Every clinic already runs software. It is just made of paper: an appointment book, a stack of case sheets, a cash register and a recall diary that lives mostly in one person’s head. Those four systems work, in the narrow sense that the clinic functions. What they cannot do is agree with each other.
Four registers, four versions of the truth
Ask a practice what a patient owes and you will get an answer from the cash book. Ask what was done and you will get an answer from the case sheet. The two are reconciled by a human being at the end of the day, which means they are reconciled approximately, and only on the days that end early enough.
Software that digitises each register separately reproduces the problem with better fonts. The disagreement moves from paper to screens, and now it is harder to fix because nobody can see all four at once.
What changes when there is one record
When treatment, billing, prescription and recall are facets of a single record, a class of work disappears rather than getting faster. Nobody reconciles anything. The bill is a view of the treatment. The recall is a consequence of the procedure. The patient’s copy is the same object the clinic wrote, not a photocopy of it.
- The invoice cannot disagree with the chart, because it reads from it.
- The recall date is set by the procedure, not by someone remembering.
- The patient’s app shows the plan the clinician actually approved.
- A second clinician sees the whole history, not the part that was transcribed.
The ownership question
Once you accept the single record, you have to answer who it belongs to. The convenient answer for a software company is: us. Data that cannot leave is retention you never have to earn. We think that is both wrong and, in the long run, commercially fragile.
The clinic is the custodian of the record. The patient is its owner. The software vendor is neither.
So we built export before we built lock-in. A clinic can take its patients, treatments and financials out of Osdent in a standard format without asking us. A patient can carry their history to a dentist in another city. We would rather compete on whether the product is good than on whether leaving is painful.
Why this matters for AI
The intelligence layer everyone is racing to build is worth very little without a clean, longitudinal, consented record underneath it. A model that drafts a treatment plan from a fragmented chart drafts a fragmented plan. The unglamorous work — one record, correct, complete, and portable — is what makes the interesting work possible.