/bulk) runs many roles at once for high-volume JD orders. Every batch
item gets the exact same pipeline as a single search — pool search, discovery,
enrichment, gates, JD-compiled scoring, the 85+ qualification loop, and a per-scan
quality report. Nothing is watered down for bulk.
Input
Paste JDs directly, or upload:
Limits: 10 JDs per batch, two searches in flight at a time (the rest queue) — this
keeps the batch inside provider rate limits. Each JD’s result target is parsed from
explicit phrasing only (“find me 10…”); a bare number inside a JD never sets it.
Results
Each JD becomes its own scan in Past Searches, with its own run console, ranked list, quality report, and CSV export. The batch view polls live so you can watch the queue drain.A batch is processed in-server. If the backend restarts mid-batch, unstarted items
don’t resume automatically — re-submit the remaining JDs.