AI in dentistry: What’s real and what’s noise?

By executive editor Andy Sloan

Most of the AI conversation in dentistry is about what the technology can do. The more useful question is whether your business is ready for it.

There’s plenty to be excited about. Imaging tools that flag pathology more consistently than the human eye. Engagement platforms that spot a patient about to lapse. Scheduling systems that learn from cancellations. Billing tools that catch anomalies before they turn into write-offs.

Most of it works, in a well-run business. That condition matters more than the technology sector likes to admit.

Where it falls down

A mid-sized group piloted an AI follow-up tool. It picked up patients who hadn’t answered recall, flagged lapsed treatment plans and sent personalised reminders at sensible intervals. In the pilot, conversion went up.

Three months into the rollout, adoption was patchy. Some practices had built it into their day and were seeing results. Others had quietly stopped using it, out of habit or scepticism, with nobody helping them make it work. Two were running it alongside their old manual recall process, so the work was being done twice.

When the group looked into it, the problem was the workflow underneath. The practices that used it well had someone who owned follow-up, a clear process for responding, and a practice manager who understood what the tool was for. The others had none of that. The AI didn’t create the gap. It showed it up.

Buying the tool was the easy part. Redesigning the process so the tool had something consistent to work with was harder and worth more, and it should have happened first.

What AI is good at

None of this is an argument against AI. The tools are good and getting better quickly. It just helps to be precise about where they help.

AI is good at consistency. Give it a clearly defined process and it will run it faster and more reliably than people can, at scale. Recall, treatment plan follow-up, appointment confirmations and clinical notes all benefit when the process underneath is sound.

It’s good at showing you variation. In a multi-site group, some of the most valuable uses never touch a patient. Which sites are slipping on recall? Which associate completes treatment plans at a very different rate from colleagues? Which practices have documentation gaps that will matter when the CQC visits? AI can show you that almost as it happens, instead of in a quarterly report that arrives too late to act on.

And it helps with decisions where the criteria are clear, such as scheduling, demand forecasting and pricing, as long as there’s enough history and the business can act on what it learns.

What it can’t do is fix ambiguity. An unclear process just becomes a faster unclear process. A badly designed operating model gets scaled, noise and all. And if management won’t act on what the data shows, better data changes nothing.

Where the CMA comes in

The CMA’s study of private dentistry gives all this a deadline. Its July update set out what it is examining: how patients find and compare information, how practices compete, whether there is pressure selling or overtreatment, and whether complaints routes work. Its emerging thinking is expected in November, and the final report is due by 4 March 2027.

I don’t expect a new enforcement regime. I expect pressure for consistency: prices patients can compare, plan terms they can understand, and treatment planning and consent records that stand up when someone asks to see them. Those are the areas where AI tools can help most, if the process underneath is right.

That leaves groups with two paths. One uses AI to make records and patient communication more consistent before the CMA asks. The other adds AI-branded tools on top of the same inconsistent processes, and ends up with a faster version of exactly what the regulator is looking for.

Which path a group takes depends far less on the software it picks than on whether its leaders know what they are building.

The question worth asking

When someone pitches you an AI tool, skip “what does it do?” Ask what your business needs to look like before it will work.

That moves the conversation from features to how you operate. Is recall actually written down? Does someone own compliance? Is your data clean enough to use? Does the management team have time to act on what it learns?

Groups that answer those questions first will find adoption fairly straightforward, and the gains will last. Groups that skip them will add another tool to the list of things that were meant to help and never quite did.

AI won’t decide which dental groups win over the next five years. Well-built businesses will, and AI will make them stronger. For everyone else, it’s an expensive way to find out the foundations were never there.

Andy Sloan is the founder of Sloan Strategic Advisory and Executive Editor (Dental) at Dentist Times. He has spent more than 15 years at the commercial centre of UK dental consolidation, including board roles at Portman Dental Care and the Dental Buying Group, and writes The Accidental Dental Strategist.