About

The Houston Open Data Explorer shows what happens when an AI analyst sits on top of data the public already owns. The City of Houston publishes hundreds of GIS layers and millions of records, but each lives in its own portal, format and coordinate system. Here they are loaded into one PostGIS database, joined with Census, CDC, FEMA and EPA data, and made answerable in plain English.

How an answer is made

1. You ask a question.
2. The model reads a data dictionary of 120+ datasets, picks the relevant ones, and inspects their fields.
3. It writes SQL, runs it against a read-only copy of the data (30–45 second limit per query), and checks the results.
4. It answers with numbers, a chart or a map, and the caveats that apply. Every step and every query is shown so you can check the work.

Limits worth knowing

• Answers are only as good as the source data: crime is geocoded to the hundred-block, FEMA claims are released only at ~10 km precision, and census figures are survey estimates.
• Population for neighborhoods is estimated from census tracts by land area.
• AI models can make mistakes. Check the queries behind any number you plan to rely on.
• Questions and answers are logged to improve the service. Don't include personal information.

Sources

City of Houston GIS (geohub.houstontx.gov, houston-mycity.opendata.arcgis.com) · Houston 311 public data extracts · Harris County Flood Control District high-water marks · FEMA National Flood Hazard Layer, NFIP claims and Individual Assistance · EPA Toxics Release Inventory · CDC PLACES and CDC/ATSDR Social Vulnerability Index · U.S. Census Bureau ACS 2020–2024 and TIGER.

Built for the Houston Hackathon 2026.