Is Your Practice Ready for AI?

A full policy document is not a precondition, but a short written statement of boundaries genuinely is. One page covering what AI may do, what it must never do, who approves suppliers, and how patient information is handled is enough for most small and medium practices, and it can be expanded later. The value is less in the document than in the conversation that produces it, because that is where clinical and administrative staff discover they had different assumptions.

Scope creep across the clinical boundary. A system introduced to book appointments gradually starts answering questions about symptoms, medications or results because patients ask and the system is capable of responding. Without an explicit boundary and a tested escalation path, this happens incrementally and nobody notices until something goes wrong. The second largest risk is privacy: engaging a supplier without establishing where health information goes, which remains your accountability under the Privacy Act regardless of any contract.

Administrative automation such as appointment booking, reminders and call handling does not ordinarily require notification. However, practitioners remain bound by their professional obligations and codes of conduct regardless of the tools a practice uses, and software that makes or influences clinical decisions may fall within the Therapeutic Goods Administration’s regulatory framework for software as a medical device. Where a tool touches clinical decision-making rather than administration, take specific advice rather than assuming. This is general information and not legal or regulatory advice.

Fix the part the automation will actually touch, and do not let the rest delay you. If you are automating appointment reminders, valid mobile numbers and duplicate patient records are the things that matter; the tidiness of clinical notes from 2019 is irrelevant to that project. A targeted clean-up of the specific fields involved usually takes days rather than months, and it prevents the most common early failure, which is automated messages going to contact details nobody has updated in years.

Give them a fast, obvious route to a person and make sure it works. In practice the proportion of patients who object is smaller than most practices expect, and it drops further when the system is genuinely useful, patients care considerably more about being able to book quickly than about who books them. What generates complaints is not the AI itself but being unable to escape it. Test that escalation path specifically, including out of hours, and brief reception on what arrives escalated and why.

No. The assessment runs entirely in your browser, nothing is transmitted or stored on our side, and there is no email gate. You can screenshot the result for a practice meeting if that is useful. Do not enter any patient information. The questions ask only about practice systems and processes, and no free-text patient data is requested or accepted anywhere in the tool.