How AI-led interviews work
An AI-led interview is a set of behavioral questions that Amazon Connect Talent generates for each competency a job requires. Amazon Connect Talent runs the interview with the candidate, asks follow-up questions when useful, and captures the conversation for you to review.
Topics
Competency-based questions
Amazon Connect Talent generates behavioral questions from the job's competencies. Each question is grouped under the competency it measures and carries a short rationale that explains why it assesses that competency. Questions are written to be clear, fair, and focused on a single competency, and Amazon Connect Talent can generate them in multiple languages. For the languages that Amazon Connect Talent supports for an evaluation, see Set the evaluation language.
In the results, each question card shows a competency badge so you can see which competency the question measures.
Dynamic follow-up questions
During the interview, Amazon Connect Talent can ask a follow-up question to explore a candidate's answer in more depth. In the results, follow-up questions appear as nested items beneath the primary question.
Edit interview questions
Before you publish an evaluation, you can review the questions, remove any you do not want, and add new AI-generated questions. To add a question, choose a competency and, optionally, enter a short instruction to guide the AI. The competency dropdown lists up to 15 competencies that Amazon Connect Talent identifies for the job, including competencies that are not among the five shown during evaluation creation. You can restore a removed question while you work.
Each interview module starts with a default set of questions, and Amazon Connect Talent estimates the interview duration from the number of questions.
Interview playback
After a candidate completes the interview, you can review a two-column playback: the list of questions alongside a synchronized transcript, with optional audio. A timeline bar shows how much of the interview the candidate spoke compared with the interviewer.