AI Interviews · Methodology
Structure is the fairness feature.
Decades of selection research agree on one thing: structured interviews predict performance better and discriminate less than free-form conversations. ELAREH encodes that structure into the product so it cannot erode under time pressure.
Four rules
Enforced by the product,
not the calendar
Every candidate for a role receives the identical interview — deviations are impossible, not discouraged.
What a 3 and a 5 look like is defined before anyone interviews — not reverse-engineered after a likeable candidate.
Each score cites the transcript passage that earned it — reviewable in one click.
AI drafts; a named reviewer confirms; disagreements are logged and reported.
Rubric anchor · “Discovery questioning”
Written, not improvised
Depth varies.
The rubric never does.
Follow-up questions are drafted into the template at different depths when it's built — nothing is generated live during your interview. What counts as a 5 was locked before the first interview ran.
Auditability
Every interview leaves a paper trail
Transcript, rubric version, scores, reviewer identity and any overrides — exportable per candidate. If a decision is challenged, the evidence is one click away.