Inside open-ended AI coding interviews
See how candidates use their own editor and AI tools to scope ambiguous work, steer the model, and verify the result.
How candidates are working
AI harness mix
Share of sessions by the tool candidates drove
Workflow habits
How candidates put their tools to work
14%Planned first
24%Used subagents
14%Used skills
78%Ran tests
Drive, don't ride
Outcomes by how many messages the candidate sent
Hire or better
No hire
Browse Strong Published sessions
input