A dental practice runs on two things: clinical time in the chair, and a surprising amount of writing. Recall reminders, referral letters to specialists, and treatment plan letters that explain what a patient needs and what it will cost. Most of this lands on a practice manager or front desk coordinator who is also answering the phone, chasing health fund pre-approvals, and rebooking cancellations. It is steady, repetitive work, and it is exactly the kind of work Claude is good at.
Claude is the AI assistant built by Anthropic. For an Australian dental practice, the useful way to think about it is as a fast, careful writer that already knows your tone and your templates. It does not replace your team or your clinical judgement. It takes the first draft off their plate so they can check and send rather than start from a blank page. This guide walks through three jobs where that saves the most time: recalls, referrals, and treatment plan letters.
Why dental admin suits an AI assistant
The writing a practice produces is high volume, formulaic, and personal all at once. A reminder for a six month checkup follows the same shape every time, but each one should use the patient's name, their last visit, and the right tone for someone who is nervous versus someone who is two years overdue. Producing that by hand is slow. Producing it with a plain mail merge feels robotic. Claude sits in the middle: it writes each message individually from a short brief, so the letter reads like a person wrote it, without a person writing every word.
There is a second reason it fits. A typical front desk coordinator in Sydney costs a practice around $65,000 a year once super is included. If even six hours a week of that role goes to drafting routine letters and reminders, that is roughly $9,000 of salaried time a year spent typing. Redirect a good chunk of it and you either free the person for higher value work or you handle more patients without adding a head.
Recalls: turning an overdue list into real outreach
Every practice has a recall list, and most have a longer overdue list than they would like to admit. Those patients are not gone, they have just fallen off the radar. The barrier is rarely the patient, it is the effort of writing something that does not read like spam. Claude turns a plain export of overdue patients into individual, friendly reminders you can review and send.
A practical setup looks like this. You give Claude your recall templates and a short description of your practice voice. Then for each batch you hand it the fields that matter, and it writes the messages:
Patient first name and the treatment they are due for, so a checkup reminder reads differently from a hygiene visit or an orthodontic review.
How overdue they are, so a patient three months late gets a gentle nudge while one two years late gets a warmer note about how good it would be to see them again.
The channel, so an SMS stays short and an email can carry a little more detail and a booking link.
Any clinical note worth a mention, such as a filling that was flagged to watch, phrased in plain language rather than dental shorthand.
The revenue case is easy to see. If a reminder run brings back thirty patients in a quarter who would otherwise have drifted, and each books a checkup and clean worth around $300, that is $9,000 a quarter you were leaving on the table. Do that every quarter and it becomes a real line on the P&L, from work that used to sit at the bottom of someone's list.
Referral letters that do not eat a clinician's evening
Referrals to an oral surgeon, periodontist or orthodontist need to be accurate, professional, and reasonably quick. Dentists often write them at the end of the day from memory, or copy an old one and edit it. Claude drafts a clean referral from the clinical facts you give it, keeping the structure a specialist expects: reason for referral, relevant history, current findings, and what you are asking the specialist to do.
The clinician stays in control. You provide the facts, Claude assembles the letter, and you read it and correct anything before it goes out. Because you are checking a draft rather than composing from scratch, a letter that took fifteen minutes can take three. Across a busy week of referrals, that time adds back up to real clinical capacity.
Treatment plan letters patients actually understand
This is where good writing changes whether a patient says yes. A treatment plan letter has to do two hard things at once: explain the clinical need honestly, and set out the cost without burying it or making it feel like a sales pitch. Patients decline treatment they do not understand. Claude helps you write plans in plain English, with the numbers laid out clearly.
Say a patient needs a crown at around $1,800 and has been quoted for an implant closer to $5,000. Claude can turn your clinical notes into a letter that explains what each procedure is, why it is recommended, what happens if it waits, and how the fees break down, including where a private health fund rebate might apply. You still set the clinical content and the prices. What you get is a version a worried patient can read at the kitchen table and follow.
Explain the problem in the patient's terms first, then the recommended treatment, then the alternatives including doing nothing.
Show the fee for each item and the likely out of pocket cost after a typical health fund rebate, clearly marked as an estimate.
Note the relevant item numbers so the patient can check their cover, without turning the letter into a form.
Close with a clear next step and who to call, so the letter ends in a booking rather than a shrug.
Keeping patient information safe
Dental records are health information, which counts as sensitive information under the Privacy Act 1988 and the Australian Privacy Principles. Practices are also bound by their obligations to AHPRA and the Dental Board of Australia. Using an AI assistant does not change those duties, so the sensible approach is to build the workflow around them from the start.
In practice that means being deliberate about what you put into any tool. For most letter writing you do not need a full identified record. You need a first name, the treatment, and the relevant clinical facts, and you can keep out anything that is not required to write the message. For a practice that wants a firmer footing, the enterprise versions of Claude offer stronger data handling commitments, and that is worth getting right before you scale a workflow across your whole patient base. It is the sort of thing worth a short conversation before you begin, not after.
A sensible first step
You do not need to change your practice software or sign up for a big platform to start. Pick the one job that hurts most, usually the overdue recall list, and run a single batch with Claude drafting and a team member checking every message before it goes out. Measure two things: how much time it saved, and how many patients came back. That gives you a real number for your own practice rather than a vendor's promise.
If you would like help setting that up, or working out which of these three jobs to start with, book a short call with us. We work with Australian practices to put Claude to work on the admin that is quietly costing them time and patients.
