What questions does a clinic chatbot handle well?
Chat suits questions with one correct answer that you can write down in advance. Opening hours, location and parking, the new patient process, which forms to bring, telehealth instructions, accessibility, fees from your approved script and how to book or cancel all fit. If the chatbot is connected to your booking system it can offer times; if not, it can collect a structured request for staff. Start by listing the questions your reception team answers most often by phone and email. If the list is short and stable, the chatbot will have a narrow job you can test.
- Hours, location, parking and access
- New patient process and forms
- Fees and payment, from the approved fee script
- Booking, cancelling and rescheduling requests
- Telehealth connection instructions
What should a clinic chatbot never do?
A clinic chatbot should not assess symptoms, suggest diagnoses or advise on treatment. The Australian Commission on Safety and Quality in Health Care notes that an AI tool used for diagnosis, prevention, monitoring, prediction, prognosis or treatment must comply with Therapeutic Goods Administration regulation, which puts it in a different category from an administrative assistant. Medication requests also need care: the Medical Board's telehealth guidelines say prescribing for a patient without a real-time direct consultation is not good practice, and that this includes medication requests made by live chat outside a real-time consultation. Set fixed responses for these topics, direct emergencies to triple zero (000) and hand the conversation to staff.
What privacy rules apply to chat conversations?
Patients type personal and health information into chat, often without being asked. The OAIC's guidance on commercially available AI products says privacy obligations apply to personal information put into an AI system and to the output it produces, that privacy policies and notices should explain the use of AI, and that public-facing AI tools such as chatbots should be clearly identified as AI. It also recommends that organisations do not enter personal information into publicly available generative AI tools. Before launch, confirm where conversations are stored, who can read them, how long they are kept and whether the vendor uses them for any other purpose.
Do advertising rules affect what the chatbot says?
Chat replies that describe your services are part of how you promote them. Ahpra's advertising guidelines say advertising of a regulated health service must not be false, misleading or deceptive, must not offer a gift, discount or inducement without stating the terms, must not use testimonials, must not create an unreasonable expectation of beneficial treatment and must not encourage unnecessary use of health services. A chatbot that writes its own wording can drift into claims about outcomes or comparisons. Write the promotional answers yourself, have them checked against the guidelines, and stop the chatbot from improvising beyond them.
How do you tell whether chat is worth it?
Measure before you buy. Sort two weeks of calls and emails by reason and count how many are questions a chat window could answer from your website content. Check your website analytics for the pages patients visit before they call. If most routine questions are about one or two topics, a better page or an online booking link may solve them. If written questions arrive after hours and patients want an answer then, chat has a clearer case. After launch, track chats per week, the share resolved without staff, booking requests completed and calls on the same topics.
Who reads the conversations, and how quickly?
Every handover from the chatbot needs an owner. Route handovers into a shared practice queue that reception already works from, with a response standard for business hours and an automatic after-hours reply that gives your hours, emergency direction and when the message will be read. Review a sample of transcripts each week for wrong answers, missed handovers and questions the chatbot could not handle, and update its approved answers. Assign one person to approve changes so the chatbot's content stays in step with your website, fee script and phone message.