Watch an LLM read the job market the way a tired applicant wishes it would.
Pick a candidate. Real postings from this morning's sweep stream through deterministic eligibility gates, the survivors get scored live by Claude with the reasoning shown, and the numbers are honest — most jobs are a bad fit, and it says so.
loading today's sweep…
Who's looking?
Processed in-memory for this one scoring run. Never stored, never logged, never used for anything else.
one Claude call · ~1¢ · rate-limited + budget-capped (that's part of the demo)
Eligibility gates
Deterministic, instant, free — the pipeline's own prefilter verdicts on today's real postings. No tokens spent saying no.
Live scoring
The gate survivors go to Claude with JobScout's honest rubric: clean title + seniority + eligibility baselines ~70, niches are a bonus never a requirement, and a poor fit gets called a poor fit.
scoring with Claude…