Automated first-round voice screening that scores every candidate against the same rubric
What it does
An AI-powered voice interview agent that automates first-round candidate screening and evaluates every candidate against the same structured scoring rubric. HR teams define a job's requirements and a weighted scoring rubric, then add candidates with their CVs; each candidate gets a unique, single-use interview link sent from the employer's own Gmail account. After consenting to voice recording and AI evaluation, the candidate has a natural voice conversation with the AI interviewer, which asks questions drawn from both the job requirements and the candidate's CV, so no two interviews follow an identical script. The transcript is then evaluated against the predefined rubric to produce a score and written justification per criterion, strengths, potential concerns, an overall hiring recommendation, and a weighted score out of 100, letting candidates for the same role be ranked side by side on the same methodology.
HR teams define a job's requirements and a weighted scoring rubric before screening starts.
Candidates are added with their CVs, each receiving a unique, single-use interview link sent from the employer's own Gmail account.
Consent & Personalized Voice Interview
Candidates explicitly consent to voice recording and AI-based evaluation before the interview begins.
The AI interviewer holds a natural voice conversation over the OpenAI Realtime API and browser-native WebRTC, asking questions drawn from both the job requirements and the candidate's own CV rather than a fixed script.
Automated, Structured Evaluation
The transcript is evaluated against the predefined rubric using OpenAI structured outputs.
Produces a score and written justification per criterion, candidate strengths, potential concerns, and an overall hiring recommendation.
Deterministic Scoring & Ranking
The AI evaluates each response, but the final weighted score out of 100 is calculated programmatically from the employer-defined rubric, keeping the scoring arithmetic deterministic and consistent.
Candidates applying for the same role are ranked side by side using identical scoring methodology.
Impact
Automated first-round candidate screening with a natural, personalized voice interview instead of a fixed script.
Applied one structured, weighted scoring rubric consistently across every candidate for a role.
Kept scoring arithmetic deterministic by computing the final weighted score programmatically rather than trusting model output directly.
Delivered per-criterion scores, justifications, strengths, concerns, and an overall hiring recommendation for every interview.
Enabled direct side-by-side ranking of candidates applying to the same role.
