Prototype — proposed scoring & evaluator additions to the RMA form builder (app.hiredata.com). Per-question scoring (expected answer, Must-have / Nice-to-have, AI judge note) lives in the form definition. On completion the same trigger runs both the AI Summarizer (writes candidate notes) and the new AI Evaluator, which judges each answer and writes the outcome to a chosen output field — here the Carerix Match: Stage, mapped to Rejected / Valid with Restrictions / Validated per Randstad PT's logic. The open-text question is judged by the LLM (mocked here). Nothing is sent to Carerix or HireData.