A plain-language definition of AI interviews — what they are, how the conversation and scoring actually work, and how candidates and recruiters get value from a structured, voice-first screen.
An AI interview is a structured job interview conducted by software instead of a human recruiter. The AI asks role-specific questions, listens to or reads each answer, and scores responses against a fixed rubric. It runs the same way for every candidate, on demand, and hands recruiters a ranked, evidence-backed summary in minutes.
The first screening conversation is the most repetitive part of hiring: the same questions, asked of every applicant, eating hours that recruiters could spend on finalists. Scheduling alone can stall a pipeline for days, and tired human interviewers drift — the 50th candidate rarely gets the focus the 5th did. An AI interview removes those frictions. Candidates take a short, structured interview whenever it suits them, every applicant answers the same questions against the same bar, and recruiters open their day with a ranked shortlist instead of a backlog of phone screens to book. Criba runs this as a ~5-minute voice-first interview and returns a Pass / Borderline / Reject result with direct quotes behind every score.
Candidates no longer wait days for a recruiter's calendar to open. An AI interview is available on demand, so applicants complete it on their own time — evenings, weekends, between shifts — and recruiters skip the scheduling back-and-forth entirely.
Every candidate answers the same role-specific questions against the same rubric. That consistency is what makes an AI interview fairer than ad-hoc phone screens, where the questions and the interviewer's attention drift from one call to the next.
Because scoring is immediate, candidates hear back in hours instead of weeks. A quicker, more respectful process protects your employer brand and keeps strong applicants from drifting to a competitor mid-pipeline.
An AI interview returns more than a score — it surfaces the direct quotes and signals behind each rating. Recruiters can see why a candidate passed or didn't, which makes decisions explainable and easier to defend.
Ten applicants or a thousand, each gets the same structured interview with the same depth. Hiring surges no longer create a screening bottleneck or force a rushed, shallow review.
By handling the repetitive first conversation, an AI interview frees recruiters to spend their judgment on finalists, hiring-manager alignment, and closing offers — the work that actually needs a human.
The role and questions are set
A recruiter defines the role and the 3-5 signals that matter for it. The AI turns those into a structured set of questions, so the interview measures exactly what the job needs rather than a generic script.
The candidate takes the interview
The candidate joins from a link — no app or scheduling required — and answers by voice or text. Criba runs a ~5-minute voice-first conversation, asking follow-ups in real time when an answer needs more depth.
The AI scores against a rubric
Each answer is evaluated against the same fixed rubric for every candidate. The AI assigns signal-level scores and captures the direct quotes that justify them, so the result is transparent rather than a black-box number.
Recruiters get a ranked shortlist
Results flow back as a Pass / Borderline / Reject shortlist with a bullet summary, signal scores, and red flags. Recruiters open their day with a prioritized list and the evidence to act on it quickly.
An AI interview replaces the first human screen with software. The recruiter defines the role and the signals that matter, and the AI turns those into structured questions. The candidate answers by voice or text from a link, the AI asks follow-ups when needed, and each response is scored against a fixed rubric. The recruiter receives a ranked summary with scores and supporting quotes. Criba runs this as a ~5-minute voice-first interview and returns a Pass / Borderline / Reject result.
An AI interview can be fairer than ad-hoc phone screens because every candidate answers the same questions against the same rubric, removing the drift and fatigue that affect human interviewers. But fairness depends on design: the questions must be job-relevant and the scoring auditable. Criba scores against role-specific signals and shows the direct quotes behind each rating, so decisions are transparent and reviewable rather than a black box, and recruiters always make the final call.
Treat it like any structured interview. Find a quiet space with a stable connection, and for a voice interview, speak clearly at a normal pace. Answer the actual question with concrete examples — what you did, how, and the result — rather than rehearsed generalities. There are no trick questions; the AI is measuring role-relevant signals, so specific, honest answers score best. With Criba you can take the ~5-minute interview whenever suits you, so pick a time when you can focus.
An asynchronous video interview records a candidate answering preset questions for a human to review later — it saves scheduling time but still needs a person to watch and judge every recording. An AI interview goes further: the AI conducts the conversation, can ask adaptive follow-ups, and scores the answers itself against a rubric. The recruiter reviews a ranked summary instead of watching full recordings. Criba uses a voice-first AI interview rather than one-way video.
Pricing varies by vendor and volume. Enterprise video-interview suites can run into the tens of thousands per year, while usage-based tools are far cheaper for teams that screen in bursts. Criba is free to start and priced transparently by usage, with no mandatory sales call, so you can run real AI interviews and see the value before committing. See the pricing page for current details.
No. An AI interview replaces the repetitive first screen — the same questions asked of every applicant — not the recruiter. People keep the high-value work: assessing finalists, partnering with hiring managers, and making the hire decision. The AI simply hands them a shorter, higher-quality shortlist with the evidence to act on it, so they spend their time on judgment instead of triage.