How a Physical Therapy Practice Stopped Losing Patients to Voicemail
Physical therapy is a referral and availability business. A patient gets referred, they call to book, and if nobody answers, a meaningful number of them call the next practice on the list.
That isn’t a failure of care. It’s a failure of availability at the exact moment availability matters.
The situation
Best Life Physical Therapy runs locations in Georgia and Kentucky. Like most practices, they see patients in back-to-back sessions while the front desk handles check-ins, insurance verification, billing questions and scheduling at the same time.
The phone rings mid-session. The front desk is already on another call. It goes to voicemail. The caller hangs up without leaving a message.
This happened several times a day, and the cost was invisible — there’s no record of a caller who never left their name. You can’t review a report of people who didn’t reach you.
What we built
An AI voice agent called Maya, answering every inbound call: during sessions, after hours, and at weekends.
She checks real-time calendar availability, answers routine questions about insurance and location, and books appointments directly into the practice management system. It runs on PracticeQ for scheduling and GoHighLevel for the CRM and follow-up, with Retell AI handling the voice layer — which matters mostly because the agent can read and write to the same records the front desk uses. It isn’t a separate system that someone reconciles later.
The numbers
In a typical week Maya handles more than ten calls.
Around 80% she completes without any practitioner involvement. The patient gets their question answered, their appointment booked, and a follow-up text with confirmation and directions.
The remaining 20% get routed to a person — Maya texts the front desk the caller’s exact request, so the callback happens with context instead of starting from “you called?”
The patients who would previously have hung up and dialled another practice are now on the schedule.
What it deliberately doesn’t do
This is the part worth being specific about, because it’s where these projects usually go wrong.
Maya answers the predictable calls: do you take my insurance, what’s your availability this week, where are you located, how do I book. Those follow a pattern and they’re the bulk of the volume.
Anything requiring clinical judgment gets flagged for a callback. What to expect from treatment, whether a particular condition is something the practice treats, anything about a patient’s specific situation. The agent doesn’t attempt those, and that boundary was decided before it went live rather than discovered afterwards.
For a practice, the risk was never a missed booking. It’s an AI saying something clinical to a patient in pain. Drawing that line narrowly is what made the rest of it safe to run.
What changed
Front desk staff spend less time on routine calls and more on the interactions that need a person. Practitioners aren’t pulled out of sessions to answer scheduling questions. New patient calls get answered while the practice is fully occupied.
The missed call problem didn’t get managed better. It stopped being a problem the front desk was responsible for solving.
A practice can’t hire its way to 24-hour availability. It can build it once.
Book a free call to talk through what a connected system would look like for your setup.