Mercor AI Interview (2026): How It Works, What's Asked and How to Pass

The Mercor AI interview is a recorded conversation with an AI interviewer that asks role-specific questions, produces a transcript and scores your answers for the job listing you applied to. According to Mercor's help docs, most interviews take about 20 minutes, run in English unless stated otherwise, and need a working camera and microphone in a desktop browser. You only need to complete each interview once, and the result carries over to other applications that need it. Mercor does not publish a scoring rubric, so the best preparation is to be specific about your own experience and to answer concisely.

As of October 2026, here is the short version.

At a glance

Question	Short answer
What is it?	An AI-led, recorded interview that evaluates you for a specific Mercor listing.
How long is it?	About 20 minutes for most interviews; some are shorter or longer (Mercor docs).
Who interviews you?	An AI interviewer. Mercor says AI is the only screening method; there is no chat or call alternative.
What do you need?	Chrome (recommended), Edge or Safari on a computer, a working mic and camera, and sometimes full-screen sharing.
What is asked?	Role-specific questions, often built around your resume and past work, with follow-ups.
Can you retake it?	Yes, up to three times across all applications for that interview, but it resets your application progress.
Do you get feedback?	No question-level feedback. You are notified when your application moves to the next stage.
Is it legit?	Mercor is a real, well-funded company. Candidate complaints are about slow or no responses and data use, not fake listings. Details below.

How the Mercor AI interview works

Mercor started as an AI recruiting product and now connects companies, including AI labs, with domain experts and contractors. That shift and its funding are covered in TechCrunch's October 2025 report, which reported a $350 million Series C at a $10 billion valuation. The AI interview is the screening step in that pipeline.

The flow, based on Mercor's documentation:

1. You apply to a listing on Mercor.
2. If the listing needs an interview, it appears in your dashboard. You open a waiting room with setup checks.
3. The AI interviewer asks questions out loud. Captions are on by default (you can turn them off but not enlarge them).
4. Mercor stores a recording and an AI-generated transcript to evaluate you and to recommend your profile to employers.
5. You are notified when your application moves forward. Mercor gives no fixed timeline.

Mercor says it does not sell your interview data and does not use it to train AI models. It also says it does not let you download or share recordings and transcripts, so you cannot review your own attempt afterwards.

The setup checks

In the waiting room, a working microphone is required. If the interview needs it, so is screen sharing, and only full-screen sharing is supported. Face in frame, good lighting, a sharp camera image and a steady camera are recommended but do not block you. If the Start button stays grey, read the message under it, allow camera and microphone permissions in the browser, and select Restart devices. Firefox and mobile browsers are not supported.

What is asked in a Mercor AI interview

Mercor's own docs say some interviews are technical and go deep into past experience, while others test language or reasoning. They do not publish a question bank, and they ask candidates not to share specific questions. What follows is a pattern drawn from Mercor's description plus candidate and third-party accounts, so treat it as a guide to the shape of the interview, not a list of real questions.

Resume-based questions

Several candidate accounts describe the AI building its questions from your resume: it opens with a recent role or project, then drills into specifics. One engineering-focused guide describes follow-ups aimed at claims such as what the bottleneck was, which number moved and what you would change now. That guide is a third-party write-up with no cited source, so weigh it accordingly. The consistent theme across accounts is simple: anything on your resume can be probed.

Example question types (labelled as examples, not real Mercor questions):

• "Walk me through the most recent project on your resume. What was your part in it?"
• "You wrote that you improved a process by 30 percent. How did you measure that?"
• "What would you do differently if you ran that project again?"

Domain, coding and generalist variants

Autocomplete shows people searching for Mercor interviews by variant: domain expert, coding or software engineer, generalist, finance, and AI trainer or evaluator. Mercor's official pages do not describe these variants separately, so what we can say is limited to what candidates and guides report.

• Domain expert (finance, law, medicine, science and similar): expect questions about your field, your past work and how you would judge the quality of an answer. A guide to AI evaluation roles describes questions about rubrics, factual accuracy and edge cases. That is a third-party guide, not Mercor's.
• Coding or software engineering: reports describe resume-led technical questions about your projects. The same engineering guide describes a separate live-coding phone screen later in the process, with a human engineer. Another account lists a different loop, so the number of rounds clearly varies by role and over time.
• Generalist: listings for generalist and AI trainer work describe tasks like writing realistic prompts and evaluating model outputs, so expect questions on judgment, clarity and how you check your own work. Some interviews include a whiteboard that autosaves and does not need submission.
• Language or reasoning interviews: Mercor says some interviews test language or reasoning rather than past experience.

Behavioural and scenario questions

Third-party guides describe scenario questions such as handling conflicting instructions, ambiguous guidelines, or explaining a problem with a model's output to a client. These are examples from guides, not confirmed Mercor questions. The structure to use is the same as in any behavioural round: situation, your action, a measurable result. Our guide to behavioral interviews for engineers walks through it.

How the interview is scored

Mercor says the interview is used to evaluate your performance for the listing and to recommend your profile to employers. It does not list scoring criteria, weights or pass marks. Anyone quoting a Mercor pass rate or a cut-off score is guessing.

What you can reasonably infer from the format: an AI scoring a transcript can only reward what you actually say. That favours answers that are specific (named tools, numbers, decisions), direct (answer first, detail second) and consistent with your resume.

Retakes

Per Mercor's retake instructions:

• You can retake an official interview up to three times across all applications linked to that interview.
• Retaking resets your application progress.
• Hiring managers can see your retake count.
• The waiting room asks you to keep retakes for technical problems.
• If an attempt ends much earlier than planned, it shows "Interview too short", counts as a retake and has to be redone.
• The Retake assessment button appears only after the interview is completed. If it does not show, refresh, then email support@mercor.com.

Mercor's pages are slightly inconsistent about intent: one frames retakes as an option, another asks you to reserve them for technical issues. Because hiring managers can see your count, treat the first attempt as the one that matters. Third-party blogs mention a 30-day waiting period after a failed attempt, but we could not find that in Mercor's own documentation, so treat it as unverified.

Mercor's docs also describe a free, unlimited practice interview that does not affect applications. An earlier snapshot of the docs said practice interviews had been removed, so check your dashboard to see whether it is currently offered.

Is the Mercor AI interview legit, hard, or a scam?

Legit. Mercor is a real company with major press coverage and funding (see the TechCrunch link above). Nothing we found suggests the interview itself is a scam, and Mercor says it does not sell interview data.

Where the complaints are. Candidates on forums and review sites report slow or no responses after the interview, offers that were later paused, and no feedback. Some also suspect interviews are a way to collect expert knowledge. These are candidate reports and speculation, not established facts, so judge for yourself what you are comfortable recording.

Is it hard? Mercor's interview is not designed to be a puzzle. The difficulty is mostly that it is unfamiliar: you talk to a bot, you cannot see reactions, and you have to be concise. If you can explain your own work clearly, it is manageable. If your resume overstates anything, the follow-ups will expose it.

Sensible precautions: do not share confidential client or employer information, do not upload identity documents until you are dealing with a human about a real offer, and use the official work.mercor.com site rather than links from unsolicited messages.

Rules to follow

Mercor's docs are explicit:

• Use AI tools only for grammar and wording checks. Do not use large language models to write answers, rate outputs, produce explanations or evaluate code.
• Do not share specific interview questions, screenshots or confidential project or client details.
• Keep answers concise and avoid long silences, since extended silence may prompt the interviewer to interject. If it cuts in while you are thinking, say you were not finished.

Follow these rules. The format records your face and voice, and hiring managers can see your retake count. There is no safe shortcut, and the interview only helps you if the result reflects you.

How to prepare: a 3-day and 7-day plan

7-day plan

• Day 1: Re-read your resume line by line. For each bullet, write the problem, your action, the result with a number, and one thing you would change.
• Day 2: Pick your three strongest projects or roles. Practise explaining each in 90 seconds, out loud.
• Day 3: Review the basics of your field. For technical roles, be ready to explain trade-offs, not just definitions.
• Day 4: Practise behavioural scenarios: conflict, ambiguity, a mistake you owned. Use situation, action, result.
• Day 5: Do a spoken mock with follow-ups. Our AI mock interview asks follow-up questions and ends with a written report, which is the closest thing to the feeling of being questioned by an AI.
• Day 6: Test your setup: Chrome, headphones, external microphone, steady camera, a quiet room, stable network, VPN and extensions off.
• Day 7: Light review only. Sleep, then take the interview at a time you choose.

3-day plan

Do Day 1 (resume bullets), Day 2 (three project stories out loud) and Day 5 (one spoken mock) from above, and test your setup the evening before.

Tips that match Mercor's own guidance

• Have pen and paper ready, and a calculator if your field needs one.
• Answer first, then add one concrete detail.
• Pause to think briefly, but do not sit in silence.
• Be honest about what you did yourself versus what the team did.
• If the bot stops responding, wait 30 to 60 seconds, refresh, rejoin, and if needed clear cache and disable VPNs and extensions.

Common mistakes

1. Vague claims. "Improved performance" gives the interviewer nothing to score. Name the metric.
2. Rambling. Long answers risk being cut off or diluted.
3. Overstating your resume. Follow-ups go to the details.
4. Using an AI tool to compose answers. It is against the rules and it shows in the cadence of the answer.
5. Burning retakes. Hiring managers see the count.
6. Technical setup errors. Unsupported browser, blocked permissions or an unstable network are the most common reasons an interview fails.

Practising for an AI interview

Practising with another person is useful, but it does not reproduce the experience of being followed up on by an AI. A spoken mock where the interviewer reacts to what you actually said is more realistic. Thita's free AI mock interview is a spoken, adaptive interviewer that asks follow-up questions, runs your code against test cases for coding rounds and ends with a written report. The Starter plan is free. It is built for technical interviews, not for Mercor specifically, and it does not replicate Mercor's format.

If you want a quick, no-signup warm-up, the in-article practice round is Thita's own format (three questions in about ten minutes, typed answers, scored out of ten). Think of it as Mercor-style practice at best, not a copy of it.

For a broader picture of AI interviews across vendors, read how to pass an AI interview, the AI interview questions and answers guide, and our list of companies using AI interviewers in 2026. If you are also applying through micro1, see the micro1 AI interview guide. To rehearse resume deep-dives specifically, try a resume-based mock interview.

FAQ

How does the Mercor AI interview work?

You open the interview from your dashboard, pass a setup check, and talk to an AI interviewer that asks role-specific questions. Mercor records the session, produces a transcript and evaluates your answers for the listing. You are notified when your application advances.

How long is the Mercor AI interview?

Mercor says most interviews take about 20 minutes, with some shorter or longer. Ending much earlier than planned can show as "Interview too short" and count as a retake.

Is the Mercor AI interview legit?

Yes, Mercor is a real company that has been covered by outlets like TechCrunch. The common complaints from candidates are slow or missing responses, no feedback and concerns about how interview data is used. Mercor says it does not sell interview data or use it to train AI models.

Is the Mercor AI interview hard?

It is not a puzzle-style test. It is mostly hard because answers must be specific, concise and consistent with your resume, and because you cannot read an interviewer's reactions. Practising out loud makes the biggest difference.

How do I retake the Mercor AI interview?

Go to the Assessments tab, select Retake assessment on the interview card, choose a reason and select Retake interview. You get up to three retakes across all applications for that interview, and a retake resets your application progress.

What happens after the Mercor AI interview?

Mercor notifies you when your application moves to the next stage and gives no fixed timeline. You will not see your recording, transcript or question-level feedback.

Can I use ChatGPT during the Mercor AI interview?

No. Mercor allows AI tools only for grammar and wording checks and prohibits using LLMs to write answers, rate outputs or evaluate code. Prepare with AI beforehand instead and answer in your own words.

Does the Mercor interview always use the resume?

Mercor says interviews are customized and go beyond your resume, but candidate accounts consistently describe resume-based opening questions and follow-ups. Know every line you have listed.

Is there a Mercor AI interview for software engineers or domain experts?

Mercor offers interviews tailored to the listing, including technical and language or reasoning formats. Mercor's pages do not describe separate "coding", "domain" and "generalist" interviews, so those variants come from candidate and third-party descriptions. Read the listing and the waiting-room instructions for your role.

Can I practise for the Mercor AI interview?

Mercor's docs describe a free practice interview that is not tied to any application, though it has been described as removed in an earlier snapshot, so check your dashboard. You can also practise with a spoken AI mock interview.