What is an AI Clinical Scribe? A Guide for Australian Specialist Practices in 2026
An AI clinical scribe is software that listens to a consultation and drafts the clinical note for the practitioner to review, edit and approve. For specialist practices across Australia, the appeal is straightforward: less time typing after hours, and more attention on the patient in the room. What is less straightforward is the regulatory picture, which shifted materially in August 2026 when the TGA (Therapeutic Goods Administration) moved from reviewing digital scribes to compliance action. This guide explains what an AI clinical scribe does, how it differs from dictation, what Australian regulators expect of you, and what to ask a vendor before you commit.
In this article
What an AI clinical scribe actually does
An AI clinical scribe captures the conversation in a consultation, transcribes it, and structures that transcript into a draft clinical note in the practitioner’s own template. The draft is not the record. It becomes the record only when the clinician has read it, corrected it, and approved it.
The category is often called an ambient AI scribe, because capture happens in the background while the consultation proceeds normally. There is no dictating and no pausing to narrate. The practitioner talks to the patient; the software listens and produces a structured draft afterwards.
That distinction matters when you are comparing tools, because the two are often sold under the same banner. Dictation software turns your spoken words into text, so you are still composing the note yourself, sentence by sentence. An ambient AI scribe listens to the conversation you were having anyway and produces a draft you did not have to compose at all. Both are sold as clinical documentation software, so it is worth asking a vendor plainly which of the two you are being shown.
A well-designed scribe stays inside a narrow lane: it captures what was said, structures it against a fixed template, and drafts correspondence from it. The clinical judgement, and the final word, remain with the practitioner.
Why specialist practices in Australia are adopting AI clinical scribes
Specialist practices are adopting AI clinical scribes primarily to reduce the documentation load that accumulates outside consulting hours. Ahpra (Australian Health Practitioner Regulation Agency) has explicitly acknowledged the role AI can play in reducing administrative burden, noting that scribing tools used responsibly can support workload management in clinical record keeping.
The pressure is particularly acute in specialist settings. A specialist consultation generates more than a note: it generates a letter back to the referring practitioner, and frequently a procedure record or a follow-up plan. Correspondence-heavy workflows are where documentation time concentrates, and where the effect is most obvious.
Practice managers tend to describe a different benefit. Correspondence goes onto the letter template of the relevant doctor, so letters leave the rooms faster and a new registrar has something structured to work from on day one.
How AI clinical scribes are regulated in Australia
In Australia, an AI clinical scribe is regulated as a medical device only where its intended purpose brings it within the definition in the Therapeutic Goods Act 1989. The TGA is clear that not all digital scribes are medical devices, and that the answer depends on what the product is designed and represented to do rather than on the technology inside it.
What the TGA looks at
The TGA’s published position is that a digital scribe which only transcribes a consultation, records a diagnosis and treatment as stated by the practitioner, and drafts administrative documents sits outside the medical device definition. A product that goes further, suggesting a diagnosis, recommending a treatment, or interpreting an investigation, is a different proposition.
The regulator also expects developers to reassess intended purpose whenever the product changes. If an update introduces functionality with a therapeutic use, the product becomes a medical device, whatever it was at launch.
That expectation now has enforcement behind it. In August 2026, after twelve months of engagement with vendors, the TGA confirmed that its digital scribes review had moved into compliance action. The review identified scope creep, AI functions influencing clinicians’ decisions, gaps in monitoring controls and mitigation of foreseeable risks, and limited transparency and post-market feedback once products were deployed. Updated guidance is expected to follow. The TGA digital scribes guidance sets out the regulator’s current position in full.
What Ahpra expects of practitioners
Ahpra’s guidance, Meeting your professional obligations when using Artificial Intelligence in healthcare, is the piece most often overlooked. It applies to every registered practitioner and it is unambiguous: you remain responsible for any AI used in your practice.
In practice, that means four things:
- You need to understand the capabilities and limitations of the tool you have chosen.
- You need to be transparent with patients about its use and obtain informed consent.
- You need to satisfy yourself that it meets privacy and data governance standards.
- You need to apply human judgement to every output.
Ahpra is explicit that TGA approval of a tool does not reduce that obligation. The full guidance is available on the Ahpra website.
Consent and privacy
Consent practice varies widely between clinics, and that variation has drawn attention from the Office of the Australian Information Commissioner and from consumer groups. Meaningful informed consent means the patient understands that the consultation is being captured, what happens to it, and that they may decline without consequence to their care. A sign at reception is a start, not a consent process.
Built in, not bolted on: xScribe inside Xestro
xScribe is Xestro’s own AI clinical scribe, built into the platform rather than added alongside it. That architectural choice is the substantive difference between a built-in scribe and a standalone one, and it shows up in the workflow.
The practical difference is what the scribe has access to before it starts. A standalone scribe typically starts with the conversation alone. Getting the draft into the right record is then a separate step, whether through an interface or by hand. A scribe built into the specialist practice management software opens with the patient’s file already there, the medical context, the correspondence history, the template that practice uses, and the fields the note needs to fill. There is no second application to open, nothing to copy between windows, and no re-keying a draft from one system into another.
Built to fit the way each specialist works
xScribe is set up by the practice and by each provider, rather than arriving with a fixed house style. Practices keep working with the document templates they have already built in Xestro, and each practitioner decides how their own drafts read: the structure, the level of detail, the language, and the tone. Two specialists in the same rooms can record a similar consultation and receive drafts that read quite differently, because each has configured it their own way.
Every draft is built from what was said in the room. Providers can also choose how closely the draft mirrors their own words: from restructured into template language through to near-verbatim phrasing. That control is worth asking any vendor about, because a scribe that stays within what was said is more useful than one that pads the note.
One consultation, more than one document
The clinical note is one output of xScribe, not the only one. A single consultation can produce the correspondence that follows it: letters to referring practitioners, patient summaries, referrals, certificates and reports. Each is drafted from what the practitioner stated during the consultation, and each is reviewed and approved before it goes anywhere.
Those documents follow the practice’s own templates, so what comes out looks like what the practice already sends rather than something the software imposed.
Clinician in the Loop. xScribe drafts; you decide. Nothing reaches the patient record until you have reviewed it, edited it where needed, and approved it. The AI does the typing; the clinical judgement, and the final word, remain yours.
What to ask before you choose an AI clinical scribe
The most useful evaluation questions are about scope and accountability, not about accuracy percentages. Work through these seven with any vendor.
1. What exactly does it write into the record, and who approves it? Ask to see the approval step. If the draft can reach the record without a human reading it, that is a different product.
2. Does it write into your specialist practice management software, or beside it? If the draft has to be moved from one system into another, that step stays in someone’s day.
3. How do you obtain and record patient consent? Ask whether the tool supports recording consent, or whether that sits entirely with your reception process.
4. How does it handle accents, background noise and specialty terminology? Ask for a trial in your own rooms, with your own patient mix.
5. Who is accountable for the final note? The answer is always the practitioner. A vendor who suggests otherwise has told you something important.
6. Has the vendor assessed its intended purpose against the TGA definition? Ask what their process is when features change, not just what the answer is today.
Download our checklist below:
Frequently asked questions
What is an AI clinical scribe?
An AI clinical scribe is software that captures a consultation, transcribes it, and drafts a structured clinical note for the practitioner to review, edit and approve. It does not create the record on its own. The clinician approves what enters the patient file.
Are AI clinical scribes approved by the TGA?
It depends on the product. The TGA regulates a digital scribe as a medical device only where its intended purpose meets the medical device definition, and products limited to transcription, summarising and administrative drafting generally sit outside it. Ask any vendor directly how they have assessed their product, and how they reassess when features change. That includes us: ask, and we will walk you through how xScribe is scoped and how each new feature is reviewed against the TGA’s definition.
Do I need patient consent to use an AI clinical scribe in Australia?
Yes. Ahpra expects practitioners to be transparent with patients about AI use and to obtain informed consent. That means the patient understands the consultation is being captured, knows what happens to the data, and can decline without it affecting their care.
Is an AI clinical scribe the same as dictation software?
No. Dictation software converts what you say into text, so you are still composing the note. An ambient AI scribe captures the two-way consultation and produces a structured draft against your template, which you then review and approve.
What is an ambient AI scribe?
An ambient AI scribe captures the consultation in the background, without the practitioner dictating or narrating. The term describes the capture method rather than a different category of product.
Who is responsible for errors in an AI-drafted clinical note?
The practitioner. Ahpra’s guidance is explicit that practitioners remain responsible for any AI used in their practice, and that TGA approval of a tool does not change the obligation to apply human oversight and judgement to every output.
Does an AI clinical scribe work with specialist practice management software?
That varies considerably. Some scribes integrate with clinical documentation software through an interface, some require copy-and-paste, and some, like xScribe, are built into the practice management platform itself, so once the clinician approves the draft it goes into the record without re-keying.
Can an AI clinical scribe produce documents other than the clinical note?
Yes, depending on the product. In Xestro, practices link their own document templates to xScribe, so one captured consultation can draft a clinical note, a medical certificate, a clinical summary or a compensation report from what the practitioner stated. Every document is reviewed and approved by the clinician before it is used.
Book a Xestro demo
If you are evaluating an AI clinical scribe for your practice, the most useful next step is to see one working inside a full practice management system rather than in isolation. Book a personalised Xestro demo and we will take you through xScribe in the context of your specialty, your templates and your correspondence workflow.
If you are already using Xestro and want to build confidence with the system, explore our training options. Our sessions are designed to help practices use Xestro effectively and with confidence.