AI Real-Estate Deal Sourcing

The problem

A developer was drowning in broker listings: hundreds of emails and PDFs a week, with no fast way to find the few parcels worth pursuing. We built a pipeline that scrapes the sources, structures every listing, and scores them against the developer's buy-box. It's the same site, zoning, and budget instinct a GC uses in preconstruction, just a different customer.

What we built

• Scrapers for broker sites and offering memorandums
• Structured extraction: acreage, price, zoning, traffic counts, contacts
• Scoring against a configurable buy-box (size, price band, location quality)
• Ranked dashboard with filters, map view, and at-risk flags

Signal

• MVP validated with the client
• Real listings flowing into a daily-refreshed ranked queue