Interviews candidates take on their own time

Candidates answer a short set of job-related questions in writing, when it suits them. Voice and video answers are planned, and the table below says when. A recruiter reviews the transcript and decides.

How candidates answer

Mode	Availability	How it works for the candidate
Text	Available now, the default	One question at a time. Answers save as the candidate goes, so they can stop and come back.
Voice	From release 1.0 (target: Q4 2027), optional	Recorded in the browser on a phone or computer, after a device check and a practice question that is not stored.
Video	From release 1.0 (target: Q4 2027), optional and never required	Available when an employer wants it. Candidates can always answer by text or voice instead, and video is never analyzed for faces or expressions.

Until then, candidates answer in writing. The demo shows voice and video answers with prepared transcripts. Recorded answers will allow one re-record per question. Candidates read their own transcript and can correct it before they submit.

Before anything is recorded, the candidate sees what will happen to their answers and how to ask for something different.

What is recorded, who can see it, and that recordings are kept in Canada.
That transcripts are evaluated with AI services in the United States, after names and contact details are removed.
That there is no analysis of faces, tone of voice, emotions or personality.
That recordings are deleted 90 days after the interview, unless a legal hold applies.

Candidates can ask for extra time, answer by text only, or have an interview with a person instead. Requests go to the recruiter and are tracked like any other task.

Evaluated from the transcript

The evaluation reads the words, not the person. Each score points at the sentence it relies on.

A score from 0 to 5 for each criterion in the approved rubric, with a one-sentence reason.
Every quote is checked to appear word for word in the transcript.
Limitations are stated: a short answer or a missing topic is reported, not guessed at.
The recruiter decides with their own reason. When the interview completes, the 45-day outcome clock starts.

Close-up of the AI evaluation of C-00015’s interview: evaluated from the transcript only, with no facial, voice or emotion analysis; model confidence 89%; personal care experience scored 5 out of 5, with two quotes from the written answers, each labelled with its question. — Each criterion score quotes the answer it rests on. Screenshot of the demo, with fictional people and companies.

Scheduling the next conversation

The recruiter proposes three times; the candidate picks one on their phone and gets a calendar file in their language.

Integrations behind interviews

Integration	State
Interview transcription	Simulated Planned for release 1.0, target Q4 2027
Google Calendar	Planned Waiting on: Google’s verification of the app’s calendar permission, not requested yet. It needs the company domain, verified and carrying the privacy policy, which is not in place yet. Planned for release 1.0, target Q4 2027
Microsoft Outlook calendar	Planned Waiting on: Microsoft publisher verification for the app, not requested yet and tied to the company domain, then each customer administrator’s consent. Planned for release 1.0, target Q4 2027

States as of October 9, 2026. We check them again by November 30, 2026.

Answer an interview as the candidate

Open the candidate’s view in the demo, in French, and go through the asynchronous interview on a phone. Another interview is already waiting for review, with its transcript and quoted scores.

Try the demo Contact us