AiOS→HR / People→Professional Services→Attract & Hire→HR05
Interview Debrief & Consolidation
Consolidate interview feedback into an evidence-backed debrief with areas of agreement, open concerns, and a hiring recommendation.
Evidence-backed debriefs reduce panel bias, surface unresolved concerns early, and produce a defensible hiring recommendation that stands up to later review.
GenAI Impact
46%
Faster
7.3
Hours saved
15.6
Hours without AI
Based on: 5 candidates with 3 interviewer scorecards each
Criterion-by-criterion cross-panel alignment with mandatory convergence classification ensures every interviewer's evidence is visible in the hiring recommendation, preventing the manual-debrief failure where dominant voices overshadow dissenting assessments.
Governed prompts with verification checkpoints prevent GenAI from fabricating false panel consensus where interviewers disagree, while data handling controls stop candidate scorecard PII from leaking into unapproved tools.
Before You Start
This workflow processes interviewer scorecards containing candidate performance assessments and panel member evaluation notes. Do not paste raw scorecards or personally identifiable candidate details into public or unapproved GenAI tools.
GenAI may misattribute evidence to the wrong interviewer or fabricate consensus where disagreement exists. Verify every cited quote maps to the correct scorecard and that disagreement areas are faithfully represented before finalizing.
Who's Involved
Recruiter
Collects interviewer scorecards, runs the debrief consolidation workflow, and distributes the final summary.
Hiring Manager
Reviews the consolidated debrief for accuracy, resolves disagreements, and approves the hiring recommendation.
Interview Panelist
Provides completed scorecards and clarifies feedback when the debrief flags ambiguous evidence.
Execution Steps
Before you start
Inputs
Prompt
Extract structured evidence from interviewer scorecards
CONTEXT You will be provided with the following source documents: 1. Structured Interview Guides 2. Interviewer Scorecards 3. Interview Format Guidelines TASK For each interviewer, extract the key evidence they recorded against each evaluation criterion. Produce an Individual Evidence Summaries document that preserves the interviewer's own wording and ratings. OUTPUT FORMAT Use a top-level markdown heading for each interviewer (by role label, e.g., "Interviewer 1 — Technical Lead"). Under each interviewer heading, list each evaluation criterion as a subheading. For each criterion, include: - **Rating:** The rating given (e.g., Strong / Partial / Weak) - **Key Evidence:** One to three verbatim or near-verbatim quotes from the scorecard - **Interviewer Notes:** Any additional observations the interviewer recorded CONSTRAINTS Do not paraphrase evidence in a way that changes its meaning. Do not infer ratings where the scorecard is blank — flag missing ratings as "Not Rated." Do not include personally identifiable candidate information beyond what appears in the scorecards.
Outputs
Verification: Verify extracted evidence matches the original scorecards and no ratings were fabricated for blank entries.
Before you start
Inputs
Prompt
Cite and align evidence per criterion across interviewers
CONTEXT You will be provided with the Individual Evidence Summaries extracted from each interviewer's scorecard and the Structured Interview Guides that define the evaluation criteria. TASK For each evaluation criterion, consolidate the evidence and ratings from all interviewers into a single cross-panel view. Produce a Cross-Interviewer Evidence Map that shows where interviewers converge and diverge on each criterion. OUTPUT FORMAT Use a markdown heading for each evaluation criterion. Under each heading, create a table with columns: Interviewer, Rating, Key Evidence Cited, Notes. Below each table, add a one-sentence convergence summary stating whether the panel broadly agrees, partially agrees, or disagrees on this criterion. EXAMPLE ## Technical Problem-Solving | Interviewer | Rating | Key Evidence Cited | Notes | |---|---|---|---| | Interviewer 1 — Technical Lead | Strong | "Described a systematic root-cause analysis approach" | Noted speed of resolution | | Interviewer 2 — Team Lead | Partial | "Solved the problem but skipped documentation" | Flagged process gap | **Convergence:** Partial agreement — technical capability confirmed, process discipline disputed. CONSTRAINTS Do not merge or average ratings across interviewers. Do not omit any interviewer's evidence even if it appears redundant. Do not introduce criteria not present in the Structured Interview Guides.
Outputs
Verification: Verify no interviewer's evidence was omitted and convergence summaries accurately reflect the ratings shown.
Before you start
Inputs
Prompt
Outputs
Verification: Verify disagreement classifications match the actual rating spread in the evidence map.
Before you start
Data Handling: Do not paste candidate personal contact details or compensation expectations into the prompt when adding context to concerns.
Inputs
Prompt
Outputs
Verification: Verify the AI did not fabricate concerns for criteria where all interviewers agreed.
Before you start
Inputs
Prompt
Outputs
Verification: Verify the hiring recommendation is supported by cited evidence and does not contradict the panel's recorded disagreements.
Before you start
Inputs
Outputs
Reference
Guardrails
- Evidence-Only Assessments — Every claim in the debrief must trace to a specific interviewer's scorecard — reject any AI-generated statement that cannot be sourced to an input document.
- Preserve Interviewer Voice — Use near-verbatim quotes from scorecards when citing evidence to prevent the AI from smoothing over nuance or reinterpreting feedback.
- Transparent Disagreement — Never merge conflicting interviewer ratings into an average or consensus — present each position with its supporting evidence so reviewers see the full picture.
- Recommendation Traceability — The hiring recommendation must cite at least two specific strengths and any unresolved concerns — a recommendation without cited evidence is incomplete.
Pitfalls
- Pasting full interviewer scorecards with candidate personal details into an unapproved GenAI tool without redacting sensitive information.
- Accepting the AI's convergence summary without verifying it against the actual rating distribution in the evidence map.
- Allowing the AI to fabricate consensus language when interviewers clearly disagreed on a criterion.
- Using the AI-generated hiring recommendation as the final decision without the hiring manager reviewing the underlying evidence.
- Including specific candidate compensation or personal data in the prompt context when generating the debrief summary.
Definition of Done
- The Individual Evidence Summaries contain extracted evidence for every criterion from every interviewer with no fabricated ratings.
- The Cross-Interviewer Evidence Map shows a complete comparison view with convergence summaries for each criterion.
- The Agreement and Disagreement Analysis correctly classifies every criterion and cites supporting evidence.
- The Risk and Concerns Register lists all unresolved items with severity, source evidence, and suggested resolutions.
- The Interview Debrief Summary contains a hiring recommendation that cites specific evidence from the panel's assessment.
Unlock the Full Library
Get full access to all prompts, execution steps, and downloadable examples — for this playbook and the rest of our GenAI capability framework — AGASI AiOS.
We'll send a magic link — no password needed.
AGASI AiOS · HR05 v1.0 · Apr 7, 2026