Agentic Interview
Quick definition
An agentic interview is an AI-conducted hiring conversation where the system decides what to ask next based on the candidate's prior answers. Unlike a static script, it has agency over the dialog flow.
How it works
In practice there are three layers. A speech-to-text (STT) component transcribes the candidate's voice. An LLM combines that transcript with role context, prior turns, and the scoring rubric to generate the next question. A text-to-speech (TTS) layer delivers it back as audio. This loop runs every turn — so the interview is a conditional decision tree, not a static script. The contrast with one-way video (collect answers against a fixed list) and with mock-interview chatbots (typically candidate-side practice) is that an agentic interview is a real-time decision maker.
Why it matters
Signal quality goes up: ambiguous answers get probed, irrelevant areas don't burn turns. Average interview length tends to drop because the system knows in real time when the rubric is covered. On the candidate side, the experience of being heard is meaningfully better than a one-way video, which lifts completion rates.
Related terms
- Voice AI Interviewer — software that runs the agentic loop with the voice layer attached.
- Structured Interview — an agentic interview is a modern form of structured interview.
- Behavioral Rating Scales (BARS) — the output of an agentic interview is typically scored against BARS.
- Voice AI Interviewer (product)
- How AI interviews work
- EU AI Act compliance