Setting Up a Take-Home Assessment

A take-home assessment is a coding task a candidate completes on their own time, before an interview. The AI writes each candidate’s problem statement, evaluates their submission once they upload it, and builds a scorecard for you to review — you don’t grade anything by hand.

​

Adding a Take-Home Round

​

Timed Assessments

A take-home normally runs across the whole 5-day window. Turn on Timed assessment instead, and candidates get a fixed amount of time from the moment they start, for example one hour. When you switch it on, a Time limit box appears. Enter whole minutes from 15 to 480; it starts at 60. The Deadline note changes to match: “Candidates have 5 days to start. Once they start, they have 60 minutes to finish and submit.” Changing Timed assessment or the time limit applies to the invitations you send from then on, not to candidates already invited.

​

What the Candidate Experiences

The invitation tells candidates the assessment is timed. They can start any time in the 5 days, and the brief stays hidden until they do. Once they start, they see the time left at the top of their assignment page, and the problem is sized to fit the limit. With Enable vibe coding on, setup installs everything before the timer starts (see AI-Assisted Coding).

​

When Time Runs Out

1. AI access stops. On a round with Enable vibe coding, the AI assistant stops answering.
2. Submission stays open for 5 more minutes, so an upload started in the last minute still lands.
3. Submission then closes. The candidate’s page asks them to contact the recruiter. In the Interview Hub, the interview shows as Expired.

To let the candidate hand in after that, open the job’s Interview Hub, click ⋮ on their row and choose Reopen Submission, then pick how long they have to upload: 30 minutes, 1 hour, 2 hours or 1 day. Reopening gives back uploading only. The AI assistant stays off and the timer doesn’t restart.

​

Auto vs Guided Mode

Every take-home round has an Interview Guidance Mode with two options.

​

Auto

Leave this on and the AI builds the whole assessment for you — it writes the problem statement and the evaluation rubric from the role and the evaluation criteria you’ve already set up, and invents the scenario the problem is set in. No extra input needed.

​

Guided

Switch to Guided and a free-text box appears for direction. Whatever you type becomes the AI’s top-priority instruction — it reshapes the scenario, the rubric, and what the candidate is asked to submit. If you write the problem itself in the box, it is used as you wrote it, with only the layout tidied. This suits roles where the task is a case rather than code, such as finance or operations. If you write direction only, the AI writes the problem for you. A few things worth keeping in mind when you write guidance:

• Call out the specific skill you care about most for the role, not just the general subject area
• Add any expectations that go beyond “does it run,” if they matter for this role
• Reinforce seniority expectations if you want a different bar than what the AI infers from the job posting

​

Every Candidate Gets a Unique Problem

• Every candidate gets a different problem. The AI writes a fresh scenario for each candidate applying to the same job, so candidates can’t share or look up each other’s answers.
• But the bar stays the same for everyone. The difficulty, the rubric, and what candidates need to submit stay identical across candidates for the same job — only the scenario or story around the problem changes.
• The scenario is invented, not your business. Unless your guidance sets the scene, the AI makes up the business and the story for each problem, and never names your company or its products.

​

Shaping the Evaluation Rubric

The AI automatically builds an evaluation rubric for the take-home — you never have to write one from scratch. It’s grouped into a small set of candidate-facing categories, generated from your evaluation criteria and your Guided-mode text. There’s no separate screen where you hand-edit these rubric groups directly. Instead, you shape the rubric in two ways:

1. Your Guided-mode text is the strongest lever — it shifts what the whole rubric weighs most heavily.
2. Per-criterion instructions on the job’s Interview Evaluation page. Each criterion has its own free-text field, “Guidance for Hello AI,” treated as a strict, must-follow instruction when scoring that specific criterion. See Evaluation Criteria for more on setting these up.

​

What the Candidate Experiences

Candidates get an email invitation with a link, a deadline, and instructions — you don’t need to walk them through it. They read the problem statement, solve it on their own machine, and upload a ZIP of their work. On a timed round, they start their timer first, and the problem statement opens with it (see Timed Assessments). See the Take-Home Assessment candidate guide for the full walkthrough.

​

Reviewing a Submission

Once a candidate submits, open their profile, go to the Evaluation tab, choose the take-home round from the round menu, and switch to Interview. You’ll see an Assessment Analysis section with:

• Problem Statement — the task the candidate was given
• Submission Files — their submitted files, with AI-highlighted evidence pinned to relevant code

In the submission’s header, Evaluation shows the rubric score. Open it to see the score for each rubric category, and select a category to read the AI’s reasoning and the evidence behind it. You’ll also see the same requirement match, risks and concerns, and skill proficiencies sections shared by every interview round type, on the candidate’s Overview tab.

​

Related Guides