The risk of a fluent guess
A general-purpose language model can write a plausible-sounding clinical note in seconds. It will also, when evidence is thin, produce a confident-looking drug dose, procedure code, or diagnostic rationale that has no basis in the record. In a dental practice, a fabricated dose or a wrongly coded claim is not a benign error. It is a medico-legal problem and a rejected medical-aid claim.
ProClinic is designed around a different premise: the AI is a fast, tireless colleague, but never the final source of truth. Every clinical claim in a ProClinic output is anchored to a knowledge base that clinicians have reviewed and signed off on.
What “grounded” means in practice
When ProClinic drafts a note, plan, prescription, or claim, it draws from a curated body of clinical content maintained under clinical governance — not from whatever the model happened to memorise during training. If the grounding is there, the output is generated. If it isn't, ProClinic says so and hands the decision back to the clinician, rather than filling in the blank.
The result is the same discipline a careful clinician already applies at the chair: cite what you know, say when you don't, and never sign what you can't defend.
Why this matters for the practice
Medico-legally. Every clinical claim in a generated record is defensible. If a note is challenged months later, the audit trail shows exactly what grounded the generation, and when.
Operationally. Coding suggestions are justified, not asserted. A rejected claim can be investigated in minutes, not weeks — and the same grounding shortens the path back to a corrected resubmission.
Clinically. Prescriptions are checked against the patient's allergies and interactions before the clinician signs. ProClinic doesn't just draft a script; it surfaces the reason to think twice before signing one.
The trade-off, honestly stated
Grounding is a discipline, not a feature. It is slower to add new clinical content than it would be to let the model improvise. That is the point. The payoff is a system whose outputs a practising dentist is comfortable signing — every time, at the chair.
Dr Louis Trichardt is a practising dentist in Cape Town and Co-founder & CEO of ProClinic AI.