Asked to Redo the Exercise Live Without AI: Why, and What Passes

Why Are Employers Adding a Live Redo?

Loops are adding a short, unaided round because polished take-homes stopped separating anyone. In a Robert Half survey of more than 2,000 US hiring managers, two-thirds said reviewing AI-generated applications had slowed their hiring process, and more than a third reported running more interviews per candidate in response ¹. None of that is about you specifically.

A separate vendor survey of hiring managers in the UK, Ireland and Germany found roughly a third naming detection of AI-generated work among their top challenges, and about as many saying they now run more in-person, on-site interviews ². That is the shape of the round arriving in your inbox: a stage built into loops by default, the way a phone screen or a reference check became routine, not a flag raised on your file alone.

The mechanism behind it shows up elsewhere too. On one freelance platform, after an AI cover-letter tool was introduced, the link between how closely a letter matched a job post and whether it earned a callback fell by roughly half, because the polish stopped carrying information and evaluators leaned on other evidence instead ⁴. A live, explain-it-yourself round is the hiring side's answer to that exact problem.

A few things are usually true about a live redo, when it is being run well:

• It goes to every finalist at that stage, not a chosen few
• It replaces or follows a take-home rather than punishing you for having submitted one
• It is scored against the same rubric as the original exercise, not a stricter one

Ask Whether Everyone Gets the Round

One plain question separates a routine stage from something worth pushing back on: does every finalist sit the same live round, or only you. Employers who design this stage well are told to run the live version with the assistant open and no stopwatch once the job's unit of work fits inside an hour, which is exactly what makes the round passable by anyone who did their own thinking the first time.

Most candidates are doing the same thing you are. In one vendor survey of jobseekers in the UK, Ireland and Germany, roughly three in four said they use AI to tailor a CV or application at least some of the time ³. A round built to catch an unusual minority would not need to run on every finalist; one built because polished submissions now look alike would.

If a recruiter cannot say who else gets the round, that is worth naming directly rather than guessing at what it means. A vague or defensive answer tells you more than the invitation email did.

What Actually Gets Checked in the Room?

The round checks whether you can explain your own reasoning without a script to lean on: why you framed the problem the way you did, what you tried that did not work, and what you would change if you started again now. None of that requires banning AI from the version you built at home. It requires that the person who submitted the work can still walk through it live.

What a good follow-up question is built to expose is exactly this: not how you sound under pressure, but whether the reasoning behind an answer is actually yours. A recruiter asking you to trace one decision back to its reason is running the same test a strong interview already runs, just with the earlier draft in front of both of you instead of a fresh prompt.

That is also why a live redo is not really testing how much AI you used the first time. Two candidates could have leaned on an assistant equally hard and still land on opposite sides of this round: one can say what the assistant proposed, what looked wrong, and what changed before it shipped, and the other cannot reconstruct any of it. The second case is the gap that fails a candidate.

If you genuinely cannot explain a decision in your own submission, that gap already existed before anyone asked you to redo it. The live round just made it visible, which is a different thing from the round having created it.

Prepare in Ten Minutes, Not an Evening

Skip the all-nighter. Reread your own submission once, out loud if that helps, and write down one thing you would change if you did it again today. That answer is the thing a well-run round is built to hear, because what gets scored is whether the reasoning is yours, not whether the output is flawless a second time. If a claim in your work rested on something you looked up, know which claim it was and why you trusted it.

A few habits make the room easier:

• Narrate your thinking as you go, including the parts that do not work
• Say out loud when you are unsure, rather than guessing silently
• If the invitation felt like being falsely accused of using AI and you are still unsettled, it is fair to ask how the round is scored before you start

The goal is not performing confidence. A candidate who says "I am not sure why I chose this, let me think it through" and then actually reasons it out usually reads better than one who bluffs.

The same logic holds even if the redo replaces AI use entirely for the live portion. Working unaided for twenty minutes does not erase the work you did with an assistant the first time; it just asks you to show the part of that work that was always yours, the framing and the judgment calls, without the tool sitting between you and the interviewer.

See what gets scored >