AI Medical Scribe for Therapists & Counselors

Therapists and counselors carry a documentation burden that other specialties rarely match. The work isn't just clicking checkboxes after a fifteen-minute primary care visit — it's reconstructing a fifty-minute session in narrative prose, attributing statements correctly across two or more speakers, and writing in a voice that holds up to chart audits, supervision review, and insurance scrutiny. Scriber is built for that work. This page explains what we do for individual, couples, family, and group therapy, what we capture in person and over Doxy.me, the note formats we generate by default, how we handle privacy and BAAs, and the specific questions therapists ask before they trust an AI scribe with a session.

The therapist documentation burden

Most therapists finish notes after the clinic day, not during it. A clinician who sees six to eight clients a day commonly spends one to two additional hours that evening writing progress notes — sometimes more during intake-heavy weeks, sometimes still unfinished by Friday. That math is the documentation burden in one number. It is the reason therapists stop accepting insurance, cap their caseloads, lose evenings, and burn out.

The burden is not driven by any single payer or any single rule. It is the cumulative effect of several requirements layered on top of each other. Insurance reimbursement requires session-by-session medical necessity language tied to a treatment plan. State licensing boards require records that meet retention and content standards that vary by jurisdiction. Supervisors expect notes that demonstrate clinical reasoning, not just session summaries. EHRs designed for general medicine often force therapists into checkbox templates that don't fit how a therapy session actually flows. Each requirement is reasonable on its own. Stacked, they consume an hour or two every clinic day.

Therapists have tried the obvious workarounds. Some chart between sessions in the ten minutes between clients — but ten minutes is not enough to reconstruct fifty minutes of clinical content while a new client is waiting. Some use dictation tools — but raw dictation still requires typing, restructuring, and editing into a payer-acceptable note format. Some skip detail and write the same generic paragraph for every session — which works until an audit asks why every note for every client says the same thing.

An AI scribe collapses the work without collapsing the quality. The clinician records the session with consent, the scribe transcribes and structures the content into the requested note format, and the clinician reviews and edits a draft instead of starting from a blank page. The reduction in time isn't because the AI is doing clinical thinking — it's because the AI is doing the mechanical reconstruction work that costs a therapist an hour every evening. The clinical judgment, the formulation, the diagnostic reasoning, the language about therapeutic alliance — those still come from the clinician, applied during the review pass.

What Scriber captures during a therapy session

Scriber works in two modes that matter for therapy: in-person and telehealth. In both, the goal is the same — capture the session faithfully enough that the generated note reflects what actually happened, not a generic template.

For in-person sessions, the clinician opens Scriber on a phone, tablet, or laptop in the room, gets verbal consent on the record, and starts the session normally. Audio is captured locally and streamed to the transcription engine. Scriber supports seven STT engines so the clinician can choose the trade-off that fits the practice — Azure Speech for the broadest medical vocabulary coverage, Speechmatics for diarization quality on multi-speaker sessions, AssemblyAI and Deepgram as alternates, and an xAI fast engine that returns transcripts in roughly three seconds for clinicians who want to review the transcript before the next client walks in. The session ends, the audio stops, the transcript appears, and the note generates from the transcript and the chosen template.

For telehealth, Scriber has a direct integration with Doxy.me. The clinician runs the session in their normal Doxy.me waiting-room workflow. Scriber captures the session audio through the Doxy integration, not through a generic browser screen recorder, which means audio quality is consistent and the capture does not depend on the clinician remembering to start a separate recorder. The transcript and note generation work the same way they do for in-person sessions. This is the differentiator most other AI scribes can't claim — they ask you to run a generic browser tab recorder alongside your telehealth platform, which works but is fragile and easy to forget. Scriber and Doxy.me run together.

In both modes, the diarized transcript distinguishes speakers. That matters for therapy more than it matters for primary care. A therapy note that attributes a client's suicidal ideation statement to the clinician, or that records the clinician's reflection as a client statement, is not just embarrassing — it is clinically dangerous and a defensible-charting problem. Diarization gives the note generator a structured "Client said X / Clinician said Y" stream to work from, so the generated note attributes statements correctly even when both speakers are talking quickly or speaking over each other in a couples or family session.

Note formats Scriber generates

Therapists and counselors use a wider range of note formats than most other specialties. Scriber generates each of these by default, with no template configuration required.

SOAP

Subjective, Objective, Assessment, Plan. The standard medical note format adapted for behavioral health. Subjective captures the client's reported experience and presenting concerns. Objective captures the clinician's observations of affect, behavior, and mental status. Assessment captures clinical formulation and progress toward goals. Plan captures interventions used, homework assigned, and the next session's focus. SOAP is the format most insurance carriers default to and the format most EHRs structure their note fields around.

DAP

Data, Assessment, Plan. A condensed three-section format that some payers and many group practices prefer. Data combines what SOAP splits into Subjective and Objective — what the client said and what the clinician observed. Assessment and Plan track SOAP. DAP is faster to write and faster to read; it works well for ongoing therapy where the intake history is already on file.

BIRP

Behavior, Intervention, Response, Plan. The dominant format in community mental health and many SUD programs. Behavior captures what the client did or reported. Intervention captures what the clinician did — the therapeutic technique applied, the topic addressed, the skill taught. Response captures how the client responded to the intervention. Plan captures next steps. BIRP forces the clinician to demonstrate that an intervention happened and that the client engaged with it, which is exactly what Medicaid and many state mental-health funders audit for.

GIRP

Goal, Intervention, Response, Plan. Used in programs that anchor every progress note to a specific treatment-plan goal. Goal names the treatment-plan goal addressed in the session. Intervention, Response, and Plan track BIRP. GIRP makes the medical-necessity link to the treatment plan explicit on every note.

SIRP

Situation, Intervention, Response, Plan. A variant common in case-management and wraparound contexts. Situation describes the presenting issue or context for the session. The remaining sections track BIRP/GIRP.

Intake

Initial assessment notes. Scriber generates a structured intake covering presenting problem, history of present illness, past psychiatric history, medical history, family history, social history, substance use, trauma history, mental status, risk assessment, diagnostic impression, and initial treatment recommendations. Intakes are the longest notes most therapists write — and the ones most worth automating.

Biopsychosocial

A more comprehensive intake format used by LCSWs, LMFTs, and many community-mental-health programs. Covers biological, psychological, and social domains in depth, including developmental history, cultural identity, spirituality, strengths, and supports. Biopsychosocials are often required for program admission and care-plan approval.

Treatment plan

Goals, objectives, interventions, target dates, and progress measures. Scriber generates an initial treatment plan from the intake content and updates it from progress notes when the clinician asks for a treatment-plan refresh. Treatment plans drive medical-necessity language on every subsequent progress note.

Mental status exam

Appearance, behavior, speech, mood, affect, thought process, thought content, perception, cognition, insight, and judgment. Scriber generates an MSE from session content where it is clinically grounded, and clearly leaves a section blank rather than fabricating findings the audio doesn't support.

Custom templates are supported on top of these defaults. Group practices that have standardized on a specific note structure — or individual clinicians who want their own voice consistently applied — can configure a template once and have every session generate against it.

Telehealth therapy with Doxy.me + Scriber

Doxy.me is the most-used HIPAA-compliant telehealth platform in independent behavioral-health practice. It works because it is browser-based, requires no client install, and stays out of the clinician's way. The trade-off — until recently — was that AI scribes had no native way to capture a Doxy.me session. Clinicians worked around it with browser tab recorders or by holding a phone up to the laptop speaker. Both work. Both are fragile.

Scriber has a direct integration with Doxy.me. The session captures through the integration, not through an ad-hoc workaround. The clinician opens Doxy.me as usual, starts the session, and Scriber captures cleanly with no separate recorder to remember. When the session ends, the transcript and note appear in Scriber the same way they appear after an in-person session. The clinician reviews, edits, copies the finalized note into the EHR, and moves on.

This is the differentiator most other AI scribes can't claim. A search for "AI scribe for Doxy.me" returns a long list of products that document support for Doxy.me by listing it in a marketing table. Most of them mean "you can run our generic recorder while Doxy.me is open in another tab." Scriber means "we built the integration with Doxy and tested it." For a clinician running a half-telehealth, half-in-person caseload — which is most independent behavioral-health practice in 2026 — that distinction is the difference between an AI scribe that works on your in-person days and falls over on your telehealth days, and one that works for the whole week.

Modality-specific notes

Therapy is not one modality. The same client may receive CBT one week and EMDR the next, or be in family therapy on Tuesday and individual on Thursday. Scriber generates notes that reflect the modality used, not a generic progress-note template.

For CBT, the note captures the cognitive distortion identified, the thought record completed, the behavioral experiment assigned, and the homework reviewed from the prior week. For DBT, the note captures the skill module addressed (mindfulness, distress tolerance, emotion regulation, interpersonal effectiveness), the diary card review, and the chain analysis if one was completed. For EMDR, the note captures the target memory, the SUD and VOC ratings before and after, the bilateral-stimulation set used, and the closure or incomplete-processing status — without inventing ratings the audio doesn't contain.

Couples and family therapy notes attribute statements correctly across speakers, capture the systemic dynamics observed, document the intervention applied (Gottman, EFT, structural, narrative, or eclectic), and note each participant's response separately. Group therapy notes follow the relevant payer format (often a group BIRP or progress note per participant) and capture the group theme, individual participation, and member-specific clinical content where required. Play therapy notes capture the play themes observed, the materials used, the symbolic content, and the parent-consultation portion if one occurred.

Modality-specific generation works because the clinician indicates the modality at session start, or Scriber infers it from the session content, and the prompt that drives note generation adapts accordingly. The clinician is not stuck with a one-size-fits-all template that flattens a DBT session and a couples session into the same shape.

HIPAA, BAA, and therapy privacy

Therapy notes are PHI under HIPAA, and therapy psychotherapy notes are subject to additional protections under 45 CFR 164.508. State law often adds further restrictions, especially for minors, substance use treatment (42 CFR Part 2), and reproductive health. An AI scribe operating in this space has to do the privacy work right or it has no business being in the room.

Scriber signs a BAA at signup. Not enterprise-gated. Not a sales-team-only artifact. Every clinician who creates a Scriber account gets a BAA in the signup flow. Audio is never retained. The audio file is processed for transcription and discarded — not stored, not retained for "model improvement," not held in a logs bucket. Transcripts and generated notes are encrypted at rest with AES-256 and in transit with TLS, hosted on Azure under a Microsoft-signed BAA. Access is audited; the clinician can pull an access log for their own records. There is no PHI in the browser cache. There is no PHI in the prompt logs. The architectural choices that enable this are documented on the security page; the short version is that the design starts from the assumption that any PHI exposure is a HIPAA breach and works backward from there.

State minor-consent rules vary, and Scriber does not pretend to be a substitute for the clinician's own legal review. Some states allow minors to consent to outpatient mental-health treatment without parental involvement starting at age 12, 13, or 14; others require parental consent through age 18. The same variation applies to recording consent. Scriber gives clinicians the controls they need — record consent prompts, the ability to disable recording for specific clients, the ability to delete a transcript on request — and leaves the legal determination to the clinician and their compliance counsel. The technical posture is restrictive by default: no audio retention, encrypted transit and storage, BAA-covered processing.

Common therapist questions

A few questions come up in almost every conversation with a therapist evaluating Scriber.

Will my clinical supervisor see the audio? No. Audio is never stored, so there is no audio for a supervisor to review. Supervisors review the generated text note the same way they review any other note — in the EHR, in the supervision documentation file, or in whatever workflow the supervisor and supervisee have agreed on. The audio existed for as long as it took to transcribe, then it was discarded.

How does Scriber fit my EHR copy-paste workflow? Most therapists are running SimplePractice, TherapyNotes, Valant, or a similar behavioral-health EHR. None of these have a public API that an AI scribe can write into without integration work. Scriber's current model is copy-paste: the generated note appears in the Scriber web app, the clinician reviews and edits it, and copies the finalized note into the EHR's progress-note field. It is the same workflow most clinicians already use for dictation tools. Direct EHR integrations are on the roadmap, with priority weighted by which EHRs serve the largest share of our user base. The Doxy.me integration is the first in that direction; bidirectional EHR sync is next.

Can a supervisor review my Scriber notes before they're signed? Yes. The supervisor and supervisee workflow runs against the generated text note. The supervisee finalizes the draft in Scriber, exports or copies it to the supervision review channel (an EHR with co-signature, a shared documentation file, or whatever the practice uses), and the supervisor reviews and signs as they would for any other note.

What happens if a client asks for the audio? There is no audio to provide. The transcript and the generated note are the records that exist. Clients have HIPAA right-of-access to their record, and a transcript is part of that record if the clinician retained it. Most clinicians delete the transcript after the note is finalized; some retain it for supervision. Scriber gives the clinician control of the retention setting.

What if I cover a session for a colleague and the client never consented to AI scribing? Don't record. The consent step is the clinician's, not Scriber's, and Scriber gives the clinician a one-click way to run a session without recording — the note generation features turn off, but the rest of the workflow stays available.

Try Scriber for therapy and counseling

The honest pitch is small. If you are a therapist or counselor spending an hour or two an evening on notes, an AI scribe is the most reliable way to get that hour back. Scriber is built for behavioral health specifically, generates the note formats therapists actually use, integrates directly with Doxy.me for telehealth, signs a BAA at signup, never retains audio, and lets you customize templates so the generated notes sound like you wrote them. There is a 7-day trial for clinicians who want to try a few sessions before committing, and paid plans for clinicians who have decided to make the switch.

The trial does not require a credit card. The BAA is in the signup flow. The Doxy.me integration is shipped. The seven STT engines, including the xAI fast engine that returns transcripts in roughly three seconds, are live. Custom templates are configurable. Run a few sessions, compare the generated notes to what you would have written yourself, and decide. That is the entire evaluation process.

If you have a question we have not answered here, the related solutions pages cover specific clinician roles (LMFTs, LCSWs, LPCs, couples therapists), the related specialty pages cover behavioral health and psychiatry in depth, and the FAQ at the top of this page captures the questions that come up most often.