Jurisdiction accuracy is the hard part of health-professions AI

Generic clinical content fails quietly. A scenario that lets a practical nursing student initiate an intervention their provincial scope reserves for an RN does not look wrong on screen — it looks plausible, which is exactly the problem. The student rehearses a boundary they will be tested on and disciplined against, incorrectly. Faculty then spend remediation hours undoing what the tool taught.

Health professions are regulated jurisdiction by jurisdiction. The same role carries different titles, different regulators, different entry exams, and different scopes of practice across provinces and states. Ontario's Registered Practical Nurse is Alberta's Licensed Practical Nurse; the practical-nurse entry exam is the REx-PN in two provinces and the CPNRE everywhere else; Alberta regulated Health Care Aides in 2026 while most jurisdictions still have not. A platform that ignores this is not simplifying — it is wrong in a way that is expensive to detect.

This is why we treat jurisdiction data as a first-class engineering surface rather than marketing copy. Allied Health Sciences models 57 roles across 64 jurisdiction regions, each carrying its own regulator, professional title, and entry-to-practice examination, and grades simulated practice against the scope that actually applies where the student will be licensed. The coverage is browsable at the licensure map, and every role page localizes to the student's province or state.

The discipline this imposes is mostly editorial: regulator names change, professions gain regulation, exams get replaced. Owning the content model — rather than wrapping a general-purpose model and hoping — is what makes those updates a data change instead of a retraining project. That architectural choice is the same one described in owning the stack healthcare ai, applied to the least glamorous and most consequential layer: getting the rules right, everywhere a student practises.