Hiring measures the wrong thing.

1. 01
   Interviews test memorization

   LeetCode rewards pattern matching, a skill a model already has. It measures the wrong thing.
2. 02
   Resumes are self-reported

   Titles, skills, and achievements nobody verifies: easy to inflate, easier to generate.
3. 03
   Talent is invisible

   The strongest reasoning often comes from outside the schools and networks recruiters already trust.

Process over output

Anyone can produce a plausible answer now. What's scarce is the reasoning that gets there: the dead ends, the checks, the change of mind.

Observable, not self-reported

We score behavior in the trace, not a claim on a form.

Real problems only

Open questions without an answer key, not puzzles with one.

Compounding, not lone genius

An idea gets stronger when someone else builds on it. The graph is the point.

Thinkers

Show how you reason through a real problem, build a profile that updates itself, and get found by teams who read the trace instead of the resume.

Companies

See how a candidate thinks before the first call. Post your own problems through a concierge pilot.

Discuria is an early beta. The judge is a language model scoring against a fixed rubric; it has not yet been checked against a panel of human raters. Problems, scoring, the solution graph and a first company pipeline are live; search across profiles for companies is still being built. We'd rather tell you what's not done than print a number we can't defend.