AI & Tech
Subscribe for data-driven insights. No spin, just the facts.
USAFacts + Databricks
We co-hosted the Grounded Reasoning Cup with Databricks at Databricks Data+AI Summit. Top AI research teams pit their agents against some of the most challenging government documents out there.
Jun. 17, 2026
Visit the USABench
Open-source benchmark for evaluating LLM accuracy on government data. 459 evaluation questions across 3 government data sources. Top model accuracy is currently 47.4%. Transparent methodology, extensible to any public dataset, community contributions welcome.
Ask USAFacts
Ask your own questions. We'll answer with fully sources government data.
Claude Connector
Direct access to our government data for accurate, sourced answers.
Available now
Follow the Money
Explore school district data inside of ChatGPT, USAFacts style.
Autumn
We work with foundation model companies, data firms, and government agencies to make civic data more useful and AI-ready. If that sounds like you, let’s chat.