a live and publicly raised concern rather than a hypothetical one: therapists in public commentary on the company's own marketing have asked directly whether their notes train models that could displace them, and consent, presence and data use are recurring themes in that discussion. For a product recording psychotherapy sessions at this scale, an explicit published training-data position is the minimum, and its absence is graded.

HIPAA compliance is stated prominently in product marketing, which is more than several peers in this batch offer. Graded B rather than A because no BAA terms, execution process or compliance documentation were located, and a stated compliance claim is not the same as a published agreement.

Converted from Not Rated. No independent attestation was located at review, and the distribution model makes the gap consequential in an unusual way.

No SOC 2 of either type, no HITRUST, no ISO 27001, no trust centre and no penetration testing statement was retrieved.

Most records in this pass involve institutional buyers with procurement functions that will demand this material privately whether or not it is published. This product is different. It is sold self serve to individual therapists and small practices, with more than 70,000 mental health professionals reported on the platform. A solo clinician adopting a documentation assistant has no security team, no vendor risk process and no realistic ability to assess a third party holding their session content. They rely entirely on what is published, and nothing is.

The content is among the most sensitive any vendor in this index handles. The assistant listens to therapy sessions and produces progress notes, treatment plans and psychotherapy notes. Psychotherapy notes carry heightened protection under the health privacy rule specifically because of their content, requiring separate authorisation for most disclosures.

That combination, unusually sensitive material and unusually unprotected buyers, is the reason to press rather than the company's size.

Ask for the attestation, its type and period, what session audio and transcript retention applies, and whether content trains models.

Converted from Not Rated. The prior scoping was correct and had no grade attached.

No device pathway applies. Documentation drafting and workflow support fall outside software as a medical device, and the product neither diagnoses nor directs treatment. Same treatment as the other behavioural health documentation vendors in this index.

What governs instead is threefold and none of it is addressed publicly. State licensure and scope of practice rules bear on what a therapist may delegate to an automated system and what must carry their own judgement. Payer documentation requirements determine whether a generated note supports the billed service. And 42 CFR Part 2 imposes stricter confidentiality on substance use disorder records than the general health privacy rule, requiring specific patient consent for disclosures the rule would otherwise permit, with an enforcement date reached in February 2026.

The Part 2 question is the live one for this product and it is unaddressed. The platform is sold to individual therapists and small practices at very large scale, more than 70,000 professionals reported, and a substantial share of independent mental health practice involves substance use treatment. A self serve product adopted by a solo clinician has no procurement process to catch the question, which makes publishing a position more important here rather than less.

Ask for the written Part 2 position, and whether note types or templates distinguish records that carry it.

No governance framework, subgroup analysis or bias evaluation located. The exposure mirrors Eleos: speech-to-text and language models applied to psychotherapy sessions face documented performance variation across accent, dialect and atypical speech, and the population includes clients whose speech is affected by symptoms or medication. Additionally, a product that drafts treatment plans and suggests interventions is shaping clinical content, not just transcribing it, which raises the governance stakes above pure documentation.

Two passes located no accuracy or error figure, no published limitations, no evaluation methodology and no warranty, indemnity or remediation commitment. What stands in place of a measurement is a claim shape worth naming, because it appears frequently in this segment and is weaker than the other forms this index tracks. Accuracy is evidenced through clinician testimonials.

A testimonial is selected by the vendor from a population the vendor knows, it reports a person's impression rather than a comparison against a reference, and it cannot be wrong in any way a reader could establish, so a wall of them provides no information about the rate at which the system fails. It also sits oddly against a peer in the same category that publishes its evaluation approach in peer reviewed venues, which shows the segment is capable of better.

The consequences here are heavier than in general documentation, since a psychotherapy progress note supports a clinical formulation, is submitted for reimbursement and is discoverable in later proceedings. Ask for a measurement rather than a testimonial, for how the system handles disclosed risk content, and for what the vendor commits to when a note is wrong.

Strong, and structurally similar to Eleos: a browser extension plus EHR integrations lets the Assistant run inside whatever system a practice already uses with no migration required, and a one-click note transfer to the EHR removes the copy-paste step that undermines most scribe workflows. The company also ships its own AI-native EHR for practices that do want to switch, so it serves both sides of that decision rather than forcing it. Offering a genuine no-migration path and a full replacement is unusual.

Broad and low friction: browser extension, web platform, and native iOS app allowing phone-based recording, supporting in-office and telehealth sessions, with no-credit-card trial access. Practical for solo practitioners, which is the target buyer. No data residency or hosting architecture disclosure located.

The strongest axis on this record and the inverse of Eleos. Pricing is published rather than gated: a free EHR tier as the on-ramp, with the AI Assistant sold as a usage-priced paid layer, and trial access without a credit card. A prospective buyer can determine what they will pay and start without contacting sales, which almost no behavioral health vendor in the index permits.

Note the honest framing of the model itself, disclosed in third party review: the free EHR is genuinely free but functions as the on-ramp to the paid AI layer where the value concentrates. That is a legible commercial structure, not a one.

Covers individual, couples, family and group therapy plus psychiatry workflows, with note types tailored to each, including a couples SOAP structure that generates a note per client alongside a relational process note. That is genuine modality depth.

Graded B rather than A because coverage is outpatient psychotherapy and psychiatry; it does not extend to substance use disorder treatment, case management, field based outreach, home health or hospice the way Eleos does, and it is oriented to solo and small practice rather than enterprise behavioral health organisations.