Recagn
Sign in

Guide · AI interviews

One-way interview: what it is, how it works, and what the AI does

Set the questions once; each candidate answers alone, and you watch the answers later.

Part of AI interviews: how they work, what they cost, and when to use them

What is a one-way interview?

A one-way interview is a set of preset questions that a candidate answers alone, on camera or by voice or text, at a time they choose before a deadline. Nobody is on the other side while they answer; a reviewer watches the recordings later. Vendors also call it an asynchronous, async or on-demand video interview.

It is the most common format for an AI interview, where software scores the recordings. What that scoring reads is covered further down.

The names differ by vendor. Spark Hire calls it a one-way video interview, HireVue an on-demand interview, and Interviewer.AI an async video interview, short for asynchronous: the candidate and the reviewer are never in the interview at the same time (each vendor's own page, checked 17 Nov 2026).

What the candidate goes through

The steps below follow Spark Hire's help centre and HireVue's candidate help, both checked 17 Nov 2026. Other tools run a similar flow.

  1. An email arrives with a link for that candidate and a deadline to finish by.
  2. The candidate tests their camera and microphone. Both vendors offer a practice question, and Spark Hire says the practice recording isn't shared with the hiring team.
  3. The questions come one at a time, each with its own rules: think time before the recording starts, a time limit for the answer and a number of takes. The employer sets how many retakes and how much think time to allow.
  4. The candidate records, and can review the answer and record again if the question allows another take.
  5. They submit. On Spark Hire a candidate can pause and come back to the same link before the deadline; not every tool allows it.
  6. The reviewer watches the recordings when it suits them. On most AI tools a transcript and a score come with each one.

Video, voice or text

"One-way video interview" is the common name, but the same pattern runs by voice or typed text, and one question can allow more than one format. What the reviewer gets from each format, and what each asks of the candidate, is set out in text, voice and video compared.

Asynchronous video interview vs live video interview: which suits a first round?

An asynchronous video interview suits a first round where many candidates answer the same questions and you need a recording to compare. A live video interview suits later rounds, where you need to probe and the candidate needs to ask questions back. Use async to shortlist and live to decide.

"AI video interview" now covers a third kind as well: a live AI interviewer that speaks the questions and follows up on the answers.

One-way (asynchronous)Live video call with a personLive AI interviewer
Who is on the other side while the candidate answersNobodyA recruiter or the clientSoftware that speaks, often as an avatar or a voice
When the candidate answersAny time before the deadlineAt a booked slotVaries by tool
Follow-up questionsNone, or a set follow-up on some toolsAs the conversation goesGenerated from the candidate's answer
What the reviewer getsRecordings, and on AI tools a transcript and scoreNotes from whoever ran the callA recording, transcript and score on tools that offer them
What it costs the recruiter's dayWatching time onlyEvery call, at its slotWatching time only

Source: how each format works, from Hyring's AI video interviewer page, Interviewer.AI's pricing page and Spark Hire's help centre, read 17 Nov 2026 · the last row is this guide's reading

One-way wins on comparison and on time. Every candidate gets the same questions in the same order, so their answers sit side by side. Nobody has to find a slot between a candidate's notice period and their working hours. The reviewer watches when it suits them and skips to the answer they need.

Live wins on conversation. A candidate can ask about the role, and the interviewer can chase a vague answer until it gets specific. A one-way interview gives neither, unless the tool adds a follow-up question.

Hyring, which sells both kinds, draws the line the same way: one-way for high-volume screening at the top of the funnel, live for roles that need probing (Hyring, checked 17 Nov 2026).

What is an AI video interview, and what does the AI do?

An AI video interview is a video interview in which software scores the answers, asks the questions, or both. In a one-way AI interview the software transcribes each recording and scores what was said against the role's criteria, on the tools we checked. In a conversational one, an AI interviewer also asks follow-ups as the candidate talks.

Hyring lists transcripts among its features and Interviewer.AI lists transcripts and relevancy scoring, and both offer an AI interviewer that adapts its next question to the last answer (each vendor's own page, checked 17 Nov 2026). Before you compare tools, find out which of the two jobs the AI does in each.

Words, voice or face

A scorer can read three things: the transcript of what was said, the delivery (pace, pauses, filler words), or the picture. Which one a tool reads matters more than the word "AI" on its page.

HireVue used to analyse candidates' faces. SHRM reported in February 2021 that HireVue had dropped facial analysis in March 2020, after its own research found that language analysis predicted better and that "visual analysis no longer significantly added value". A face isn't the job. A transcript of the answer is at least about the job.

Questions to ask any vendor

  1. What does your AI score: the transcript, the voice or the video picture?
  2. Who writes the questions, and can I edit every one?
  3. Does an AI ask follow-ups, and on which formats?
  4. What does the candidate agree to before the first question?
  5. Who makes the reject decision: the software or a person?

None of the vendors we checked publishes an accuracy figure with a method behind it, so this guide quotes none.

When is a one-way interview the wrong format?

A one-way interview is the wrong first step when the role needs a live back-and-forth, when you have only a handful of candidates, or when the candidates can't easily record on camera. Pick the format per role, and per question where your tool allows it.

  • Senior or consultative roles. The candidate expects to interview you as well, and a recording gives them nobody to ask.
  • A handful of candidates. With four or five people, the calls take less time than writing and setting up the questions.
  • Candidates without a camera, a quiet room or spare data. Offer voice or text instead of video, so the format doesn't decide who finishes.

Where one of these applies, start with a call and keep the one-way interview for the next role that has a long list.

How Recagn does this

Recagn is our product. Recagn runs a one-way interview: each candidate opens a link and answers the role's questions one at a time, before the deadline you set, choosing from the formats each question allows. Recagn transcribes each voice or video recording and scores the transcript against the role's criteria. You decide who moves on.

The clip shows a person performing a scripted answer on video in Recagn's interview screen, recorded on 10 Nov 2026 for the Digital marketer (Bhubaneswar) worked example: question 4 of 7.

The clip sequence

  1. 0–4s. Question 4 of 7: How do you set up conversion tracking across GA4, Tag Manager and Meta Pixel, and how do you turn weekly/monthly data into decisions? Give an example of an insight that changed what you did.
  2. 4–36s. Ujjawal chooses Video recording and answers on camera: I set GA4 events for form submits and call clicks, fire them through Tag Manager, and mirror them with the Meta Pixel. I found form conversions were double-counted because the thank-you page fired the tag twice. After I fix it, one Google campaign that looked cheap was actually average, so I moved budget to the better one. The bug show up when GA4 had more conversion than the CRM had leads. Thank you.
  3. 36–41s. He stops, then Save & next. His answer is saved and question 5 of 7 opens.

Screen recording · answer performed from a script · invented candidate

Read it against the steps above. The question opens on its own screen, "Question 4 of 7", with the four formats it allows as buttons: Type answer, Voice recording, Video recording and Upload document. The candidate picks video and records, and the countdown under the preview shows the time left: 3 minutes per recording is Recagn's default. Stop, then Save & next, and question 5 opens. Nobody is on the other side at any point, which is what makes it one-way. The take wasn't scored; the scored run used a text-to-speech reading of the same words.

Each question allows text, voice, video and document unless you narrow it, and code answers (not run) are a fifth format. The candidate picks one per question. The invite carries a deadline of 3 to 30 days, 14 unless you change it. Before the first question, a device check tests only what the interview uses, with a practice recording for voice and video.

Recagn scores the transcript, not the voice or the picture, and its scorer is told not to use or infer protected attributes, photo included. Frames sampled from video answers go to a separate check for reading off-screen or reading prepared text, which looks at behaviour, never identity or appearance. Its result isn't part of the score. In the interview review you watch each recording with its transcript under it.

On questions where you turn follow-ups on, Recagn can ask one follow-up on a typed answer. A spoken AI interviewer is on the roadmap.

You set Takes (re-records) and think time on each voice or video question, and a time limit on any question; how many retakes to allow covers the choice. An interview with any video answer uses three credits instead of one.

Running the first round and transcribing every answer come off your plate. The questions, and who moves on, stay yours. Recagn does the screening. You make the calls.

The full first round in Recagn, from invite to shortlist, is on how an automated first-round interview works.

Questions people ask

Can't a candidate just read a prepared answer in a one-way interview?

Yes, they can. A recording can't stop it, and nobody is there to change the question. Ask questions about the candidate's own work that can't be rehearsed, cap how often they can record again, and ask whether the tool checks for reading off-screen. Then ask about one of their answers in the live round.

What is a HireVue on-demand interview?

It is HireVue's name for a one-way interview. The candidate opens the link in the invitation, tests their camera and microphone, can try a practice question that isn't saved, then answers each question after a prep-time countdown, within a response-time countdown. The hiring company sets the number of attempts, and a HireVue live interview comes as a separate invite, going by HireVue's candidate help (updated 28 Aug 2024, checked 17 Nov 2026).

Does a one-way interview on Recagn cost anything if the candidate never starts it?

No. Recagn charges credits only when an interview is scored, so a candidate who never starts costs nothing. A scored interview uses one credit, or three if any answer is video, plus webcam proctoring where you turn it on.

See it run on one of your roles.