Agentic interview

These three phrases describe the same hiring product. An AI interviewer is the software that conducts the interview. An AI interview agent is that interviewer when it chooses the next question from the last answer and from the skills still open on the rubric. Agentic interviewing is the name for running the round that way.

A fixed list asks every candidate the same questions in the same order. Agentic interviewing uses the rubric to decide which skill is still untested, and the last answer to decide how deep the current skill goes. A thin answer draws a follow-up. A strong answer moves on.

For a backend role whose rubric includes API design, the AI interviewer asks how the candidate would paginate a list endpoint. An answer that names cursor pagination and then stops gets a follow-up on what happens when a row is inserted during the scan. An answer that only says “limit and offset” gets a follow-up on what breaks at a large offset. Later questions come from whichever rubric skills are still open, such as data modeling or how the candidate would explain the choice to another engineer.

You paste the job description. Cointervue turns it into a rubric for that role: skills, seniority, and whether DSA or system design is in scope. Candidates join from a link. No app install.

• Invite candidates with a link, a bulk upload, or your ATS.
• The candidate runs a device check, then talks with the AI interviewer. They can pause, ask for a question to be clarified, or ask to hear it again.
• Live coding sits in the same session, in 20+ languages, when the rubric includes it. DSA problems are calibrated to the level of the role.
• System design, behavioral questions, and culture fit stay in that same session when the rubric calls for them.
• If the connection drops, the session keeps the progress already made.
• The scorecard is ready within 60 seconds of the interview ending. Your team reviews it. The product does not reject a candidate on its own.

A first technical round is usually a fixed prompt list or an agentic interview. A human interviewer still fits the final round. Technical interview as a service compares that model on availability, turnaround, and price.

The scorecard is the reason to run the round this way. A pass or fail line is not enough for a hiring manager who was not in the room. Cointervue ties each score to the rubric and to evidence from the session.

• Rubric scores for the skills you set on the role.
• The code the candidate wrote, with notes on the scorecard.
• Transcript highlights for the follow-ups, so you can see why a skill was marked the way it was.
• A hire or no-hire recommendation. Your team can disagree with it.
• Integrity flags when proctoring is on: tab switches, pasted code, and signs of outside help. Candidates are told about monitoring before the interview starts. A person reviews the flags.

It fits the round with the most candidates. What an L1 interview means is that first technical screen: fundamentals, coding, and a check that the resume holds up, usually in 45–60 minutes. Many teams also move an early L2 once they trust the scorecards. Final rounds, and the offer decision, stay with your own people.

Tech companies use it to take those screens off senior engineers. A 45-minute screen also costs prep, write-up, and context switching. The cost of technical interviews guide prices that time. A coding test still has a place when you only need a timed exercise with no follow-up. The AI interview vs coding test page separates the two.

Staffing firms use it when a mandate brings in more candidates than their interviewer panel can see in a week. One Cointervue pilot delivery lead at a Gurgaon IT staffing firm sent a shortlist of 6 in 4 business days instead of the usual 9–11. White labeling is an add-on, so candidates can see the firm’s brand.

Handing that round to an outside provider is technical interview as a service. Cointervue is ₹199 per completed interview pay as you go, ₹169 with 100 or more a month, and from $5 in the US and EU. The first 5 interviews are free.

An AI interview agent is an autonomous software system that conducts conversational technical screens, assesses real-time code execution, and selects follow-up questions dynamically based on what the candidate just answered and what rubric skills remain untested.

Unlike rigid question scripts or static test forms, an AI interview agent engages in active technical dialogue. It probes architectural trade-offs, asks why an engineer chose a specific data structure, and evaluates how they handle edge cases. When paired with an AI coding interview, the agent evaluates practical execution alongside verbal reasoning in a single session.

• Adaptive conversational questioning tailored to candidate responses.
• Live code execution in a browser sandbox across 20+ languages.
• Objective scoring mapped to role-specific JD rubrics.
• Decision-ready scorecard delivered within 60 seconds of completion.