Give candidates a role-based assignment with any AI tool allowed. Review the work, AI traces, focused debrief, and evidence-grounded report.
Candidate evidence pack
Illustrative
AI-fluency report
Launch a product nobody has heard of
78
Fluent
01
Assignment
02
AI traces
03
Debrief
04
Report
Hiring signal
They catch contradictions before they become campaign copy. Hire for judgment. Review the first draft.
A résumé or take-home can look excellent while hiding how much the candidate understood. Interview Lab makes the process part of the evidence.
The assignment
Tests work that is close to the role.
The trace
Shows what the candidate asked, checked, and changed.
The defense
Shows whether the candidate owns the decisions.
Candidate experience
Candidates receive the assignment and deadline. They can use their preferred AI tools, submit the deliverable and traces, then discuss specific decisions in a focused 30-minute debrief.
They review the sharing terms before the hiring team receives their evidence and report.
See what candidates get →Candidate trace-export instructions →Plain-language definition
An AI trace is the saved conversation from a tool such as ChatGPT, Claude, Gemini, or Cursor. It includes the prompts, the AI responses, and the follow-up instructions used to create the deliverable.
Together, these exchanges show how the candidate directed, checked, and improved the work.
Direction
“Compare these three launch options against the brief.”
Verification
“That claim conflicts with the source. Check it again.”
Iteration
“Keep the structure, but revise the recommendation.”
The format treats AI use as normal work behavior. It tests whether the candidate can direct the tool and remain accountable for the result.
Step 1
Use a curated challenge or bring a role-specific brief. The candidate gets a real work product to create.
Step 2
The candidate can use ChatGPT, Claude, Gemini, Cursor, or another tool. AI use is part of the test.
Step 3
The candidate submits the deliverable and the AI conversations behind it. Unusable traces return before scoring.
Step 4
A focused debrief probes specific trace moments. You receive the report, traces, and transcript with candidate consent.
The report scores five dimensions. It connects each assessment to the candidate’s traces, deliverable, or debrief answers.
Illustrative AI-fluency report
Marketing challenge · About 75 minutes
78
/ 100
Fluent
82
Direction
71
Verification
74
Iteration
80
Judgment
84
Ownership
Verified from the traces
“The fact sheet says $39/month unlimited. The contractor FAQ says ten swaps then $2 each. These can’t both be true. Flag it before any ad copy ships.”
If a hiring panel asked
They catch contradictions before they become campaign copy. They still let the model draft too much of the first pass. Hire them for judgment. Review the first draft.
This is an illustrative report. An actual report uses the candidate’s submitted traces, deliverable, and debrief answers.
Verification carries the most weight. A fluent candidate must do more than write effective prompts.
Verification
25%
Did the candidate test claims, catch errors, and challenge AI output?
Direction
20%
Did the candidate split the task well and give the model clear instructions?
Judgment
20%
Did the candidate make the tradeoffs that the AI could not make?
Ownership
20%
Can the candidate explain and defend the work as their own?
Iteration
15%
Did the candidate recover from weak output and improve the approach?
After Interview Lab, you can review a candidate’s public projects, verified skills, certificates, and AI Readiness Score. These signals support the work-trace evidence.
Tell us what you are hiring for. We will confirm the challenge, run the Interview Lab flow, and deliver the evidence pack.
Limited hiring-team pilot · Per-interview pricing · No software setup
Candidate practice is live. The employer-run service is a pilot for a small number of hiring teams.
Yes. The pilot can use your role-specific brief or one of our curated challenges. We confirm the scope before a candidate starts.
Any AI tool. The format tests how candidates direct, verify, improve, and own AI-assisted work.
Give them one document. It covers ChatGPT, Claude, Gemini, Cursor, Claude Code, and Codex. Copy the addendum from the enterprise workspace or send https://10xcareer.ai/interview-lab/share-traces.
Candidate practice stays private. For an employer-run interview, the hiring team receives the evidence with candidate consent.
The pilot uses per-interview pricing. Contact us with the role and expected candidate count for the current terms.
Bring one role. We will show you how Interview Lab can fit the next stage of your hiring process.