Scriber — AI Medical Scribe for Behavioral Health Clinicians

AI scribing isn't a universal tool. There are legitimate scenarios where it adds friction, creates compliance headaches, or doesn't fit your workflow. Knowing when not to use it is as important as knowing when to use it.

Session Type: When to Skip Recording

Family or couples therapy where recording consent is complex. In couples or family sessions, you need consent from every participant. Some partners or family members decline recording, and that's their right. You can't force AI scribing on a session. For these cases, you might:

Dictate a summary post-session (AI scribe still helps here, but it's slower than session recording)
Write the note from scratch (skip AI entirely for this one session)
Have standard consent forms that allow recording for some sessions but not others

Group therapy. Recording a group of 6–10 people and generating an AI note that captures individual contributions is harder. Speaker diarization (who said what) becomes more complex. Some clinicians skip AI recording for groups and write notes from notes taken during the session.

Involuntary or forensic-adjacent sessions. If you're documenting court-ordered evaluations, custody assessments, or other forensic work, recording (even with consent) can introduce legal complexity. Check with your attorney before using AI recording in these contexts.

Highly sensitive disclosures. Sessions involving acute trauma, recent abuse, or crisis disclosure sometimes feel different when both participants know a recording is happening. Some clinicians will pause AI recording during especially sensitive moments and resume after, or skip the session entirely.

Clinical Context: When Accuracy Matters More Than Speed

Detailed medication management sessions. Psychiatrists and psychiatric NPs often need notes that precisely document:

Current medication list with exact doses, frequencies, side effects
Rationale for medication changes (why you're switching from sertraline 50mg to paroxetine 20mg)
Patient tolerance and response (whether the medication helped, side effects, compliance barriers)

AI notes capture this, but review time might be 10–15 minutes for a medication management session (longer than a therapy session). If you have 4–5 med management appointments per day, the time savings are diminished.

Complex presentations with multiple diagnoses or treatments. A patient with bipolar disorder, substance use disorder, and trauma might need a note that integrates findings across multiple domains. AI can do this, but you'll probably spend more time verifying that the synthesis is accurate.

Sessions where you're making significant clinical decisions. Hospitalization, medication increase, change in diagnosis, or discontinuation of treatment. These notes carry legal weight, and you should probably take extra time with them regardless of whether they're AI-generated.

For these high-stakes notes, you might use AI to generate a draft, but expect review time of 15–20 minutes instead of 5 minutes.

Compliance and Consent: When Recording Complicates Things

Patients who specifically decline audio recording. Some patients will say: "I don't want you to record this session." You must respect that. You can either:

Use dictation (non-session recording) to generate the note with AI
Write the note from scratch
Ask why they decline and see if their concerns can be addressed

But don't record against their wishes.

Settings with mandatory consent procedures. Some group practices, agencies, or institutions require a lengthy consent process for session recording (not just a checkbox, but a detailed conversation about what happens to the audio, who accesses it, retention policies). If your setting requires this, the overhead might outweigh the benefit. You'd still need to have that conversation; AI scribing just adds a tool to it.

Legal hold or litigation scenarios. If a patient has sued you or threatened to, you're under legal hold: you must preserve all communications and documentation. Recording introduces another data stream (the audio) that you must preserve. Some lawyers recommend pausing AI recording during litigation to simplify evidence preservation. Check with your counsel.

Technical Fit: When Integration Breaks Down

EHRs with no import capability. Some EHRs only accept notes through their proprietary input screens and don't support copy-paste imports. If your EHR is like this, every AI-generated note requires manual retyping or data entry. This eliminates time savings.

Before implementing AI scribing, verify your EHR supports pasting structured text into the note field. If it doesn't, AI scribing won't work for you.

Offline-first practices. If you work in a setting with unreliable or no internet (rural clinic, field-based services), real-time AI transcription won't work. You'd need to record locally and sync later, introducing latency. This might still be worth it, but it's a different workflow.

Settings with restrictive data policies. Some government agencies, military, or classified healthcare settings have data policies that don't allow using cloud-based transcription services. If your setting has this restriction, AI scribing might not be compliant. Check before implementing.

Volume: When You Don't See Enough Patients

If you're seeing fewer than 5 patients per week, time savings are minimal (20–40 minutes per week). The $99/month cost might not be worth it. You'd save money just writing notes yourself.

If you see 5–10 patients per week, the math is borderline. You get some time back, but not dramatically. At 10+ patients per week, AI scribing clearly makes sense.

Specialty: When Your Workflow Is Already Optimized

Practices using structured templates that are already fast. Some clinics use fill-in-the-blank templates for all notes (checkboxes, drop-downs, minimal free text). If note completion already takes 2–3 minutes per session, AI scribing doesn't add much value.

AI shines for narrative note-writing. If you're already using heavily structured templates, you might get minimal benefit.

Training or Supervision: When You're Learning

New clinicians still learning documentation. If you're a graduate student, intern, or early-career clinician still learning how to write good notes, AI scribing can actually slow you down. You need to practice generating structure and clinical language yourself. Relying on AI too early might impede learning.

That said, using AI + a supervisor's feedback is a valid learning tool. The AI does the draft; the supervisor teaches you how to improve it. This can accelerate learning. But passive AI use without feedback probably limits your development.

Workflow: When You're Dictating, Not Recording

Practices built on dictation, not session recording. Some clinicians (especially older practitioners) never record sessions. Instead, they dictate a summary post-session to a transcriptionist: "Session with John, discussed anxiety, used CBT techniques, plan is to continue therapy, increase coping strategies."

If you're dictating a summary, AI scribing still works (it can turn the dictation into a structured note), but it's slower than generating notes from a full session recording. The value proposition is weaker because the AI doesn't have the full session context.

If your practice is fundamentally based on dictation, not recording, you might get more value from a traditional transcriptionist service or from an AI that works better with dictation workflows.

Financial: When You Can't Afford It

This is honest: if $99/month is a financial burden, don't use it. Your time is valuable, but not more valuable than meeting your own basic needs.

If you're a solo clinician in a lower-cost area seeing 8 patients per week at $75/session, your revenue is ~$600/week or ~$2,400/month gross. $99 is 4% of that. It's affordable.

If you're seeing 3 patients per week at the same rate, you're at $900/month gross. $99 is 11%. That's higher relative cost, and the time savings might not justify it.

Honestly assess your budget before committing.

Your Decision Matrix

Ask yourself:

Can I record sessions with full consent? (Yes → move to Q2; No → skip AI scribing)
Do I see 8+ patients per week? (Yes → move to Q3; No → reconsider)
Does my EHR support copy-paste note import? (Yes → move to Q4; No → skip AI scribing)
Can I afford $99/month sustainably? (Yes → use AI scribing; No → wait until you can)

If you answer "yes" to all four, AI scribing makes sense. If you answer "no" to any, be honest about it and either find a workaround or skip it.

The Bottom Line

AI scribing is genuinely useful for many clinicians, but it's not a universal solution. If your practice has high volume, standardized note formats, reliable internet, and full session recording consent, AI scribing is probably worth it. If any of those factors are missing, be honest about the friction before you commit.

A tool that doesn't fit your workflow is just another subscription expense. Better to skip it and use that $99/month elsewhere.

When Not to Use AI Scribe | Scriber Wiki