How the AI Interview Works

Last updated: October 30, 2026

This notice explains how artificial intelligence ("AI") is used when you take an interview or assessment on Cohera on behalf of an employer. It supplements our Privacy Policy.

1. What the AI does

During an AI-led interview, an AI voice asks you questions about your background and the role. During a coding, ML, system-design or take-home assessment, an AI interviewer may ask about your approach while you work.

In an AI-Assisted assessment you work in a cloud workspace with an AI coding tool, Claude Code or Codex, the way engineers work day to day. Afterwards an AI interviewer asks you about your work; in a design assessment it talks with you throughout. Using the AI tool is expected and is not held against you; the evaluation looks at how well you checked, directed and took ownership of the result.

An AI-Assisted build assessment may begin with an onboarding call: a voice call of up to 15 minutes with the task's point of contact, a colleague played by AI or an AI interviewer, who explains the task and answers your questions. During the call the project's files are read-only and the AI coding tool is off. If your assessment begins with one, the same point of contact asks you about your work at the end.

Some AI-Assisted tasks are set in a simulated team. The colleagues you can message are played by AI, and during the round Cohera may make scripted changes on a colleague's behalf, such as a commit to the code you're working on, a message or an incident report. You're told when each one happens. They're scripted the same way for every candidate on the task, and how you respond is part of the assessment. If you're stuck, tell a colleague what you've tried and ask for help: they may give you a hint, a nudge in the right direction, never the solution. A round allows only a few hints, and the hiring team sees each one and takes it into account.

A WorkVerse mission is a longer AI-Assisted assessment set at a simulated company. It opens with the onboarding call, then runs in timed parts: you build a change, respond to a review of your pull request, handle a scripted incident, review a teammate's pull request and write a handoff note, before the debrief. The colleagues are played by AI, and every change or message Cohera makes on their behalf is scripted the same way for every candidate. Your review comments and replies, your incident summary and your handoff note are analysed with the rest of your work.

After the session, an AI model evaluates what you said and produced against the rubric the employer configured for the role — typically the role's required skills, problem-solving, communication, and the employer's stated must-haves. The output is a structured evaluation: per-competency scores, a written summary, strengths, concerns, and suggested follow-up questions for the hiring team.

2. What is analysed

Depending on the round, the AI analyses:

• the transcript of your spoken answers (speech is transcribed to text by a speech-to-text provider);
• the code, notes, whiteboard content and design documents you produce in an assessment workspace;
• files or a public repository you submit for a take-home assignment;
• in an AI-Assisted assessment, how you used the AI coding tool: your prompts and a summary of its responses and of what its commands returned, the commands you ran, the files you opened, the changes you made (including which of them the AI made), and the results of automated checks run on your work, including checks you didn't see while you worked. If the task has colleagues you can message, played by AI, your questions and their answers are included too; if it has scripted changes, so are when each one's notice was shown to you and what you did after it. If the assessment began with an onboarding call, the call is recorded and transcribed, and its transcript is analysed like the debrief's, including what you asked and what you were told in it.

In an AI-Assisted assessment your entire screen, camera and microphone are recorded, and we keep snapshots of your code as you work. Please don't paste personal information, passwords or other secrets into the AI tool.

3. What is not analysed

• No facial analysis. Your camera feed is recorded so the hiring team can watch the interview; it is not analysed by AI for facial features, expressions, attention, or identity.
• No emotion, sentiment or personality inference. The AI does not score how you look or sound, only the content of your answers and work.
• No voice biometrics. Your voice is transcribed to text; it is not used to identify you or to infer characteristics about you.
• No use for training. Your interview data is not used to train Cohera's or our providers' AI models.

4. Human review

The AI evaluation supports, and does not replace, the employer's decision. The hiring team reviews the recording, transcript and evaluation and makes the hiring decision. By default, a below-the-bar result does not reject you automatically: your candidacy is held for the employer's hiring team, who review the interview before confirming any rejection. An employer may turn that default off for its workspace; where it does, a rejection email may be sent based on the AI evaluation. In either case the employer remains responsible for the decision and you may request a human review as described below.

If a technical problem on our side interrupts your assessment, you are never rejected because of it: the round is flagged for us to fix and the hiring team decides what happens next. If something went wrong on your side, tell us through the help link on your interview page.

When Cohera recruits for the employer. Some interviews are Cohera's own screening step for a role Cohera is recruiting for. In those, Cohera's recruiting team reviews the recording, transcript and evaluation and decides, against the employer's requirements, whether to put you forward; the employer runs its own interviews afterwards. The employer receives a written summary and the recording of the screen, through Cohera's recruiting partner where there is one, only if you approve being put forward for that role. Accommodation and review requests (sections 5 and 6) go to Cohera, which decides on them.

5. Accommodations

If you need an accommodation — extra time, a format without AI coding tools, a text-based or camera-off format, a different day, or anything else — ask before you start, by emailing support@coheraai.com or through the help link on your scheduling or interview page. We forward your request to the employer, who decides on it, and we help arrange what it agrees to. See our Accessibility Statement.

6. Requesting a review, correction or deletion

You may, within 30 days of your interview or assessment (or later where the law gives you a longer period):

• ask for a human review of an AI-assisted evaluation;
• ask for a correction of inaccurate personal information used in the evaluation;
• ask for a copy of the personal information used;
• ask for deletion of your recording and evaluation, subject to the employer's legal retention obligations.

Email privacy@coheraai.com from the address you interviewed with, naming the employer and the role. We forward the request to the employer, who controls the hiring decision and is responsible for responding, and we help the employer respond within the time applicable law requires; where the law requires Cohera itself to respond, we will. Requests may also be made through the "Need help" link on any interview page.

7. Providers involved

Your interview is processed by the service providers that operate the platform — a realtime speech and language-model provider for the conversation and transcription, a speech-synthesis provider that turns the interviewer's written replies into its voice (it receives that text only, never your audio), a language-model provider for evaluation, and secure cloud storage for recordings — under contracts that restrict their use of your information. In an AI-Assisted assessment, your prompts and the code your AI coding tool reads are also sent to the model provider behind that tool (Anthropic for Claude Code, OpenAI for Codex), and your workspace runs on Vercel. The current list is on our Subprocessors page.

These providers process your information mainly in the United States and Canada. Information held in the United States may be accessible to courts, law enforcement and national security authorities there under US law.

8. Retention

See also the data retention section of our Privacy Policy.

9. Questions

Contact privacy@coheraai.com or write to Cohera AI, 9088 Halston Court, Burnaby, BC V3N 0A7, Canada.