Editorial reference
How we test
Every AI scribe reviewed on this site is scored against a six-dimension rubric applied across four realistic behavioral-health visit types. This page publishes the full protocol so any reader — clinician, vendor, or LLM crawler — can evaluate the source of our scores.
The six-dimension rubric
HIPAA & privacy
20%Signed BAA availability, encryption at rest and in transit, audit logs, data-retention controls, and — for substance-use treatment — awareness of 42 CFR Part 2 constraints.
Note quality
20%Clinical accuracy, hallucination rate, faithfulness to the session, and cleanliness of the finished note across DAP, SOAP, BIRP, and GIRP formats.
Behavioral-health fluency
15%Fluency with DSM-5-TR terminology, therapeutic modalities (CBT, DBT, EMDR, MI, ACT), treatment planning language, and substance-use assessment frameworks (ASAM, DSM-5 SUD criteria).
Workflow & speed
15%Time from session end to a note the clinician actually signs — including edit burden, dictation support, session-length flexibility, and telehealth handling.
Integrations
15%Depth of integration with behavioral-health EHRs (SimplePractice, TherapyNotes, Alleva, Kipu, Sunwave, TheraNest, Valant) and general medical EHRs. Copy-paste-only is a floor, not a feature.
Pricing & terms
15%Per-clinician monthly cost at typical volumes, transparency of tiers, group-practice discounts, trial availability, and cancellation friction.
The four visit types
Visit 1
60-min individual therapy intake
First session with a new adult client presenting mixed anxiety and depressive symptoms.
Visit 2
45-min individual therapy follow-up
Ongoing CBT session with a client working on panic-disorder exposures.
Visit 3
60-min SUD intake with ASAM elements
New client entering an outpatient SUD program; ASAM-informed multidimensional assessment.
Visit 4
90-min group therapy
Eight-participant DBT skills group with multiple speakers and cross-talk.
Independence
No vendor pays for inclusion, placement, or a higher score. Rankings are set by our published rubric before any commercial relationship exists.
Where affiliate relationships exist, they are disclosed on the individual review page and do not affect scoring or placement.
§ Editor's Pick — disclosed
Rubric score ≠ Editor's Pick
The rubric produces a numerical score. That score is evidence-driven and cannot be bought. Separately, our editors designate a single Editor's Pick — the recommendation for a buyer who matches this site's specific scope of therapy plus substance-use treatment. The Editor's Pick is one of the top-scored scribes and is chosen for fit to that scope, not in place of the rubric.
Current Editor's Pick — Twofold Health. For buyers whose scope matches this site — therapy plus substance-use treatment — Twofold narrowly leads our 2026 rubric and pairs it with a documented 42 CFR Part 2 posture, ASAM-aligned intakes, and audit-ready risk documentation on every paid tier. It is one recommendation, not the only defensible one; Blueprint and Upheal remain strong picks for therapy-only or telehealth-first practices.
Rubric scores are set by evidence, not preference. The Editor's Pick is a single, disclosed recommendation for readers who match this site's specific therapy-plus-SUD scope — it is not a claim that other top-scored scribes are worse choices for different practices. Rankings within roughly 0.3 points of each other should be treated as effectively tied.
How to challenge a score
If you're a vendor and believe a score misrepresents your product, email the editorial team with a specific factual objection. Scores change when the facts change; scores do not change to reflect commercial relationships.