FastOpenDataFastOpenData is a two-person project. There's no company and no funding round behind it — just two co-founders building an ETL pipeline and API for public U.S. address, geographic, and demographic data, in the open, incrementally.
The idea is straightforward: public data — Census, TIGER, OpenStreetMap, and more — is scattered across CSVs, shapefiles, and government FTP servers, and every project that wants it ends up rebuilding the same address-to-tract joins from scratch. The goal here is to build that join once, properly, and expose it twice: as a REST API for programs that know what they want, and as an MCP server for agents that have to find out. The second one is not a wrapper around the first — an agent needs to search a variable dictionary and name its columns, and it gets metered on how much of the dataset it reads rather than on how often it asks.
Today, that goal is partly built: the geocoding endpoint works, the address lookup returns tract, county, state and PUMA attributes, and the ETL pipelines that ingest Census, TIGER, and OpenStreetMap data already exist in the codebase for a separate graph-processing project. The agent interface is deployed and answers protocol requests, but nobody can sign into it yet — the piece that issues tokens does not exist, and the endpoint refuses rather than pretending. Closing that gap — an agent interface with no way in — is the next phase of work, not a finished feature.
This site will get more honest, not less, as the project matures — pages will move from "planned" to "live" as the work actually lands.