--- title: "One Company Let Candidates Use AI in Interviews and Says…" canonical: "https://www.metaintro.com/blog/company-let-candidates-use-ai-hiring-got-better" language: "en" author: "drashtigarach" published: "2026-07-24T20:07:10.000Z" modified: "2026-07-25T11:18:58.015Z" --- [Back to Blog](/blog) [News](/blog/tag/news)[Interview](/blog/tag/interview)[AI](/blog/tag/ai)[Job Hunting](/blog/tag/job-hunting)[Career](/blog/tag/career) # One Company Let Candidates Use AI in Interviews and Says Hiring Got Better A tech company let candidates use Cursor, Claude Code and ChatGPT in interviews and hiring improved. What AI-allowed interviews test and how to prepare. [![Drashti Garach](https://cdn.metaintro.com/rs:fill:40:40/q:72/plain/images/5719d740-e510-42bc-8017-e040d145f35f_1766029465094.png)Drashti Garach @DrashtiGarach](/blog/author/drashtigarach) [July 24, 2026](/blog/archive/2026/07)Updated July 25, 202613 min read ![One Company Let Candidates Use AI in Interviews and Says Hiring Got Better](https://cdn.metaintro.com/rs:fill:1200:675/q:78/plain/images/company-let-candidates-use-ai-hiring-got-better-hero.png) [https://x.com/intent/tweet?text=One%20Company%20Let%20Candidates%20Use%20AI%20in%20Interviews%20and%20Says%20Hiring%20Got%20Better&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better](https://x.com/intent/tweet?text=One%20Company%20Let%20Candidates%20Use%20AI%20in%20Interviews%20and%20Says%20Hiring%20Got%20Better&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better)[http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better](http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better)[https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better&title=One%20Company%20Let%20Candidates%20Use%20AI%20in%20Interviews%20and%20Says%20Hiring%20Got%20Better](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better&title=One%20Company%20Let%20Candidates%20Use%20AI%20in%20Interviews%20and%20Says%20Hiring%20Got%20Better)[mailto:?subject=One%20Company%20Let%20Candidates%20Use%20AI%20in%20Interviews%20and%20Says%20Hiring%20Got%20Better&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better](mailto:?subject=One%20Company%20Let%20Candidates%20Use%20AI%20in%20Interviews%20and%20Says%20Hiring%20Got%20Better&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better) Warp redesigned its hiring so that candidates are free to use AI tools during interviews rather than being restricted from them, [according to Fast Company](https://www.fastcompany.com/91573546/we-let-job-candidates-use-ai-it-made-hiring-better). For engineering candidates the format is one hour, real tools like [Cursor](https://www.cursor.com/), [Claude Code](https://www.claude.com/product/claude-code) or [ChatGPT](https://chatgpt.com/), and the task of building a real feature from the company's actual product. At [Metaintro](https://www.metaintro.com), we track roughly 50 million job postings, and this shift matters far beyond one company. As AI becomes standard in knowledge work, more employers will test how you use it rather than whether you can work without it, and the preparation that wins those interviews is different from what most candidates expect. ## What Did the Company Actually Change? It changed the thing being measured. Instead of trying to prevent AI use, the company set out to measure the judgment to know when and how to use AI. That is a fundamentally different assessment target. The reasoning is practical. The company argued that AI has become part of the modern toolkit for knowledge work, and that pretending otherwise during an interview creates an artificial environment that optimises for skills people will rarely rely on once hired. An interview that bans the tools of the job tests an imaginary version of the job. The format follows directly from that logic. Engineering candidates get one hour, may use [Cursor](https://www.cursor.com/), [Claude Code](https://www.claude.com/product/claude-code), [ChatGPT](https://chatgpt.com/) or other AI tools of their choice, and are asked to build a real feature from the company's actual product with complete freedom in how they approach it. Every element of that design is deliberate. Real tools mean the test matches the job. A real feature means the problem is authentic rather than a puzzle. Complete freedom means the candidate's approach, not a prescribed path, is what gets observed. The interview becomes a compressed simulation of the work itself. ## What Separates Strong Candidates From Weak Ones? The finding is the most useful part of the whole account, because it tells you exactly what to practise. It was not prompt cleverness. The strongest candidates began by taking the time to understand the problem itself, asking thoughtful questions, clarifying constraints, identifying edge cases, and building a mental model before they started executing. The differentiator was thinking first, prompting second. The weaker performers did the opposite. They immediately began prompting, accepted AI-generated output too quickly, and only later realised they had misunderstood the underlying problem or generated a subpar answer. Speed to the tool, without comprehension, produced confident wrong answers. That contrast is the entire lesson. The most impressive candidates did not necessarily write the cleverest prompts. They exercised judgment about the problem before reaching for the tool, and judgment is what the format was built to reveal. The candidates who treated AI as a shortcut past understanding revealed exactly that. ## Why Are Employers Moving Toward AI-Allowed Interviews? Because banning the tools of the job produces bad information, and employers are noticing. An interview is meant to predict on-the-job performance, and prediction improves when the test resembles the work. If people will use AI daily once hired, an AI-free interview measures a skill they will rarely use in isolation and misses the skill they will use constantly, which is deploying AI well. The artificial environment optimises for the wrong thing. There is also an integrity dimension that AI-allowed formats quietly solve. When tools are banned, some candidates use them covertly and the process becomes a detection game. When tools are permitted and the assessment is designed around them, there is nothing to hide, and the evaluation can focus on judgment rather than policing. The problem of concealed AI use largely disappears when use is expected. For candidates, the strategic read is that this is a trend, not a one-off. As more employers reach the same conclusion, more interviews will hand you the tools and watch how you wield them. Preparing for that world is preparing for where hiring is going, not just for one unusual company. ## How Should You Prepare for an AI-Allowed Interview? Prepare your judgment, not your prompts, because judgment is what these formats reward. Practise understanding before executing. The single clearest finding was that strong candidates clarified constraints and identified edge cases before prompting. Build the habit of spending the first portion of any timed exercise defining the problem, listing what could go wrong, and forming a plan. That discipline is visible to an interviewer and it is exactly what they are scoring. Ask questions out loud. The strongest candidates asked thoughtful questions, and in an interview that behaviour is not a sign of weakness, it is evidence of the judgment being measured. Clarifying scope and assumptions demonstrates that you engage with a problem rather than rushing at it. Learn to evaluate AI output critically rather than accepting it. Weaker candidates accepted AI-generated output too quickly. Practise reading generated work with suspicion, checking it against the actual requirement, and catching where it is subtly wrong. The ability to reject or correct a plausible-but-flawed answer is a core skill in this format. Get genuinely fluent with the common tools. Being asked to use [Cursor](https://www.cursor.com/), [Claude Code](https://www.claude.com/product/claude-code) or [ChatGPT](https://chatgpt.com/) under time pressure is not the moment to learn them. Fluency lets you spend your attention on the problem rather than the interface, which is where the evaluation actually happens. Be able to explain your decisions. In a format built around judgment, narrating why you chose an approach, why you prompted a certain way, and why you accepted or rejected an output is often as important as the output itself. Practise thinking aloud so your reasoning is visible. ## What Are the Limits of This Approach? It is worth being balanced, because an AI-allowed format solves some problems and introduces others, and understanding both helps you read what an interview is really testing. The clearest benefit is realism. An assessment that hands you the tools you will actually use produces a truer prediction of on-the-job performance than one that bans them, which is the whole reason the company argued an AI-free interview creates an artificial environment. Realism is the strongest argument in favour of the approach. The clearest limit is standardisation. When every candidate approaches a real feature with complete freedom in how they tackle it, comparing two candidates becomes more subjective than scoring a fixed test. That places a heavier burden on the interviewer to assess judgment consistently, and not every organisation does that well. As a candidate, this means the quality of your interviewer matters more in these formats, and making your reasoning explicit helps a less-experienced evaluator see your strengths. A second limit is that a one-hour build under observation is still a performance, not the job itself. It rewards people who think clearly under time pressure, which is a real and relevant skill, but it can understate candidates who do their best work more slowly or collaboratively. If you are someone who excels over days rather than in a single timed hour, it is reasonable to ask whether any longer or team-based component exists, and to make sure your process is visible within whatever time you are given. None of these limits undo the core insight. Measuring how someone uses AI is more informative than measuring whether they can avoid it. But knowing the limits lets you prepare for the specific format in front of you rather than a generic one, and it tells you which of your strengths you most need to make visible. ## What if the Interview Still Bans AI? Many still do, and you need to read which kind you are in and adapt, because using the wrong strategy for the format is a self-inflicted rejection. Clarify the rules before you start. Ask explicitly whether AI tools are permitted, because guessing wrong is costly in either direction. Using AI where it is banned can end your candidacy, and refusing to use it where it is expected can make you look slow or out of step. A single clarifying question removes that risk. In an AI-free interview, the underlying skill still matters most. Even where tools are permitted, the strongest candidates succeeded through problem understanding rather than tool use, so the fundamentals you would show without AI are the same fundamentals that win with it. Comprehension, edge-case thinking and clear reasoning transfer across both formats. Treat a ban as information about the employer. An organisation that forbids the tools its people use daily may be testing an artificial environment, exactly the mismatch the AI-allowed approach was designed to fix. That is not disqualifying, but it is worth noticing as a small signal about how the company thinks about the actual work. The through-line is that judgment is the durable skill regardless of format. Whether or not you are handed AI, demonstrating that you understand a problem deeply before acting is what separates strong candidates, and that is within your control in any interview. ## What Does This Mean for Your Career Beyond Interviews? The interview is a mirror of the job, so the same judgment that wins the interview is what makes you valuable once hired. The market is shifting toward valuing judgment about AI over raw output. When anyone can generate a plausible answer instantly, the scarce and rewarded skill becomes knowing which answer is right, where it fails, and when the tool is the wrong choice entirely. The candidates who accepted output too quickly are a preview of the employees who will struggle as AI use spreads. Building that judgment is a deliberate practice, not a passive one. Every time you use an AI tool in your own work, resist the urge to accept the first output. Interrogate it, find its errors, and understand why it produced them. That habit compounds into exactly the capability employers are now designing interviews to detect. Think of AI fluency as table stakes and AI judgment as the differentiator. Knowing how to prompt is becoming common. Knowing when to prompt, how to check the result, and when to override it is rare, and rarity is what commands both offers and pay. The company in this account did not reward the cleverest prompters, it rewarded the clearest thinkers, and that is the skill worth building for the long term. Finally, position yourself honestly around this. If AI-allowed interviews are spreading, the winning move is not to hide your AI use or to lean on it as a crutch. It is to become the kind of professional who uses these tools with obvious judgment, because that is what the interview, and the job, now reward. ## How Do You Show Judgment Under Time Pressure? The hardest part of an AI-allowed interview is that judgment has to be visible inside a single observed hour, and there are concrete ways to make sure it is. Narrate a short plan before you touch the tools. Spending the first few minutes stating what the problem actually is, what the constraints are, and what could go wrong signals exactly the behaviour the strongest candidates showed. It also protects you from the trap the weaker candidates fell into, of prompting immediately and only later discovering they misunderstood the task. A visible plan is both a performance signal and a genuine safeguard. Check the first output out loud. When an AI tool returns something, resist accepting it and instead say what you are verifying and why. The weaker performers accepted AI-generated output too quickly, so demonstrating the opposite, a deliberate review of whether the output actually meets the requirement, directly displays the skill being scored. Leave a trail of your reasoning. In a subjective, freedom-based format, the interviewer is reconstructing how you think. Small spoken decisions, why you chose this approach, why you rejected that suggestion, why you tested this edge case, give them the evidence they need to rate your judgment highly. Silence forces them to guess. Manage the clock as part of the judgment. Deciding what to build first, what to leave out, and when to stop refining is itself a judgment call, and doing it visibly shows maturity. A candidate who delivers a smaller, correct, well-reasoned result within the hour usually reads better than one who chases a perfect answer and runs out of time. Taken together, these habits turn an intimidating format into a showcase. The interview is designed to reward clear thinking about how to use AI, and the candidates who make that thinking visible are the ones who win it. --- ## Related Articles - [What an AI Hiring Bias Study Reveals About Screening](https://www.metaintro.com/blog/ai-hiring-bias-study) - [What Happens When a Hiring Manager Catches You Copying Your Work Sample](https://www.metaintro.com/blog/hiring-manager-catches-you-copying-work-sample) - [The AI Skills Employers Actually Want](https://www.metaintro.com/blog/ai-skills-employers-actually-want) - [A 90 Day Plan to Build AI Skills](https://www.metaintro.com/blog/90-day-plan-to-build-ai-skills) - [Why AI Skills Are in Demand in the Public Sector](https://www.metaintro.com/blog/why-ai-engineers-in-demand-public-sector-2026) - [How to Negotiate a Job Offer](https://www.metaintro.com/blog/how-to-negotiate-a-job-offer) - [How AI Is Reshaping the Interview](https://www.metaintro.com/blog/ai-interviews-candidate-dropout-2026) - [Employers Care More About Fit Than a Polished Resume](https://www.metaintro.com/blog/employers-hiring-obsession-over-polished-resume) --- ## People Also Asked ### Q: Can I use AI in a job interview? A: Sometimes yes and sometimes no, so always ask first. Some employers now explicitly allow tools like [Cursor](https://www.cursor.com/), [Claude Code](https://www.claude.com/product/claude-code) or [ChatGPT](https://chatgpt.com/) and design the assessment around them, while others still ban them. Clarifying the rules before you start is essential, because using AI where it is prohibited can end your candidacy. ### Q: What do AI-allowed interviews actually test? A: They test judgment about when and how to use AI, not prompt cleverness. In the account here, the strongest candidates understood the problem, asked questions and identified edge cases before prompting, while weaker ones prompted immediately and accepted flawed output too quickly. ### Q: How do I prepare for an interview that lets me use AI? A: Practise understanding a problem before executing, get fluent with common tools in advance, and learn to check AI output critically rather than accepting it. Being able to explain why you made each decision matters as much as the result, because judgment is what these formats reward. Prepare to stand out. [Metaintro](https://www.metaintro.com) helps you match to roles where your real skills and judgment fit, so you can walk into any interview format prepared. [Create your free profile](https://www.metaintro.com/signup) and find the jobs that value how you actually work. ### Share this article [https://x.com/intent/tweet?text=One%20Company%20Let%20Candidates%20Use%20AI%20in%20Interviews%20and%20Says%20Hiring%20Got%20Better&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better](https://x.com/intent/tweet?text=One%20Company%20Let%20Candidates%20Use%20AI%20in%20Interviews%20and%20Says%20Hiring%20Got%20Better&url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better)[http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better](http://www.facebook.com/sharer.php?u=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better)[https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better&title=One%20Company%20Let%20Candidates%20Use%20AI%20in%20Interviews%20and%20Says%20Hiring%20Got%20Better](https://www.linkedin.com/sharing/share-offsite/?url=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better&title=One%20Company%20Let%20Candidates%20Use%20AI%20in%20Interviews%20and%20Says%20Hiring%20Got%20Better)[mailto:?subject=One%20Company%20Let%20Candidates%20Use%20AI%20in%20Interviews%20and%20Says%20Hiring%20Got%20Better&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better](mailto:?subject=One%20Company%20Let%20Candidates%20Use%20AI%20in%20Interviews%20and%20Says%20Hiring%20Got%20Better&body=https%3A%2F%2Fwww.metaintro.com%2Fblog%2Fcompany-let-candidates-use-ai-hiring-got-better) For job seekers ## Spend your time on the work, not the search. Free to start. No résumé needed. Real jobs, picked for you, by text. [Get Started Free](/signup)[Search matching jobs](/jobs/search) - MatchCompare live roles against your current evidence. - PositionTurn proof projects into role-specific applications. - ImproveUse market feedback to keep the skill plan current. [Return to navigation](#main-navigation)