Will candidates actually take an AI interview?
The thing that usually loses a candidate between shortlist and screen is scheduling, and an AI interview removes it — they get a link and join at a time that suits them within 48 hours, on a phone or a laptop. That is the real argument for acceptance. We do not publish a completion-rate benchmark, because we have no permissioned customer data to publish one from; for your own jobs, the dashboard shows how many candidates were invited and how many completed the interview.
What is removed from the candidate's side
No scheduling coordination, no calendar back-and-forth, no time-zone negotiation. The candidate joins from a link, at a time they pick within 48 hours of the invite, in a browser on a phone or a laptop.
A pre-join check confirms microphone, camera and speaker before the session starts, so device problems surface before the interview rather than during it.
The session is planned for 15 minutes by default, 20 at most, so the ask is bounded and legible up front rather than open-ended.
What is asked of them, honestly
They talk to software instead of a person, and some candidates will decline on that basis alone. That is a real cost and it is not distributed evenly — senior candidates with options are the most likely to decline, and they are often the ones you most want.
They need a quiet space, a working microphone and a camera that stays on for the length of the session, which is a genuine barrier for some people and some places.
They are told how their data is used before the interview begins. The interview is transcribed and the transcript is kept; on the Growth and Enterprise plans it can also be recorded.
Where the drop-off actually sits
The measurable risk is between invitation and join, not mid-interview: a candidate who joins and passes the device check has already accepted the format. That means the lever is the invitation — how the role is worth their 15 minutes, and whether it is clear a human reads the result.
For each of your own jobs the dashboard shows how many candidates were invited and how many completed the interview — the only numbers about your candidates that are actually about your candidates.
Where this stops
We quote no acceptance or completion benchmark, and you should be sceptical of vendors who do — the number depends entirely on role seniority, market and how the invitation is written, so an average across someone else's customers tells you very little about yours. Haira publishes no customer counts or outcome statistics at all, by design, until there is permissioned data to publish.
Will candidates actually take an AI interview?
The friction that usually loses candidates is scheduling, and there isn't any here — they get a link and join at a time that suits them within 48 hours, on a phone or a laptop, and a pre-join check tests their microphone, camera, speaker and connection before the session starts. The session is planned for 15 minutes by default, 20 at most, and candidates get to explain context a resume can't carry. Your dashboard shows, for each job, how many candidates were invited and how many completed the interview.
Last checked against Haira's own implementation on 2026-09-30.