AI Won’t Fix a Broken Dental Practice

The ADA’s 2026 research found that 43.3% of responding dentists used AI for at least one practice task. The most common uses were imaging and diagnostics, insurance verification, and explaining clinical findings to patients. Fewer than 5% said they used AI to suggest treatment. So dentists are not waiting for software that plans care. They are using AI in the places where it can take friction out of the day.

That is probably the right approach, but it comes with a catch. A practice with poor workflows can buy AI and still have poor workflows. When you automate a broken process, you get the same broken process at a faster speed. The tool cannot fix a step that should not exist in the first place.

The order to work in

A better approach follows four steps in order. First, define the problem in plain terms. Second, measure what that problem costs you today in staff hours, errors, and lost revenue. Third, redesign the workflow so the waste comes out before any software goes in. Only then should you ask whether AI improves the numbers on the cleaner process.

The third step is the one most practices skip. Often the redesign alone saves much of the time, and the AI tool turns out to be a smaller win than the vendor promised. That is a good outcome, because you found the savings without taking on a new monthly bill.

What to measure before you buy

Insurance verification is a good place to start, because the hours are easy to count. Have the front desk log the time they spend on it for two normal weeks. Note how many checks they run, how long each one takes, and how often a claim later comes back because the check was wrong. Those three numbers give you a baseline to measure any tool against.

Then put the same questions to every AI product you look at. Ask how much time it will save, and whether that time turns into fewer paid hours or into other useful work. Ask what it costs to subscribe, set up, and train the team. Ask how often a person must step in to check or fix its work, and what its error rate is in offices like yours. Service quality matters too, so ask how patients will notice the change.

A worked example

Here is an example with round numbers. Say your team spends 30 hours a week on insurance verification. At a loaded cost of $30 an hour, that is $900 a week, or about $46,800 a year. An AI tool claims it can cut that work by two thirds. It costs $800 a month, which is $9,600 a year, plus $3,000 to set up.

On paper, two thirds of $46,800 is $31,200 in saved time. But staff still have to review the tool’s output, and say that review takes 6 hours a week. That leaves 14 hours saved instead of 20, which is worth about $21,800 a year. After the $9,600 subscription, the net gain is about $12,200 a year, or $9,200 in the first year once you pay for setup. That is still a fair return, but it is less than half of the $21,600 net the vendor’s math implies.

The savings are only real if the hours go somewhere. If the 14 hours come off payroll or move to work that brings in money, such as recall calls, the gain shows up in profit. If the hours just become slack time, the practice pays for the tool and keeps the same costs.

Treat AI like any other investment

AI should go through the same review as a new hire or a new piece of equipment. Set a test that runs for 90 days, with one clear measure of success such as hours saved per week or claims denied per month. If the tool hits that mark, keep it and track it each quarter. If it misses, cancel it before it becomes one more subscription nobody checks.

The aim is a better practice, and AI is one of the tools that can help you build it. The owners who get the most from AI will be the ones who measured the work first and know where the tool saves money.

What to do next

Pick the one office task that takes the most staff time each week. Log the hours and errors for two weeks so you have a baseline. Look for steps you can cut before you add any software. Then fill in the worksheet below for any AI tool you are weighing, and set your success measure before you sign the contract.