Case Study · 02 / 05
AI Voice Agent For Inbound Calls.
After-hours overflow handled by a voice agent that books, reschedules, and triages.
Context
Where they were coming from.
A primary care network was losing roughly 18% of inbound calls outside business hours. The voicemail box was a black hole. Staff did not return them, and patients did not call back.
The work was straightforward. Confirm a name, find an open slot, ask three triage questions, escalate if anything sounded urgent. It was just happening at midnight.
What We Built
The agent, specified.
A voice agent built on Vapi orchestration, ElevenLabs voices, and a thin booking layer that integrates with the practice management calendar. The agent answers, books, reschedules, and routes anything flagged urgent to an on-call human.
- Vapi for telephony orchestration
- ElevenLabs voice cloning (with a real clinician's permission)
- Real-time slot lookup against the practice management system
- Triage classifier with conservative escalation
- Full call transcripts written to the chart
Results
The shape of the curve after launch.
Illustrative metrics pending client sign-off. The shape of the curve matters more than the exact number. These are the moves we aim for.
73%
After-Hours Calls Resolved Without A Human
−18%
Voicemail Abandonment
92%
Patient Satisfaction In Post-Call Surveys
Have a workflow like this?
Bring us the workflow that never quite gets done. We'll spec the agent that does it.