PAGA Monitor

Turns California's PAGA labor-filing database into scored, searchable leads.

PAGA notices are public records, but they sit behind a Salesforce search portal with no API, no feed, and no alerts. For plaintiff firms, seeing a filing first is the whole game. PAGA Monitor polls the portal, dedupes what it finds, and scores each case so the best leads surface immediately.

Call the portal's own API, not its HTML

The portal is a Salesforce Visualforce app. Instead of parsing rendered tables, a headless browser submits the real search form (required for the signed session) and then invokes the site's own remoting endpoint, getting a month of filings as structured JSON in about three seconds.

Ingestion and serving are split on purpose

A hosted, browser-based scraper is unreliable, so the Playwright worker runs separately and writes to Postgres; the Flask dashboard is stateless and only reads it, so it deploys cleanly to Vercel with zero knowledge of how the data got there.