Composite case study based on real patterns from Australian allied health practices. Names are illustrative.

Priya runs a four-practitioner physiotherapy and occupational therapy practice in Brisbane. About 40% of clients are NDIS participants. Before AI tools, clinical administration was consuming an estimated 15+ hours per week across the team: progress notes, NDIS documentation, patient communication, and reporting. Six months in, they’ve cut that by roughly a third. Here’s how.

The NDIS Documentation Problem

NDIS progress notes and reports are time-consuming to write and must meet specific formatting requirements. The practice uses a structured ChatGPT prompt that takes a practitioner’s dot-point session notes and produces a draft NDIS-compliant progress note. The practitioner reviews, edits, and signs off: the AI never sees identifying patient information (first names and session details only, no Medicare numbers, DOB, or addresses).

Time saving: approximately 10–15 minutes per NDIS client per session. At 20 NDIS clients per week across the practice, that’s 3–5 hours saved weekly on documentation alone.

Patient Communication

ChatGPT drafts appointment reminders, post-session care instructions, and re-engagement messages for clients who’ve dropped off. The receptionist provides the key details; ChatGPT produces the draft; it’s reviewed and sent. Clear, warm, professional: and consistent, which wasn’t always the case when each practitioner wrote their own.

Referral Letters

Referral letters to GPs and specialists used to take 20–30 minutes each. With a structured prompt template, they now take 8–10 minutes. The template captures: patient presentation, treatment provided, response to treatment, recommendations, and follow-up requirements. Practitioners add clinical specifics; AI handles the structure and language.

What They’re Careful About

Health data is sensitive and the Privacy Act 1988 applies strictly in healthcare. The practice’s rule: no full names, no Medicare numbers, no diagnoses in AI prompts: session notes are entered as clinical descriptions only, with identifying details added by the practitioner after the AI draft is produced. All AI output is reviewed by a qualified practitioner before sending or filing.

The Numbers

Tool cost: ChatGPT Team plan, ~AU$38/user/mo for 2 users
Estimated admin time saved: 4–5 hours/week across the practice
Equivalent cost: less than one billable hour per week

The practice is now looking at AI transcription tools (like Heidi Health, built for Australian allied health) for session notes: the next frontier for clinical admin reduction.

📖 Full AI guide for allied health → | AI and privacy →