Australian Allied Health Is Sitting on an AI Revolution It Refuses to Use

A scoping review published this year in the European Journal of Physiotherapy has confirmed something practitioners in this sector have quietly known for years: there’s a widening gap between what AI can do for allied health businesses and what those businesses are actually doing with it.

The review mapped AI adoption across physiotherapy, occupational therapy, speech pathology, podiatry and dietetics. It found that despite evidence AI can improve care delivery, workforce readiness to use it in clinical practice hasn’t followed. Tools exist. Evidence is building. Adoption is not.

That gap is costing practices real money. And the maddening thing is that most of the tools sitting unused aren’t clinical AI at all.

The admin problem nobody wants to talk about

Allied health runs on time. A physiotherapist billing at the NDIS rate earns around $193.99 per hour doing clinical work. Every hour spent on scheduling, note-writing, insurance claims, or chasing no-shows is $193.99 in forgone revenue.

Conway Consulting Group ran the numbers on this earlier in 2026. A clinician seeing patients for 5 billable hours per day generates around $213,400 per year. Push that to 6 hours, one extra hour of clinical work typically recovered by cutting admin, and you’re at $256,100. That’s a $42,700 difference per clinician per year. In a practice with four clinicians, that’s $170,800 sitting on the table.

Now layer in no-shows. Clinics that have adopted AI appointment management are cutting no-show rates by 40%. The average practice losing $50,000 to $100,000 annually to missed appointments isn’t an outlier. It’s standard.

Then there’s the documentation load. A physio seeing 20 patients per day, spending 10 minutes each on notes after the session, burns through more than 16 hours of note-writing per week. That’s not clinical time. It’s unpaid admin that bleeds into evenings. AI scribes cut that to a two-minute review of a first draft.

88% of clinics are still doing it manually

Research from Source Digital in April 2026 found that only 12% of Australian clinics have adopted AI in any meaningful way. The other 88% are still managing bookings by phone, writing notes from scratch, and sending generic reminder texts the night before appointments.

This isn’t a capability gap. The tools exist, they’re affordable, and most were built specifically for the Australian market. What’s missing is the decision to start.

The National AI Centre’s data from early 2026 shows health and education is one of the better-performing sectors for AI adoption among Australian businesses, with more than half of health businesses using AI in some form. But zoom in from “health businesses” to “allied health private practices” and the picture changes. Most of that adoption sits in larger hospital systems and digital-native providers, not in the physio clinic running three clinicians and a shared receptionist.

Why practices aren’t moving

The barriers are real, and worth taking seriously rather than writing off as technophobia.

Privacy is the legitimate one. Patient health information is sensitive. Using a general-purpose AI tool that stores prompts on overseas servers, with identifiable patient data included, is a genuine compliance risk under the Australian Privacy Act. That concern is well-founded.

TGA regulation creates uncertainty around clinical tools specifically. The TGA’s framework is purpose-based: if software is intended to diagnose, treat, or prevent a medical condition, it may need to be registered as a Software as a Medical Device. Most AI scribes that only transcribe and draft notes sit outside that definition, but the AHCRA flagged in May 2026 that many clinics adopted AI scribes without checking. When a tool’s latest update starts offering differential diagnoses, the classification changes and so do the compliance obligations.

Digital literacy gaps are real in smaller practices that haven’t had to think much about software beyond their practice management system.

And liability is a genuine question. If an AI scribe drafts a note that contains an error and a clinician signs off without catching it, who’s responsible? Under current AHPRA guidance, the clinician is. The AI doesn’t hold registration.

But every one of those barriers applies specifically to clinical AI: diagnostic tools, treatment recommendation engines, clinical decision support. Not one of them applies to an AI handling appointment reminders, answering questions about parking, or drafting referral letter templates.

The distinction that matters

Clinical AI and admin AI carry completely different risk profiles, and treating them the same is what’s holding most practices back.

Clinical AI involves patient data in ways that can directly affect care decisions. It sits within regulatory frameworks, carries real liability implications, and requires careful evaluation. Caution here is appropriate.

Admin AI handles scheduling, communication, note transcription (not diagnosis), billing workflows, and back-office tasks. The risk is comparable to using any other cloud software. A practice already using Cliniko or Nookal for bookings is already running a cloud system with patient data. The question isn’t whether to use software; it’s whether the software is fit for purpose and privacy-compliant.

The tools built specifically for Australian allied health have already navigated this distinction:

• Appointment booking: HotDoc, Cliniko, and Nookal are Australian-built practice management platforms with online booking, automated reminders, and waitlist management. These aren’t experimental.
• Clinical notes: Heidi Health and Lyrebird Health are both Australian-built AI scribes. They transcribe the consultation and draft the note; the clinician reviews and edits. Heidi serves over a million clinicians. Lyrebird is free for Best Practice Premier users.
• Billing and claiming: Tyro Health and Medipass handle Medicare, DVA, WorkCover and private health claiming without the manual processing that currently eats receptionist hours.

None of these require a TGA submission. None involve AI making clinical decisions. They handle the administrative layer between clinician and patient, and that layer is eating hours every day in practices that haven’t moved.

What the research is actually saying

The 2026 scoping review isn’t arguing that practitioners should adopt every AI tool available. It’s identifying a structural problem: the gap between what’s possible and what’s being done is growing, not because the tools are inadequate, but because the sector hasn’t built a way to evaluate them with confidence.

The Equal Experts APAC study of 100+ Australian digital health leaders found that 65% of organisations rated their operating model maturity at the lowest levels. It’s not a money problem or an availability problem. Nobody has mapped out a decision process. When there’s no framework for evaluating a new tool, inaction is the default.

That’s fixable. A practice that wants to start doesn’t need to solve TGA compliance and clinical AI ethics on week one. It needs to pick one administrative pain point, find the right tool for it, run it for 90 days, and measure what changed.

The physio practices using AI appointment management aren’t doing anything complicated. They’re doing something their competitors aren’t. And the data on no-show reduction and admin time recovery is clear enough that the case for inaction gets harder to hold every quarter.

Start with admin, not clinical

Allied health professionals are trained to be careful and sceptical of interventions that haven’t been properly evaluated. Those are the right instincts in a clinical context.

They’re the wrong instincts to apply to appointment booking software.

If your practice is losing hours every week to phone tag and post-appointment note-writing, the tools to fix that are available, Australian-built, and already running in practices like yours. The clinical AI questions can wait. The administrative ones have been answered.

The gap the researchers found isn’t inevitable. It’s a decision, and it comes with a price tag.

📖 More deep dives: News Deep Dives

Related: Companies That Replaced Staff With AI Are Already Rehiring. Here’s What That Means for Australian Small Business | Lyrebird Health vs Heidi Health: Which AI Clinical Scribe Is Right for Australian GPs?

📬 Subscribe to our weekly newsletter | 💼 Free resources & AI prompt packs

📖 More deep dives: News Deep Dives

📊 Compare AI tools side by side | 💼 Free resources & AI prompt packs

More on AI for allied health: Allied Health AI Hub →

📊 Compare AI tools side by side | 💼 Free resources & AI prompt packs

More hot takes: Hot Takes →