Hiring at scale is broken because the only mass-scale screener most teams have is résumé filtering — and a résumé is a weak, unfair proxy for who can actually do the job. The usual fix, competitive-coding screens, selects for puzzle-solving, not for the ability to ship real production code.

Then agents changed what matters. When an AI can write the loop for you, the durable, high-value skill is knowing what to build and how to structure it — and being able to own the result. So we built a screener around that: ship real code against a realistic, role-specific challenge (use any agent you like), pass a deterministic gateway, get your code graded against a rubric, and then defend your reasoning in an unaided, adaptive conversation.

We deliberately don't try to detect AI. The signal isn't “did a human type every line.” It's “can this person ship, reason about, and own real work in our stack — with or without an agent.” That's the only question worth answering, and it's the one we answer.