Table of contents
A 12-provider clinic was losing providers to burnout. Clinicians spent 2.5 hours per day on documentation. No-show rate was 22%. New patient intake required 25 minutes of in-appointment paperwork.
The deployment#
- Scribe (Documentation): Transcribes voice recordings into structured notes (SOAP for GPs, DAP for psychologists), assigns ICD-10 codes.
- Aria (Scheduling): Manages appointments, sends context-aware reminders, handles cancellations, manages waitlist.
- Welcome (Patient Intake): Guides patients through digital registration, insurance upload, medical history, consent forms.
Results (90 days)#
| Metric | Before | After | Change |
|---|---|---|---|
| Clinician admin time/day | 2.5 hours | 35 minutes | -77% |
| Same-day chart completion | 40% | 94% | +135% |
| No-show rate | 22% | 14% | -36% |
| In-appointment intake time | 25 minutes | 5 minutes | -80% |
| Provider satisfaction (1-10) | 5.2 | 8.1 | +56% |
What mattered most#
Clinicians stopped taking work home. Notes are waiting in the EHR within minutes of dictation. Review and sign-off takes 2-3 minutes per chart.
Aria's no-show reduction was driven by personalized, context-appropriate reminders — not generic messages. The specificity of the communication made patients feel recognized.
HIPAA and privacy#
All agents operate on HIPAA-compliant infrastructure. Scribe doesn't store raw audio. Welcome processes intake data in an encrypted pipeline with no persistent storage of unprocessed PHI.
Written by
Bitontree Team
AI Workforce Engineers
Bitontree designs and deploys teams of AI employees for businesses across legal, healthcare, accounting, real estate, recruitment, SaaS, and e-commerce. We write about what we learn building and shipping AI workforces in production.