A Second Brain for the Ortho Cases That Never Started

A treatment coordinator in a two-doctor Lima practice can tell you almost everything about the families who started this month. Ask about the ones who said yes in March and never came back for bands, and the answer gets vague. The file is somewhere. The reason is in somebody's memory, or a sticky note, or a phone log nobody has opened since spring.

That pile is the most expensive thing in an orthodontic office, and it is invisible.

The gap is real and it is measured

Planet DDS published its 2026 Dental Industry Outlook this spring, built on more than 2,500 orthodontic practices running Cloud 9. Average case acceptance came in at 68.3 percent for 2025, up from 64.4 percent the year before. Case completion came in at 47 percent, up from 42 percent.

Those are two different measures and they are not a subtraction problem. But the direction tells you something anyone who has worked an ortho front desk already knows: a lot of plans that get a yes in the consult room never turn into a banded case. For a practice averaging 600-plus new patient exams a year, even a small slice of that is real money walking out the door in a minivan.

The reason those plans stall is almost never mysterious. It is just scattered. The financial policy lives in a PDF. The insurance verification note lives in the practice management system. What the mom actually said — that her deductible resets in January, that dad handles the money and wasn't at the appointment — lives in a coordinator's head or a free-text field nobody searches. When you follow up four months later, you are starting from zero.

What a Second Brain does with that

A Second Brain is an AI knowledge system built on your documents and nothing else. The build starts with a spec: what goes in, who is allowed to ask what, and which questions it has to answer on day one. Then an ingestion pipeline pulls the material in, and your team asks questions in plain English. The answers come back with the source attached.

For an orthodontic practice chasing unstarted plans, the material worth ingesting is unglamorous:

• The financial policy, payment plan terms, and down payment exceptions the doctor has approved before
• Ohio Medicaid orthodontic criteria and your own verification notes on what got approved and what got denied
• Consult summaries and the treatment presentations you actually handed parents
• The doctor's phasing rules — which cases wait on eruption, which don't
• Notes on past follow-ups: which objections came up, and what actually moved them

Then Monday morning looks different. The coordinator asks which families were presented in the last ninety days without a start date, and what the stated hold-up was. Then: what did we tell the ones who mentioned cost, and which of those eventually started? The system answers from your records and shows which note each line came from. Ten seconds to verify. No guessing about whether the AI made something up, because grounded chat that can't find an answer says so instead of inventing one.

That is the whole trick. It doesn't call anyone. It doesn't book anything. It makes the practice's own memory searchable, so the person picking up the phone already knows the history.

Keep the doctor in command

In January 2026, the American Association of Orthodontists published a position paper on responsible AI in clinical practice, built around what it calls the Human-in-Command principle: the licensed orthodontist keeps final authority, and AI functions as a supportive tool that enhances but never replaces professional expertise.

That paper is aimed at clinical AI, and a Second Brain over your administrative and financial documents isn't diagnosing anyone. But the principle travels well. Retrieval and organization are the machine's job. Deciding what to offer a family is yours. Grounded answers that cite their source are how you keep that line clean, because your team can always check the original before acting.

Two practical notes. Anything touching patient information needs a business associate agreement and real access controls, same as any other vendor. And start narrow: one folder, thirty documents, the pending-starts question. You will know inside a week whether it earns its keep.