thinkingtrace
Sign in Book a demo

The technical interview platform for the agentic era.

Book a demo

The job has changed, but interviews have not.

Before

Outdated interviews build outdated teams.

SWEs and FDEs analyze data, scope solutions, assign tasks to agents and test agent work. Today's interviews test none of it.

No understanding of AI-fluency.

Without encouraging and having visibility into AI usage, interviews don't reflect performance in the agentic era.

Take-homes and work trials are time consuming.

Candidates spend hours or weekends, and some refuse work trials altogether. Interviewers then spend hours evaluating.

Thinkingtrace

Up-to-date interviews for the agentic era.

Candidates solve real, system-scope problems in their local dev environment with coding agents; how engineers work today.

Full visibility into how candidates work with AI.

Every submission comes with a clear trace of agent interactions, models and tools used by the candidate.

Async support, with rich evals in minutes.

Solve ThinkingTrace interviews anytime, anywhere. Submissions are auto-evaluated on technical judgment and AI collaboration.

ThinkingTrace offers AI-native interviews that reveal how candidates think.

For hiring teams

ThinkingTrace fits your hiring loop.

Works where you do

Integrates with your ATS. Assign and review interviews in Slack.

Real SWE, FDE and PM work

Turn folders, repos, or GitHub issues into scoped interviews in minutes.

Evolve with the times

Candidates interview in their local environments instead of a static platform. So you are always evaluating how they work with the latest tools and agents people actually use.

Async or live

Use it as a take-home, a paired session with a co-interviewer, or a follow-up deep dive.

Custom rubrics

Add any custom scoring criteria to our eval suite.

Rich evals in minutes

Submissions are immediately evaluated on our suite of rubrics for technical judgment and AI collaboration.

Tests and rubrics are calibrated as coding agents and tools improve.

Evaluation

Gauge technical judgment and agent usage at scale.

Thinking Traces

See what candidates assume, question, design and test to reveal engineering judgment.

AI fluency

Full visibility into agent usage, interactions, management and more to find AI-native candidates.

Cross-candidate

Each submission is compared to every other one to show you what makes this candidate's approach unique.

Hard to fake

Open-ended problems and AI-use visibility shows what each candidate did.

For candidates

A frictionless candidate experience.

  1. Open a ThinkingTrace interview in your local dev env

    No synthetic sandbox, just your favorite IDE, terminal and workflows.

  2. With your preferred coding agents

    Claude, Cursor, Codex, anything else.

  3. On real, system-scope problems

    Showcase how you scope, design, manage and test.

  4. Asynchronously or live

    Work on ThinkingTrace anytime, anywhere.

Don't evaluate candidates on what AI can do.
Learn how they think and use AI instead.