A multi-branch surgical day-care hospital in South India uses an AI voice agent to call patients after their visit or procedure and book the follow-up appointment the doctor asked for. Its telecallers were busy with new enquiries, so follow-ups were the calls that slipped. The agent now calls in Kannada by default, switches to Tamil, Telugu, Malayalam or English when the patient prefers, and books directly into the hospital's appointment system. This post describes how it works, without naming the hospital. It is the AI receptionist for medical clinics used for outbound follow-ups.
See how it sounds in your language: Start free on Edesy — Rs 50 free credit, no demo needed.
The problem: follow-ups are the first thing to slip
The hospital runs around ten centres. Each has a small telecalling desk handling new enquiries, appointment changes and reminders. Follow-up calls after a procedure are important for the patient and for the hospital, but they never feel urgent, so they get pushed to "later" and often don't happen. Patients who aren't reminded often don't come back for the review visit the doctor asked for.
What the AI agent does
- Calls the patient a set number of days after the visit or procedure, in their language.
- Checks in briefly and politely, without giving medical advice.
- Offers the follow-up appointment with the right doctor at the right branch.
- Finds free slots from the hospital's appointment system, books the one the patient picks, or reschedules or cancels an existing booking.
- Transfers to a person for any clinical question or complaint.
- Saves the outcome: booked, declined, call back, wrong number, plus an English transcript and structured fields for the hospital's records.
One script per condition
The hospital treats several conditions, and patients speak about each one differently. Instead of one generic script, there is one script per condition, written with the clinical team: what to ask, which words to use and avoid, and when to hand over to a person. It keeps the call short and respectful, and avoids the agent improvising on medical topics.
Connecting to the appointment system
The agent books through the hospital's appointment API using a set of tools:
| Tool | What it does |
|---|---|
| Find location | Nearest or preferred branch |
| Find specialty and doctor | The doctor the patient saw, or another in the same specialty |
| Get free slots | Available times for that doctor |
| Identify patient | Match the caller to their record |
| Book / reschedule / cancel | Change the appointment in the system |
A practical lesson: set the time zone on every branch. When one location was missing it, the agent read out slot times that were off. Small configuration gaps like this show up in the first days of a pilot, which is why listening to calls early matters.
For the general pattern of API-based booking, see how to automate appointment booking with voice AI.
Languages
Kannada is the default for most branches, with Tamil, Telugu, Malayalam and English available. The agent switches when the patient speaks another language. Setting the default per branch, rather than one language for the whole group, keeps first sentences natural.
How the pilot is measured
The hospital measures one thing: follow-up appointments booked. Not calls made, not minutes.
The comparison that matters to a hospital is not the per-minute price. It is the cost per booked appointment against the telecaller desk. A telecaller in a South Indian city typically costs Rs 18,000-25,000 a month and makes a limited number of calls a day alongside everything else. The AI takes the follow-ups off that desk entirely.
Another lesson: a large share of numbers in any hospital list are wrong, switched off or hang up instantly. Pay attention to connected calls and booked appointments, not dialled numbers. On Edesy, unanswered calls are not charged.
Starting small
The pilot started at around 20 calls a day at one set of branches, with the team listening to recordings daily and adjusting scripts, before increasing volume. For clinics planning the same, our post-discharge follow-up calls guide and how voice AI reduces hospital no-shows cover the wider picture.
Try it for your hospital
Create a free account and hear a follow-up call in Kannada, Tamil, Telugu or Malayalam, or talk to us about connecting your appointment system. For after-hours calls, read AI night receptionist for hospitals.