SessionScribe
An on-device AI dictation app for therapists and psychologists that turns a rambling 50-minute session into a structured SOAP or DAP note — audio and transcripts never leave the clinician's iPad or Mac.
solo therapists and group-practice psychologists documenting sessions
- Push-to-talk session recording that runs entirely on Apple Silicon with no network calls
- Auto-structured note templates (SOAP, DAP, BIRP) generated locally from the transcript
- On-device redaction engine that flags and masks client names, locations, and identifying details before the note is saved
- Encrypted local archive with biometric lock and one-tap export to EHR clipboard formats
Murmur's GitHub traction (77 stars in one day) shows developers and privacy-focused professionals are actively seeking cloud-free alternatives; combined with mounting HIPAA and AI-act anxiety, on-device clinical dictation is a demand signal clinicians have been waiting for.
Multiple paid AI therapy-note tools already in market (Freed, TheraScribe, Mental Note AI, AIDoctorNotes) with active users; CBC confirms therapist adoption of AI transcription in clinics.Freed AI Review for Therapists | Pricing, Features, Alternatives ↗Therapists say AI can help them help you, but some see privacy concerns ↗
Therapy AI-scribe space is saturated and well-funded ($800M VC in 2024 for Abridge/Freed/DAX); however, genuinely on-device solutions are rare — only VoiceScriber and basilai content positioning that angle as undifferentiated by big cloud incumbents.Abridge vs Ambience vs Freed | Sacra ↗Healthcare & Therapy Notes on iPhone: On‑Device Transcription for Clinicians ↗
Proven willingness to pay: Freed charges $39–$99/mo per clinician and reports users saving 3+ hrs/week justifying the cost; group practices bill standard rates with custom enterprise quotes.Freed Pricing 2026: $39/mo Starter, $79/mo Core Plans ↗Freed Pricing | Freed Scribe Help Center ↗
HIPAA, PIPEDA and emerging AI-act regulations are tightening around clinical audio; on-device posture is a durable structural advantage unlikely to be obsoleted, though Apple Intelligence or OS-level dictation could compress differentiation over time.
Buildable using on-device Whisper.cpp + a small structuring LLM on Apple Silicon for SOAP/DAP output, but requires meaningful engineering for 50-min session reliability, speaker separation, and battery/memory constraints on iPad.