Medication interactions, capacity checks, and safeguarding flags run on hard rules, not on a model guess.
Deterministic checks before any AI gets near a clinical decision.
A deterministic rules engine validates medication, mental capacity, safeguarding, and clinical thresholds against UK guidance before any generative reasoning runs. Rules are versioned, reviewable, and signed off by a clinician.

Medication interactions, capacity checks, and safeguarding flags run on hard rules, not on a model guess.
Every rule has a named clinical owner, a review date, and a versioned change log.
Each red flag links back to the rule, the guidance source, and the data that triggered it.
When in doubt, the engine escalates to a human, not to a confident-sounding model.