About

Policy intelligence,
built on the record.

State Affairs is the policy intelligence and operating infrastructure for the people who work in, track, and shape state policy. We start from primary sources — the bills, administrative rules, and statutes themselves — and build everything else on top of them.

52States and chambers
615Documents indexed
741Retrievable passages
290.7KWords of legal text
DailyIngestion cadence
How it works

Three layers, in order.

01

Ingest

A nightly job pulls every state's current session, compares a change hash per bill, and fetches only what moved. Text versions are decoded and stripped to plain text; administrative rules and statutes are collected from the states' own codes.

02

Structure

Documents are split along their own boundaries — SECTION, ARTICLE, §, RULE — so a retrieved passage is a coherent provision rather than an arbitrary window. Each passage is indexed twice: lexically in Postgres, semantically as a vector.

03

Answer

A question runs against both indexes and the results are merged by rank. The model sees only those passages and is told to cite them — and to say so when they do not contain the answer.

What we believe

Nonpartisan, and
traceable to the source.

Principled nonpartisanship

The product reports what the documents say. There is no editorial thumb on the scale, in either direction.

Show the work

Every generated answer names the sections it came from. If you cannot check it, you should not rely on it.

Primary sources first

Summaries drift. The text does not. Everything is anchored to the document as filed.

Say when you do not know

A system that admits a gap is more useful than one that fills it confidently.

Want to see it?

Bring an issue and a state. We will show you the documents behind the answer.