Override the AI rank from the finalist card, Override. The override prompts you to log a rationale before it commits. The rationale is logged to the audit trail and to the candidate-side decision note.
When to override.
- The evidence span on a top-three competency does not support the score. The transcript reads weaker than the rubric called.
- A deal-breaker check was passed by the rubric but the candidate-side answer flagged a hidden constraint (right-to-work change, salary expectation drift, location change).
- Domain context the rubric cannot model: the candidate is fluent in a specific tool you require, the candidate has worked at a directly relevant prior company, and the rubric did not weight that strongly enough.
What the override does.
It changes the rank within the finalist set, and it logs a structured rationale. The candidate-side decision note shows the rationale (paraphrased to remove protected-characteristic mentions, if any). The override is visible to admins and owners on the audit log. Override patterns across the workspace are summarised on the workspace fairness report.
Overriding does not adjust the rubric weighting on the role. To change the rubric, edit it on the role page; that does not retro-rank existing candidates.
What to do next: override sparingly. Two overrides per role is healthy. Five overrides per role is a sign the rubric needs an edit.