Every night while I sleep, a Playwright scraper hits 14 Texas county deed record portals and logs 125 qualified leads to my CRM.
Here is how CrawlOS works.
The Problem With Paid Data
Real estate lead services charge $200 to $500 per month for the same public data that county appraisal districts publish for free.
I built a scraper instead.
The Architecture
CrawlOS is a Python + Playwright system running on lb-telecom-01 (my Vultr VPS in Dallas).
It hits the public-facing portal for each county, extracts recent deed filings, filters by acquisition criteria, and inserts qualifying records into a Supabase table.
The 14 Counties
Harris, Fort Bend, Montgomery, Brazoria, Galveston, Chambers, Liberty, Waller, Austin, Colorado, Wharton, Matagorda, Jackson, and Victoria.
All Gulf Coast and surrounding markets — the territory I know from six years of O&G structural work in the region.
The Filter Logic
Not every deed transfer is a lead. The scraper applies filters:
- Transfer type: warranty deed, not quitclaim
- Price range: within acquisition parameters
- Property type: residential or light commercial
- Recency: filed within the last 30 days
Anything passing all four filters gets inserted. Everything else is discarded.
The Output
125 leads per night on average. Each record includes parcel ID, grantor/grantee, transfer amount, address, and filing date.
Load Bearing Capital works the list. Petroleum Noir runs a parallel filter on mineral rights transfers from the same pipeline.
Why This Matters
The data is public. It has always been public. The advantage is not access — it is automation.
Anyone can pull Harris County deed records manually. Almost nobody does it every night across 14 counties without lifting a finger.
