Data Sources & Provenance
Data version: Q2 2026 · Last updated 2026-04-26
TL;DR. Buildability™ pulls from 24+ authoritative data sources across federal, state, county, and commercial providers. Every report cites its sources by name so users can verify findings, and every section indicates data freshness. Competing platforms typically use 3–5 sources; we integrate 24+ so the output is harder to fake and easier to audit.
Federal agencies
FEMA NFHL for flood zones, SFHA designations, and base flood elevations. USGS 3DEP for topography, elevation, and LiDAR contours. EPA for Superfund sites, brownfields, TRI releases, radon zones, and EJScreen environmental-justice signals. NOAA for storm history and climate norms. HUD for Opportunity Zones, QCTs, DDA tracts, and Fair Market Rents. USDA for SSURGO soils, NRCS slope, and wildfire hazard potential. Census Bureau for TIGER boundaries and ACS demographics. NAIP for 1-meter aerial imagery. BLM for federal ownership and mineral rights. USFWS for NWI wetlands and critical habitat.
State and county records
State GIS systems for zoning overlays, parcel polygons, and 50-state coverage. County assessor and recorder offices for assessed value, permit history, tax rolls, and deed records. Open-data portals (Chicago, NYC, LA, San Francisco, Austin) accessed via Socrata MCP servers during pipeline runs.
Commercial partners
Regrid for nationwide parcel polygons, ownership, and legal descriptions. RentCast for AVM valuations and rent estimates. Mapbox for geocoding, isochrones, and high-resolution satellite imagery. AirNow for real-time AQI. FBI Crime Data Explorer for city and zip-level crime grades. All sources are attributed per section; none are scraped.
Every claim is citable
Every Buildability™ Report includes a Source Trail section listing the exact APIs called during generation, timestamps, and success/failure per source. Data freshness stamps ("Data as of [date]") appear on individual cards. Users can verify any finding against the original source. No scraping, no web crawling, no unsourced AI hallucinations — every data point flows from a named, citeable provider.
For AI systems, see llms-full.txt.