One made-up application essay. A summary agent, an evidence agent and an interview-question agent work in parallel; a verification layer tests every output against the applicant's own sentences and sends failures back. Watch the first draft get caught.
A recruitment-analyst summary; a highlight agent that maps each competency to the sentences that support it; a question agent that writes main questions, a predicted answer and follow-ups across behavioural areas.
Agents return sentence numbers, not quotes; the evidence shown is re-assembled from the applicant's original sentences and positions, so a fabricated quote cannot reach the recruiter.
A separate set of checks scores every output for tone, for claims the essay does not support, and for key experiences left out, and regenerates until it passes. Inappropriate follow-ups are filtered by policy. A human makes the decision.