The Sourcing Agent turns your job description into a structured spec, runs a live search, and screens every profile against your criteria with the biased fields stripped out. Then it critiques its own coverage criterion by criterion and re-sources the gaps — up to five rounds — so you always get at least 35 ranked profiles, each with per-criterion evidence in full sentences.
It reads the role and produces a structured spec — titles, seniority band, must-haves, locations, dealbreakers, and four to six gradeable criteria. Not a keyword string: a plan you could hand to a human sourcer.
After every round it measures how many of the slate actually meet each criterion. Where coverage is thin, it writes new targeted queries for that specific gap and searches again — up to five rounds, then it stops on its own.
Every profile arrives with a per-criterion verdict and a full sentence explaining what in their background earned it. You can disagree with the call, because you can see the reasoning.
Or point it at a job you already posted.
Screens, scores, finds its own gaps, searches again.
With the evidence behind every position.