A legislative-intelligence system that reads how a whole chamber votes.

Context

Understanding how a state legislature actually behaves (who votes with whom, which blocs really exist, who’s quietly absent on the hard votes, whose money moves them) is buried across bills, roll-call votes, campaign-finance filings, and lobby registrations that don’t talk to each other.

What we built

We built a legislative-intelligence platform that ingests all of it (bills and roll-call votes via a legislative API, the legislature’s own author reports, and campaign-finance and lobbying data) and computes the analytics on top: head-to-head agreement between any two legislators, voting-bloc detection, party-loyalty and contrarian scoring, absenteeism including “strategic absence” detection, a bipartisan-bill finder, and donor and lobby influence mapping, plus configurable scorecards. A Python pipeline handles ingestion, entity matching, and dedup; a Next.js dashboard (bills, legislators, votes, finance, lobby, analysis, scorecards) surfaces it, over real data spanning five legislative sessions.

The value

• Turns four disconnected public data sources into one queryable intelligence system. Analysis that used to take an analyst weeks is now a dashboard.
• Surfaces patterns humans miss: voting blocs, strategic absences, and money-to-vote correlations computed across an entire chamber.
• Configurable scorecards let anyone hold legislators to a consistent, transparent standard instead of anecdote.
• Built on real data across five sessions (2017–2026), so trends and trajectories are visible, not just snapshots.