How matching works
A candidate and a posting match when every one of the following checks passes. Each check is a stated requirement or a stated term — matching is deterministic rule-checking, computed identically for everyone.
- The posting is open and discloses its compensation range.
- The posting's work model is one the candidate accepts, and the currencies match.
- The top of the posting's range meets the candidate's stated minimum (if they set one).
- Every evidence requirement on the posting is met by the candidate's approved claims at the required strength or higher — stated, has a source, or confirmed by a collaborator.
- No application or connection already exists between them.
What is deliberately absent
- No scores, no rankings: the employer's qualified pool is alphabetical, and no candidate is ever rated against another.
- No machine decisions about people: AI writes explanations of which checks passed; humans make every accept, decline, and hiring decision, and every outcome is notified.
- No hidden signals: nothing outside the criteria above — not activity, not typing speed, not who paid — influences who matches.
How much of the pool an employer sees
Being discoverable puts you in a pool, not in a directory. Verifying a domain proves an employer controls it; it does not prove the posting is real. So the number of candidates who match is free to read, and the names are rationed:
- A newly verified organization can open 25 candidate cards a day. That rises once a candidate has accepted a connection from them and their domain has been verified for a week.
- An organization with a posting taken down, or with reports from several visitors, can open none until that is resolved.
- An employer can only invite someone whose card they were shown.
- Every card opened is recorded — which organization, which candidate, which day — and that record is readable by the candidate it names.
Discoverability is opt-in and reversible. Connection requests expire visibly after 14 days. AI-generated content on this site is always labeled. Privacy · Terms