The research
Anyone can call a model.
CaseClarify pairs a state-of-the-art model with a document layer built from real dentistry — a reasoning record, a curated canon, and a misconception library. That layer is why it sounds like a colleague and not a chatbot, and why it never invents a clinical fact.
The problem, measured
The failure isn't the diagnosis. It's the translation — the moment a dentist's causal reasoning has to cross into an anxious person's understanding, and mostly doesn't. Everything below is built to carry that crossing.
1 · The reasoning record
A dentist doesn't hold facts — they hold a chain of relationships. The software's core structure is that chain, made explicit. Every fact the dentist enters is typed by its role in it, and the model is required to reconstruct the links, not recite the facts — because a fact without its because is noise to a patient.
2 · The canon
What a crown is, what a root canal feels like, the honest analogy that lands in one breath — these never change, so the model never re-derives them. They're written once, at a seventh-grade reading level, verified against general dentistry, and handed to the model as facts it may personalize but never contradict. Seventeen procedures, across restorative, cosmetic and preventive care.
3 · The misconception library
Every procedure carries the predictable ways patients misunderstand it — and the case facts decide which ones to gently repair. An asymptomatic tooth forces the film to address “it doesn't hurt, so it can wait.” A cost-worried patient triggers a calmer handling of the fee. The software chooses which gap to close; the model only writes the repair, in this patient's language. This is how it constructs an understanding instead of reciting a script.
The psychology it runs on
Picture superiority
People remember what they see far longer than what they're told. The finding is revealed as an image before it's named, so understanding feels self-generated — and self-generated conclusions are the ones patients keep and defend.
Narrative transportation
Information delivered as a story measurably lowers counter-arguing. The case is told as a six-scene film, which is why it can persuade without a single pressure tactic.
Autonomy & reactance
Perceived choice is the strongest lever on treatment acceptance. Three options in equal light — including an honest wait-and-watch — and a decide-by date the patient sets, reduce the reactance a hard close creates.
The teach-back
The oldest tool in medicine: “tell me what you'd tell your partner.” The presenter prompts it and the take-home arms the patient for it — a human confirmation that understanding actually landed.
Where the research becomes a rule
Trust is warmth times competence — and it collapses at the first exaggeration.
So the anti-pressure architecture isn't a marketing posture; it's the research made mechanical. Costs appear once, calmly, never as a rising escalation. The urgency scene is a health timeline — a validator physically rejects dollar signs from it. Every option carries an honest tradeoff, the recommended one included. The moment a patient catches one manufactured urgency, every earlier scene retroactively becomes a sales pitch. The software is built so that moment can't happen.
Sources, honestly labeled
Profession & benchmarks
ADA clinical-communication guidance; case-acceptance benchmarks from Dental Economics, Levin Group and ADA HPI (via GrowthRx); market and pricing data from IBISWorld and PracticeSignal. Industry-reported, not audited.
The decision literature
Peer-reviewed work on health anxiety, information-processing styles, shared decision-making, picture superiority, narrative transportation, and the documented gap between hearing and understanding.
Practicing dentists
Structured input from working clinicians on what they wish patients understood, where conversations break, and how the best communicators in the profession actually talk.
One promise about numbers: the benchmarks above explain why we built this. The only acceptance figures we will ever publish under our own name are the ones a real practice's own dashboard produces — counts, not percentages, a clear-eyed “no” included. We never get ahead of the study, because the study is the brand.