Network & Relationship Intelligence

A member's influence, voting bloc, and likely next cosponsor, computed, not looked up.

Network and relationship intelligence is an evolving application of AI to legislative connections: PageRank on cosponsorship, Louvain community detection on votes, and similarity scoring on bill history. Applied well, it answers who actually holds power, which coalitions are real, and who is likely to join a bill next, not just who sits on which committee.

Start for free

No login required.

The underlying data is a knowledge graph of 200,000-plus entities: members, bills, committees, PACs, organizations, and the relationships between them. Graph algorithms run on that structure every six hours: PageRank and Betweenness Centrality for influence and bridge position, Louvain for community detection, Jaccard similarity for cosponsorship overlap. A directory or a keyword search over the same data cannot surface any of this. It has to be computed from the structure of the relationships, not read off a single record.

What gets computed

Five of the sixteen live capabilities

PageRank + Betweenness Centrality

Member influence ranking

Ranks members by two different kinds of power: PageRank surfaces the legislative hubs, members well connected to other well-connected members. Betweenness Centrality surfaces the brokers, members who sit on the shortest path between otherwise-separated groups. A committee roster shows rank. This shows who actually moves things.

Louvain community detection

Voting bloc identification

Groups members into the coalitions they actually vote with, based on shared roll-call votes, not the party label on their door. The interesting result is usually at the edges: a moderate who clusters with the other party on one issue area, or a faction that splits from its own leadership.

Community detection + bridge scoring

Legislative broker detection

Finds the members who connect two voting blocs, not just the most influential member overall. A broker is not always the most powerful name in the room, but they are the one with real relationships in both camps, the person to approach first when a bill needs votes from both sides.

Jaccard similarity on cosponsorship history

Cosponsor prediction

Given a bill's current sponsors, forecasts who is statistically likely to sign on next, ranked by how closely a member's past cosponsorship pattern matches the bill's existing coalition. Narrows 535 members to a realistic target list for a whip count or coalition build.

Cross-community pair analysis

Unusual alliance detection

Surfaces member pairs who cooperate more than their party or voting bloc would predict, a signal of emerging consensus, geographic overlap, or shared constituent interest before it becomes a public position.

See all sixteen capabilities in the full documentation →

A real question, answered

Ask for the coalition, not the committee list.

Deep Thinking
Powered by Apogee Policy Graph

Why this matters for a whip count or a coalition build

A whip operation or an advocacy campaign eventually asks the same question: given the sponsors already on a bill, who else is realistically gettable, and who is the right person to approach first for a hard-to-reach faction. Cosponsor prediction narrows 535 members to a ranked list based on demonstrated behavior, not guesswork. Broker detection identifies the specific member with real relationships in both camps, the one whose support is most likely to move others with it.

None of this is visible from a committee roster or a title. It is visible only by computing it from the actual structure of who votes and cosponsors with whom, which is what this capability set does.

See the coalition behind any bill.

Free to start. No sales call. Works with Claude, ChatGPT, Gemini and Copilot.