The take-home assessment platform

Run take-homes properly, without the operational tax.

The take-home assessment platform. Send one tokenised link for any format (writing, strategy, design, code). Claude reads the submission against your grading guide and drafts the review, turning hours of grading into minutes.

No credit card. Five AI reviews a month on the free plan (enough to try it on your next few submissions).

See the review
Inside a Graden review

This is what lands on the reviewer's screen.

Pick a role you're hiring for, then step through the three seats that matter: the admin who sends the brief, the candidate who does the work, and the reviewer who opens a review Claude has already drafted.

Correctness

78/100

Concurrency

62/100

Tests

85/100

Solid small ledger. Domain modelling is clear, the transfer API is well-named, and the spec suite covers the happy paths and a few sharp edges. The weakest spot is concurrency: the transfer isn't wrapped in a transaction, so a crash between withdraw and deposit would leave accounts out of step.

Spec suite is properly scoped. transfer_service_spec.rb covers overdraft, self-transfer and zero-amount cases, which most mid candidates skip. transfer_service_spec.rb:14-48 · tests
Withdraw and deposit run outside a surrounding transaction. A crash between the two lines leaves the ledger unbalanced. The candidate called this out in their answers, so ask how they'd fix it. transfer_service.rb:40-41 · concurrency safety
transfer_service.rb lines 38–42
38  def call(from:, to:, amount:)
39    from.lock!
40    from.withdraw(amount)
41    to.deposit(amount)   # ← no surrounding transaction
42  end

Wrap both sides in ActiveRecord::Base.transaction. As written, a crash after the withdraw silently loses the deposit.

Want the long version, with source viewer, questionnaire cross-check and audit trail?

Portrait of Sev, founder of Graden

At Shopify, Monday.com and Uswitch, we all defaulted to HackerRank and LeetCode — not because we thought they predicted great hires, but because they were the only thing we could operationally manage. Running real take-homes meant hours of engineering time per candidate, and nobody had that to spare. Graden is the tool I kept wishing existed.

— Sev, founder

One platform for the whole take-home loop.

Brief, invite, submission, review, debrief. No spreadsheets, no chasing, no reviewer scrambling the morning of the call.

1

Send the brief

Pick from the library or upload your own with a grading guide. The candidate gets a tokenised link, a clear deadline, and no account to create.

2

Track and collect

See who's opened, who's working, who's submitted. Whatever the format (inline writing, files, links, a GitHub repo), the work is locked at submission so later edits don't move the goalposts.

3

Review, prepared

Claude reads the submission against your grading guide and drafts scores, flags and talking points. You read the work yourself, keep what lands, overrule what doesn't.

Powered by Anthropic's Claude.

Most take-home reviews happen ten minutes before the call. Ours don't.

The brief is the easy part. The review is where quality usually dies. We fix that.

One pipeline, any format

Writing samples, campaign decks, strategy briefs, design exercises, GitHub repos. Same workflow, same grading guide structure, same calibrated review.

A coding-test tool for your engineer. A shared doc for your marketer. A Figma link for your designer. Graden runs all of them through the same prepared review.

Human in charge, always

Your grading guide, your criteria, your weights. Claude drafts the scores and talking points, the reviewer reads the work, agrees or overrules, and makes the call. Scores are inputs to judgment, not verdicts.

The reviewer lands prepared

Every review arrives with a summary, flags tied to each grading criterion, and specific things to discuss. Integrity checks on the candidate's explanations of their own work. No more opening the submission ten minutes before the call.

The questions you'd ask in the demo call

Short, direct answers. If something's not here, ask us.

Does the AI decide who to hire?

No. Graden drafts scores, flags and talking points against your grading guide. The reviewer reads the work, agrees or overrules, adds their own notes, and makes the call. We never surface a "verdict" to the reviewer or the candidate.

What kinds of take-homes does this work for?

Anything you can score against a grading guide. Writing samples, content briefs, strategy decks, marketing campaigns, design exercises, analytics tasks, code submissions. If you can score it against criteria, Graden can run the whole loop. The pipeline is the same regardless of format.

Is our candidates' work safe?

Yes. Submissions are captured the moment they're sent and stored in encrypted storage, scoped to your organisation. For code, the candidate installs the Graden GitHub App on the single repo they're submitting (read-only, per-repo, revocable). Nothing is shared outside your workspace.

What if the AI gets something wrong?

It will, sometimes. That's why every talking point links back to the grading criterion it's scoring and, where relevant, the specific file, page or passage. You mark a point as agreed, push back on it, or add your own. The review you save is yours, not the model's.

How do you handle late submissions?

Flagged, not blocked. The reviewer sees that a submission was late and by how long, and decides what to do. We're not in the business of disqualifying candidates on your behalf.

Free to start. Paid plans from £49/month. Priced per review, not per seat. See pricing →

The take-home platform your reviewers will actually use.

Send the brief, track the submission, open a review that is already drafted. Take-homes, without the operational tax.