CentriWebDental AI helps most with the front desk and the follow-up that never quite happens: answering the phone when the team is with patients, recalls and reminders, and chasing accepted-but-unbooked treatment plans. In Australia it is not a substitute for the dentist’s judgement, and anything it writes in your name is covered by the same health advertising and privacy rules as everything else the practice publishes.
This article is for practice owners and managers. It covers where AI earns its keep, what a clinician must still do, and the rules that apply. It is general information, not clinical, legal or compliance advice.
Where a dental practice loses time and chairs
A practice has a particular problem: the people who answer the phone are also the people with a patient in front of them. Calls go to voicemail at lunch and after 5pm. Recall lists are worked when someone has a spare hour, which is rarely. A patient who was given a treatment plan, said they would think about it, and then heard nothing, is a treatment that was never done and a chair that stayed empty.
These are admin and communication problems. That is exactly where AI is lowest risk and highest return.
Dental AI for phones and bookings
An AI phone assistant can answer when reception cannot, take the reason for the call, offer real appointment times from your diary and book them, answer routine questions from information you supply (hours, parking, which health funds you are connected to, what to bring), and pass emergencies straight to a person or the on-call number. For a dental practice, the design decisions that matter are:
- Emergency handling. A caller describing facial swelling, uncontrolled bleeding, trauma or severe pain should reach a person or be given your emergency instructions at once, not enter a booking script.
- No clinical advice. It does not diagnose, suggest medication or say whether something can wait.
- Real diary, real rules. It books only appointment types that fit the time, the dentist and the patient’s status as new or existing.
- Clean handover. The reason, name and number are passed on, so the patient does not repeat themselves.
The detail of voice assistants is in AI receptionists for dental practices. For missed no-shows, see how to reduce appointment no-shows.
Recalls and reminders that actually go out
Six-month recalls are repeat, predictable and mostly patient-driven once prompted. A system that reads your practice management software, finds who is due, sends a text with a booking link and follows up once or twice if there is no reply will recover appointments the manual list never reaches. The AI part is the wording and the replies (“can I come on Thursday?”). The dependable part is the scheduled automation.
Respect consent. Reminders for existing patients about their own care are different from promotional messages, and marketing needs consent and an unsubscribe. See SMS marketing rules in Australia and talk to your compliance adviser about which messages are which. Our patient recall and reactivation guide covers the process side.
Treatment plan follow-up
When a patient accepts or considers a plan and does not book, a gentle sequence helps: a message a few days later asking if they have any questions, an offer of a call with the dentist or the practice manager, and a check-in later. The message should come from the practice, be factual and avoid pressure. It can say what was discussed. It should not argue the clinical case, claim outcomes or imply urgency that the dentist has not expressed. Any wording that describes a treatment’s benefits is advertising, so it must stay accurate.
What the dentist must still do
AHPRA’s guidance on AI in healthcare is blunt about responsibility. It says the practitioner remains responsible for delivering safe and quality care, and that practitioners must apply human judgement to any output of AI (AHPRA, AI in healthcare, opened November 2026). It also says patients should be told about AI use, with the level of explanation depending on how it is used, and that informed consent should be obtained where AI involves a patient’s personal data, ideally noted in the health record. Recording consultations needs more explanation than reading images, for example.
In practice that means:
- Clinical records, diagnoses and treatment planning stay with clinicians.
- AI-drafted notes or summaries are checked and signed off before they enter the record.
- Patients are told when AI answers calls, sends messages or records anything.
- Staff know what they may and may not paste into any AI tool.
Privacy: patient information is health information
AHPRA’s page says practitioners must protect confidentiality and privacy as required by privacy and health record legislation, including checking how data is collected, stored, used and disclosed, and whether patient data may be used to train the AI model. Health information is sensitive information under the Privacy Act, and states have their own health records laws. Ask any vendor in writing where data is stored, whether it is kept or used for training, and who can access it. Do not put patient details into a free public chat tool. Our AI and Privacy Act guide explains what to ask.
Advertising rules apply to AI-written content
Health services advertising in Australia is regulated, and the practice is responsible for what is published in its name, whoever or whatever drafted it. AHPRA’s advertising hub states that consumers should have information that is accurate, not misleading and supported by acceptable evidence (AHPRA advertising hub, opened November 2026). The hub links to separate requirements on testimonials, evidence and social media, which you should read before using AI to produce website pages, social posts, review replies or recall messages that describe treatments.
That has a few direct consequences:
- AI-written treatment pages are checked by a person who knows the rules, before publishing.
- AI is never used to write or edit reviews or testimonials, or to reply to them with claims about treatment.
- No before-and-after claims, guarantees or outcome promises appear in automated messages.
Our guide to AHPRA advertising guidelines for a clinic website goes through the practical side.
Making sure it only says what is true
An assistant that answers patient questions should work from your approved information: services, fees you choose to publish, hours, preparation instructions. It should say “I will pass that to the team” when unsure. Test it with questions you would not want answered, such as whether a symptom is serious or how much a procedure will cost for someone’s specific case. The method is in how to reduce AI hallucinations.
A worked example: the Monday recall list
Picture a practice with three hundred patients due for a check-up in the next two months. Working that list by phone takes a receptionist hours that never exist. An automated sequence sends each patient a text with a booking link, sends one reminder a few days later and flags anyone who replies with a question. The receptionist deals only with the replies. If the practice recovers even a handful of extra booked appointments from a list like that, the value is clear; the exact number depends on your patients, so measure it for your own practice rather than trusting a figure from elsewhere.
Questions to ask any vendor
- Which practice management systems do you connect to, and what exactly is read and written?
- Where is patient data stored, who can see it, and is it used to train any model?
- What does the assistant do when a caller describes an emergency?
- Can I see and edit every message template before it goes out?
- Who looks after it when something breaks, and how quickly?
- Can I export my data and end the arrangement without being locked in?
Get the answers in writing. If a vendor cannot answer the privacy questions clearly, that is the answer.
A sensible order of work
- Missed-call text-back or after-hours phone answering with emergency routing.
- Recall and reminder automation tied to your diary.
- Treatment plan follow-up with clinician-approved wording.
- Admin summaries and internal drafts, reviewed before use.
CentriWeb builds and looks after AI phone and messaging systems for health practices that connect to the real diary and hand over to a person. If your practice management software already sends reminders and recalls, switch those on first and add AI only where a gap remains.