May 20, 2026· Automation · Healthcare
After-Hours AI Receptionist
An AI voice agent that answers after-hours calls for a boutique dental practice.
StatusLive in production. Gathering call-volume and callback data.
Built with
- Retell
- n8n
- Google Sheets
- Claude
The problem
Golden Bear Dental (alias), a single-dentist practice, had no real after-hours solution. Calls either went to voicemail or rang the owner's personal cell, leaving patients with an inconsistent experience and the dentist on call around the clock.
What I shipped
An AI voice agent that answers after-hours calls with a script grounded in this practice's policies, services, and intake requirements. It triages emergencies with live transfer to the dentist, captures structured intake data, and surfaces every call in a morning digest for the front desk.
Architecture
Example: New Patient Looking for a Routine Cleaning
Sample after-hours call
0:000:00
Design decisions
- Handles unique caller personas. A single agent identifies what the caller needs (routine visit, emergency, price or FAQ) and adapts the conversation. Every path converges into one intake so the front desk can process follow-ups the same way each morning.
- The agent does not book. Letting the agent schedule directly means messy PMS integrations and a wider HIPAA surface. This design stays clean: it captures intent and preferences and leaves scheduling to the front desk, so the workflow stays PMS-agnostic across most practice management systems.
- Emergencies offer a choice. Severe enough to need the dentist tonight? The patient chooses direct transfer to the on-call dentist or callback. Removes the agent from making medical judgments.
- Upfront AI disclosure. Required by California AB 489 as of Jan 2026.
The data layer
