Handling Supervision with AI Scribes

Short answer: Supervisors review the AI-generated text note in your EHR, the same way they review any note. Audio is never stored in Scriber, so there is no audio for a supervisor to listen to. If your supervision workflow requires audio review, you will need a separate audio recording system (not Scriber). For text-based supervision, Scriber's group plan supports supervisor review and co-signature of notes.

Supervision is how behavioral-health clinicians learn and how practices ensure quality. A supervisor reviews a clinician's notes, listens to audio from sessions, discusses clinical reasoning and technique, and signs off on the clinical and billing accuracy of the notes. The addition of an AI scribe creates a question: does the supervisor review the AI-generated note, the raw transcript, or the audio? And does audio retention change because of AI scribing?

The supervision workflow with AI scribes

Scriber does not retain audio. The audio is streamed to a transcription engine, transcribed, and the audio file is discarded. That choice is intentional: audio retention creates a new HIPAA obligation, increases breach risk, and does not scale. So the supervision workflow with Scriber has two lanes:

Text-only supervision: The clinician generates a note in Scriber, reviews it, exports or copies it to the supervision channel (SimplePractice, TherapyNotes, Valant, or a shared document), and the supervisor reviews the text note. The supervisor may ask questions ("Why did you use cognitive restructuring instead of exposure here?"), but the supervisor is reviewing the clinical content, not the audio.

Text plus external audio: If your supervision requires audio review, you run two recorders. One recorder is Scriber (for transcription and note generation). A second recorder is whatever system your practice uses for supervision audio (a secure digital recorder, a telehealth platform's recording feature, your phone recording app if you have consent). At the end of the session, the supervisor has the text note from Scriber and the audio from the other recorder.

Neither approach is unique to Scriber. Many supervision workflows already run text-only (the supervisor reviews notes, discusses with the clinician, does not listen to audio). Some supervision workflows use a separate audio-retention system alongside note documentation.

Text-only supervision (most common)

In a text-only supervision model, the supervisor reviews the note and discusses clinical decisions with the clinician.

How it works:

1. Clinician records a session in Scriber (with client consent).
2. Scriber transcribes and generates a note.
3. Clinician reviews the note, edits if needed.
4. Clinician exports the note to the supervision channel (copies into a shared SimplePractice or TherapyNotes folder, or puts it in a shared document).
5. Supervisor reads the note.
6. Supervisor asks questions or flags concerns (in the EHR, via email, or in supervision meeting).
7. Clinician discusses the case and makes any note edits.
8. Supervisor co-signs or approves the note.

Supervision questions the supervisor can ask from the text note:

• "Why did you choose EMDR instead of CBT for this trauma trigger?" (The note should indicate which modality was used and why.)
• "The client reported suicidal ideation in the last session. What's the safety plan?" (The note should document risk assessment and safety planning.)
• "Did you address the homework from last week?" (The note should include whether homework was reviewed.)
• "Why did you move to medication-change discussion in this session?" (The note should show the clinical progression.)

Supervision questions the supervisor cannot answer from the text note:

• "What was the client's tone when they said that?" (The supervisor would need to hear the audio.)
• "How quickly did the client respond to the exposure exercise?" (The supervisor would need to hear the pacing.)

Text-only supervision is sufficient for many clinical purposes. It assesses clinical thinking, diagnostic reasoning, and whether the note reflects what actually happened. It does not assess technique or pacing in the way that audio review does.

Text plus external audio (when required)

If your supervision requires audio review (because you are training a clinician in a specific modality like EMDR or CBT, or because your licensing requires it), you need a separate audio recording system.

How it works:

1. Clinician records the session in Scriber (for transcription and note generation).
2. Clinician also records the session in a separate system (secure digital recorder, telehealth platform's recording feature, etc.) that retains the audio.
3. Scriber transcribes, generates a note, and discards the audio.
4. The separate system retains the audio (encrypted, secure, with access controls).
5. Clinician reviews the Scriber note.
6. Supervisor reviews the Scriber note AND the audio from the separate system.
7. Supervisor gives feedback on both clinical content (from the note) and technique (from the audio).

This approach gives you the speed and specificity of AI note generation plus the audio review that supervision sometimes requires. The cost is managing two recording systems.

Scriber's supervision features

Scriber has a group/practice plan that supports supervision workflows.

Group notes folder: Multiple clinicians upload their notes to a shared group folder in Scriber. Supervisors in the group have read access to the folder and can review notes.

Note versioning: If a supervisor suggests edits, the clinician can update the note and create a new version. Both versions are retained, so there is an audit trail of what the clinician originally wrote and what changed after supervision review.

Co-signature workflow: If your practice uses Scriber's notes (not copy-pasting to SimplePractice), the supervisor can co-sign the note within Scriber. The note is then marked as supervised and signed.

Custom supervision template: Some practices customize the Scriber template to include a "Supervision Notes" section where the supervisor can document what they reviewed and any feedback. This becomes part of the final chart.

If you are using Scriber primarily to generate notes that you copy-paste into SimplePractice, TherapyNotes, or Valant, supervision happens in those EHRs — you copy the note to the EHR, the supervisor reviews it in the EHR and co-signs in the EHR. Scriber's group features are secondary.

Consent and privacy for supervision

When you get consent to record a session for an AI scribe, you are getting consent for the audio to be transcribed. You are not getting consent for the audio to be reviewed by a supervisor. Those are separate.

Best practice:

Modify your consent script to be explicit:

Standard consent: "I will be recording this session to generate clinical notes. The audio will be transcribed and then deleted. Do I have your permission?"

Consent for text-only supervision: "I will be recording this session to generate clinical notes. The audio will be transcribed and then deleted. My clinical supervisor will review the text notes I generate. Do I have your permission?"

Consent for audio supervision: "I will be recording this session. The recording will be used to generate clinical notes and will also be reviewed by my clinical supervisor for training and quality assurance. The recording will be securely stored and deleted after [time period, e.g., 90 days]. Do I have your permission?"

Note that the third consent is a much stronger privacy commitment. You are explicitly telling the client that the audio will be retained for supervision. Many clients are fine with that (it is standard in many clinics). Some clients object. The honest consent script respects the distinction.

HIPAA and supervision

Supervision is a legitimate clinical activity under HIPAA. A supervisor who is part of your practice (or is a contractual business associate) can review clinical notes and audio as part of their supervisory duties. That review is not a HIPAA violation.

However, the supervisor must have a legitimate need to know, and the audio and notes must be stored securely. If the supervisor works at a different organization (e.g., you have a consultation supervisor outside your practice), that person needs to be a business associate with a BAA, or you need consent from the client to share the note and audio.

Scriber's BAA covers its processing of the transcript and generated note. If you use Scriber's group features (uploading notes to a shared group folder), the supervisor accessing the note in Scriber is accessing notes covered by Scriber's BAA. That is clean.

If you copy-paste the Scriber note into SimplePractice, TherapyNotes, or Valant, and a supervisor reviews it there, the supervisor is reviewing it in that EHR, and that EHR's BAA covers it. That is clean.

If you retain the Scriber transcript separately for supervision, you are now retaining PHI, which means you need secure storage, access controls, and eventual deletion. That is doable but adds liability.

Supervision questions and answers

Q: Will my supervisor hear the audio?

A: No. Scriber does not retain audio, so there is no audio for your supervisor to hear. If your supervision requires audio review, you need a separate audio-recording system.

Q: Can my supervisor see the transcript?

A: The transcript is technically a business record in Scriber, but supervisors typically do not need to see it. They review the final generated note, which is the clinical document. If you want to share the transcript with your supervisor, you can export it from Scriber and send it to them securely (encrypted email, secure file-sharing service).

Q: Can I delete the transcript after the note is generated?

A: Yes. You can delete the transcript from Scriber after the note is generated and signed. The generated note persists; the transcript is deleted. This minimizes PHI retention.

Q: What if my supervisor wants to listen to a specific part of the session?

A: You need a separate audio recording system. Record the session in both Scriber and your supervision audio system. Your supervisor listens to the supervision recording, not the Scriber recording (which is discarded).

Q: Is supervision of AI-generated notes different from supervision of notes I write?

A: The supervision process is the same, but the content is different. When you write a note, the supervisor is reviewing your charting and clinical thinking. When the AI writes a note, the supervisor is reviewing the AI's draft and your review of the draft. Some supervisors prefer one or the other. Most supervisors say the AI-generated note is faster to review because it is already well-structured and complete, so they can focus on clinical reasoning instead of chart quality.

Best practices for supervision with AI scribes

1. Be clear about consent. Modify your consent script to explicitly state whether the audio will be retained for supervision or discarded.

2. Use text-only supervision if possible. Most supervision can be text-based. The supervisor reviews the note and discusses clinical reasoning. This is faster and requires no audio retention.

3. If you need audio, use a separate recorder. If supervision requires audio, use a separate, secure recording system. Do not expect Scriber to retain audio for supervision.

4. Document the supervision workflow. Write down: who supervises, how often, what is reviewed (note, audio, both), and where the supervison documentation is stored. This clarity helps if there is ever a compliance question.

5. Retain notes, delete transcripts. The clinical note is the document that matters for the chart. The transcript is temporary. Delete it after the note is signed to minimize PHI retention.

6. Use your EHR's co-signature features. If your EHR (SimplePractice, TherapyNotes, Valant) has supervisor co-signature, use it. The supervisor's sign-off is in the EHR, creating an audit trail.

Supervision with AI scribes is straightforward because the supervision process does not change — only the source of the initial draft changes. The clinician still reviews, the supervisor still supervises, and the clinical judgment still comes from the clinician.