AI Medical Scribe for SLPs

Session Audio → Clinical Evidence → SOAP Draft

The AI medical scribe that hears the therapy—not just the conversation.

RelyCare listens for targets, first attempts, cueing, errors, retries, and self-corrections—then turns session audio into a review-ready SOAP note with the transcript evidence still attached.

Founder-led pilotNo patient information in the application

Session analysis

Mason R. · Initial /s/

Draft

Transcript evidence

SLPFirst try: sun.

MasonThun.

SLPUse your skinny snake sound, then try sun again.

MasonSun.

Atomic observation

/θ/ for /s/ on first attempt

Accurate retry · verbal cue

Target metric

8 / 17

first attempts · 47%

Linked into the Objective draft for clinician review

The second shift

The session is over. The reconstruction work shouldn’t just be starting.

Stop rebuilding the session from memory. Start with the draft, then inspect the evidence behind it.

Built for speech-language pathologists and private clinics that want documentation relief without turning clinical review into a black box.

Why SLP-specific matters

A transcript tells you what was said. RelyCare organizes what mattered in therapy.

Readable prose is not enough when the clinical record depends on which response came first, what support changed it, and where the reported number came from.

Generic conversation summary

A polished paragraph—with the sequence flattened.

“The patient practiced /s/ words and improved with cues.”

It sounds plausible, but it does not tell the reviewer what was independent, what changed after support, or which excerpt supports the conclusion.

RelyCare’s review path

The mechanics stay visible long enough to challenge them.

  • Target and task context
  • First attempt kept separate from the retry
  • Cue type and audible self-correction
  • Representative error and transcript excerpt
  • SOAP language that remains editable

What RelyCare extracts

The details that disappear when a session becomes “patient tolerated treatment well.”

When the audio supports it, RelyCare looks beyond topic-level summarization and structures the response mechanics an SLP needs to review.

Therapy context

Targets and task context

What the session was working on, at what level, and inside which activity—when that context is audible or stated.

Attempt integrity

First attempts and retries

Independent responses remain distinct from what happened after support, so a successful retry does not rewrite the first attempt.

Response mechanics

Cueing and self-corrections

Verbal or model cues, repeated attempts, and audible self-corrections are organized into the review path instead of flattened into prose.

Traceable review

Errors and supporting quotes

Representative errors and transcript excerpts stay close to the observation or metric they support, ready for clinician inspection.

One continuous evidence trail

Follow one response all the way into the note.

The point is not to make the model’s answer look certain. It is to keep enough context attached that the clinician can inspect, correct, or reject it.

  1. 01

    Transcript

    “Thun.”

    The speaker-labeled excerpt preserves what was actually audible.

  2. 02

    Observation

    /θ/ for /s/

    The response is organized by target, context, result, and attempt type.

  3. 03

    Support

    Verbal cue

    The cue stays attached to the retry instead of inflating independence.

  4. 04

    Retry

    “Sun.”

    The accurate retry is captured separately from the first attempt.

  5. 05

    SOAP draft

    Evidence linked

    The clinician can inspect the source while reviewing the Objective section.

R

Controlled demo · session 04

Mason R. · Articulation · 45 minutes

Draft · review required

Review-ready draft

SOAP note

Every section remains editable. The treating clinician decides what enters the final record.

S

Subjective

Mason participated in structured word-level articulation practice targeting initial /s/. He remained engaged following inaccurate productions and responded to brief verbal placement cues.

O

Objective

Mason produced initial /s/ accurately on 8 of 17 first attempts (47%) at the word level. Following verbal placement cues (“skinny snake sound”), he produced 7 of 7 documented retries accurately and made 2 spontaneous self-corrections. In one representative trial, “sun” was initially produced as “thun” and was accurate on retry after a verbal cue.

Mason: “Thun.” → after verbal cue: “Sun.”

Transcript linked
A

Assessment

Mason demonstrated emerging independent accuracy for initial /s/ and strong immediate stimulability with a concise verbal placement cue. The difference between first-attempt and cued-retry performance suggests the target is achievable with support but not yet consistent independently. First attempts and prompted retries should remain separate when interpreting change.

P

Plan

Continue initial /s/ at the word level and advance stable productions into short carrier phrases as clinically appropriate. Fade cues from direct models to brief placement reminders, continue tracking first attempts separately from retries, and provide a short home-practice set using familiar initial /s/ words.

Clinician review is required before finalization.

Evidence attached

From audio to approved record

A faster starting point without surrendering the final decision.

  1. 01

    Capture the session under your clinic workflow

    Record only with the clinic's consent, privacy, access, and device procedures in place.

  2. 02

    Inspect what RelyCare organized

    Review the transcript, target-level observations, first-attempt counts, cueing, retries, self-corrections, and representative evidence.

  3. 03

    Edit, approve, and finalize the SOAP note

    Correct the draft, apply clinical judgment, and approve the final record. Raw session audio is deleted after finalization.

Clinical control, where it belongs

Trust should come from visible boundaries—not a row of reassuring badges.

These safeguards matter after you have seen the workflow, because they define how the product should be evaluated and used.

1

Audio has a boundary

RelyCare is designed around what was audibly demonstrated or verbally reported. Eye gaze, posture, silent behavior, and other visual findings require the clinician.

2

The clinician has authority

RelyCare prepares a draft. The treating SLP checks the evidence, corrects omissions or unsupported wording, applies clinical judgment, and approves the final record.

3

Audio has a lifecycle

Raw session audio is deleted after finalization. Clinics should still evaluate consent, devices, access, exports, and retention procedures during a pilot.

4

BAA eligibility is explicit

A BAA is available for eligible HIPAA-regulated clinics under the applicable agreement and required operating conditions; a contract alone does not make a workflow compliant.

Read the RelyCare BAA

Common questions

What an SLP should know before trusting the draft.

What is an AI medical scribe for SLPs?+

An AI medical scribe for SLPs helps turn speech therapy session context into a documentation draft. An SLP-specific tool should also preserve therapy targets, trials, cueing, errors, corrections, and the evidence a clinician needs to review the result.

Does RelyCare automatically finalize SOAP notes?+

No. RelyCare creates a review-ready draft. The treating clinician checks the transcript evidence, corrects errors or omissions, makes clinical decisions, and approves the final record.

Can RelyCare evaluate anything that was not audible?+

No. Audio cannot establish eye gaze, posture, silent behavior, or other visual and physiological observations. RelyCare is designed to limit extraction to what was audibly demonstrated or verbally reported, and clinicians should correct anything unsupported.

Is a BAA available for US clinics?+

A BAA is available for eligible HIPAA-regulated clinics under RelyCare's applicable agreement and operating requirements. A contract alone does not make a workflow compliant; each clinic remains responsible for consent, access controls, devices, exports, and its own HIPAA obligations.

Bring the difficult sessions

Don’t buy the clean demo. Test RelyCare against your real documentation workflow.

Define the sessions, review standard, privacy requirements, and success criteria that matter to your clinic. We will help you evaluate the fit.

Apply for a Pilot

Keep exploring

See how the rest of the documentation platform fits together.