Industry guide - clinics and allied health
AI for clinics: the front-desk build sheet
Most writing about AI in healthcare is about medicine. This page is about the front desk: recalls, intake, referrals, enquiries and billing - the administrative workflows that automate cleanly because nothing clinical is ever delegated to a model.
By Tylar Edwards - the engineer who scopes and builds these workflows for practices. Updated August 2026.
The one rule: administrative, never clinical
A practice's hours disappear at the front desk: reminder calls, intake chasing, referral filing, the same twelve questions answered by phone, and invoices followed up between patients. All of that is process, and all of it automates. The boundary is absolute and designed in: nothing that diagnoses, triages or advises a patient is automated. Anything symptom-shaped routes to a person by rule, not by model judgment.
Six front-desk workflows AI automates today
| Workflow | What AI does | What stays with the practice |
|---|---|---|
| Reminders and recalls | Runs appointment reminder and recall ladders from your practice software, rebooking no-shows on your rules | Nothing - this is the purest recovered time at a front desk |
| Patient intake | Sends and chases intake forms, extracts responses into the patient record, flags incomplete answers | Clinical review of everything submitted |
| Referral and correspondence triage | Sorts referrals, reports and letters by patient and urgency, files them, drafts acknowledgments | Clinical prioritisation - urgency rules are set by the practitioner |
| Routine enquiry handling | Answers hours, fees, parking, preparation and availability questions; books appointments within your rules | Anything symptom-shaped goes straight to a person, always |
| Billing follow-ups | Chases gap payments and unpaid invoices on a ladder, reconciles into Xero or MYOB | Hardship conversations and write-off decisions |
| Letter drafting | Drafts referral replies and routine letters from your templates and file notes | Every clinical statement, and the signature |
Around your practice stack
Cliniko, Halaxy or practice software
Bookings and patient records
Clinical and appointment data needs a controlled system of record.
Email and forms
Intake and communication
Manual intake creates duplicate entry and follow-up work.
Xero or MYOB
Accounting
Billing data commonly moves into a separate finance system.
What it is worth
Using the capacity model behind our free assessment at health services rates: eight recovered front-desk hours a week models to an indicative $11,680-$18,320 a year at $80/hr - and a recall ladder that never lapses tends to be worth more than the hours themselves.
Common questions
- Can AI handle patient enquiries safely?
- Yes, with one hard rule: the AI answers administrative questions only - hours, fees, bookings, preparation instructions. Anything symptom-shaped or clinical routes immediately to a person. That routing rule is enforced in the workflow design, not left to the model's judgment.
- What about patient privacy?
- Patient data stays in the practice systems you already run - Cliniko, Halaxy or your practice software - and workflows operate inside your existing access model. Nothing patient-identifiable trains anyone's model, and the data-handling design is agreed in scoping before any build.
- What should a clinic automate first?
- Recalls and reminders. No-shows and lapsed recalls are direct revenue loss, the ladder is pure process, and the practice feels the difference in the first fortnight of appointment books.
- What is it worth to a practice?
- At health services capacity rates, eight recovered front-desk hours a week models to an indicative $11,000-18,000 a year - before counting the recall revenue a consistent ladder recovers.