Legislative Intelligence
What is legislative intelligence?
Legislative intelligence is an evolving application of AI to bill tracking, text analysis, and predictive scoring. Applied well, it answers what a bill actually says, who backs it, and how likely it is to pass, not just its current status.
Start for freeNo login required.
A real question, answered
Ask like you would ask a colleague.
No dashboard, no filters to build.
A question in plain language reaches bill text, sponsor history, and momentum in one pass.
Updated September 2026
What the term means
What legislative intelligence means
Raw AI chat answers from training data, which goes stale the moment a bill moves in committee and rarely cites a primary source. Apogee runs the same conversational interface on live retrieval against Congress.gov, GPO bill text, and CRS reports, so an answer traces back to the actual record instead of a language model's recall of it.
The distinction matters most on two fronts competitors treat as separate products: prediction and coverage. Passage prediction only works as a transparent, signal-based model here, because so few bills are ever enacted that there are not enough positive examples for a machine-learning model to train on reliably (the exact base rate is below, in the prediction section). And coverage spans Congress and all 50 states plus DC in one search, not a federal product with state tracking sold as an upsell.
No analyst firm defines "legislative intelligence" as a category. There is no Gartner Magic Quadrant, Forrester Wave, or IDC Market Guide for it; G2 files every vendor in this space under the generic "Public Affairs and Advocacy" bucket. Bloomberg Government, FiscalNote, Quorum, and others use the term because it describes what the product does, not because an analyst firm coined or ranked it. The claim above is a description of a real product, not a category an outside body has validated.
An unresolved question
What "intelligence" is supposed to mean is still unsettled
Vendors in this space do not agree with each other on what the word adds.
Aggregation
Connected search across sources that used to require separate tools.
Prediction
A viability score attached to a bill.
Human judgment
A dataset cannot replace it; a legislator takes a call because of a track record, not a dashboard.
There is no single accepted definition, and no analyst firm has settled the argument, as the section above shows. That is part of why this page treats legislative intelligence as an evolving application of AI rather than a fixed, finished category.
The gap in a status feed
Why a keyword alert is not enough
Even vendors selling keyword alerts acknowledge the failure mode. FiscalNote's own blog states the problem directly: a tracking tool's job is not to send every match, it is to send the right ones early enough to act, and most tools get the first part right and the second part wrong. The result trains the person receiving the alert to stop reading it.
Keyword matching also misses bills by design, not by accident, as one competitor's own blog concedes: a bill can impose a real, specific obligation on an industry without ever using that industry's own vocabulary; a bill that never says "data privacy" can still regulate how data is handled. A keyword alert matches the words a bill contains, not what the bill actually does.
Why live retrieval matters
Where the official record breaks down
The official record is not always as available as it should be. Congress did not publish bulk legislative data for years; GovTrack.us ran screen scrapers reverse-engineering the old THOMAS.gov system for roughly fifteen years to fill the gap, work its founder has documented in detail. Congress.gov's own public API went dark for an extended stretch in August 2025, cutting off the tools that depend on it from real-time access.
Published reviews of established bill-tracking vendors describe a related complaint: a real lag between something happening in a bill's history and that event showing up in the tool, workable for general monitoring but too slow to act on in the moment it matters. Retrieving directly from the primary source at query time, instead of on a batch schedule, is the direct answer to that lag, though no pipeline built on a patchwork of independently maintained government sites is ever perfectly instantaneous.
State legislatures make this harder still. Every state publishes bill data on its own site, in its own format, with no shared standard, which is why the open-data project Open States runs roughly 200 separate scrapers, one per state and chamber, each hand-maintained against a source that can change without notice. Fifty-state coverage is a real, ongoing engineering commitment, not a single number to claim once; Massachusetts carries the deepest state-level depth today, with live bill text, committee detail, and roll-call votes close to federal fidelity, while the rest of the 50 states plus DC are searchable at a lighter level.
Nine research capabilities
From a status field to the full picture
01
Bill search and tracking
Search federal legislation by keyword, sponsor, committee, status, or bill number, from the 93rd Congress (1973) to the current one. Semantic search matches meaning, so a question about "clean energy tax incentives" also surfaces bills that only say "Section 45 production tax credit."
02
Bill text Q&A
Ask a direct question about a bill's full text and get an answer with section and provision citations. Bill text is retrieved from GPO XML, split into sections, and searched by retrieval-augmented generation, so a 900-page appropriations bill answers as fast as a two-page resolution.
03
Bill comparison
Compare two bills side by side: shared provisions, scope differences, funding mechanisms, enforcement approaches. Built for House-Senate companion bills, competing proposals on the same issue, or a bill against its prior-Congress version.
04
Cosponsor analysis
See every cosponsor on a bill, broken down by party, state, chamber, and the date each one signed on. Pairs with cosponsor prediction to forecast who is statistically likely to join next.
See cosponsor prediction in Network & Relationship Intelligence →05
CRS report intelligence
Search and summarize Congressional Research Service reports, the nonpartisan analyses Congress itself relies on. Hybrid keyword and semantic search finds the relevant report even without the title or number; summaries are generated from the full text, not an abstract.
06
Legislative history timeline
The full chronological action record for any bill: introduction, referral, hearings, markup, amendments, floor votes, conference, signature or veto. A bill stalled in committee for six months tells a different story than one fast-tracked last week, and the timeline shows which one it is.
07
Legislative momentum scoring
A composite score built from cosponsor velocity, media coverage surge, lobbying-registration activity, hearing attention, and floor-speech mentions, normalized so bills at different baseline activity levels are comparable. Leaderboard mode ranks the full Congress; single-bill mode breaks down each signal for one bill.
08
Bill universe
One query returns everything on record about a bill: sponsors and cosponsors, committee assignments, news coverage, organizational support or opposition, hearing testimony, and lobbying registrations, traversed from the knowledge graph in a single pass instead of five separate searches.
09
Bill passage prediction
A transparent, signal-by-signal estimate of whether a bill advances, covering committee and floor movement, a companion bill in the other chamber, cosponsor count and growth, bipartisan support, sponsor influence, lobbying and media activity, and the historical base rate for that bill type. Every signal is visible, so the score explains itself.
In practice
Who uses legislative intelligence
Lobbyists
Rely most on cosponsor analysis and passage prediction: where a bill stands and where it is going.
Trade-association staff
Covering dozens of jurisdictions at once, rely most on fifty-state coverage.
Advocacy organizations
Building a public case rely most on bill text Q&A and the legislative history timeline.
In-house policy teams
Monitor what a bill means for their industry before it reaches committee.
The term itself is shifting for a specific reason. Bill-tracking software has existed for decades as a status lookup: a bill number, a stage, a link. Full bill text, committee reports, floor transcripts, and cosponsor history have since become searchable in the same place that status field lives, and a prediction can now attach to that status instead of standing alone as a separate, unexplained score. Vendors reached for "legislative intelligence" to describe that shift, because "tracker" no longer covered what the product did. The term stays unsettled, as the absence of an analyst-defined category earlier on this page shows, but the shift it describes, from a status field to a searchable, predictive record, is real and visible in what these products ship today.
The hard problem
A prediction that explains itself
Enacted bills are rare: 1.3 to 2.5 percent of introductions per Congress. That scarcity breaks ordinary machine learning, which needs many positive examples to train on. Apogee scores passage likelihood from named signals instead of a trained black box, so a low or high score always comes with a reason.
Leaderboard mode ranks every bill in the current Congress by passage likelihood. Single-bill mode breaks down one bill's score signal by signal, so the factor driving it, a stalled committee vote, a missing companion bill, a bipartisan cosponsor surge, is visible, not inferred.
Raw accuracy numbers in this category deserve scrutiny. The researcher most cited on bill-outcome prediction, John Nay, published the warning himself: a 2017 peer-reviewed paper in PLoS ONE notes that a model which always predicts a bill will fail scores roughly 96 percent accurate on historical data, because so few bills are ever enacted, without having learned anything. A separate report the following year put a commercial product tied to Nay's own venture at 98 to 99 percent accuracy, close enough to that naive baseline to raise the same question his own paper raised. Political scientist John Wilkerson, reviewing this category of model for the journal Science, called the approach promising but added a caution of his own: it does not teach anyone about process, strategy, or politics. A percentage alone does not say whether a model beats the base rate; a score that names its signals does.
Signals in the model
Every one visible on the score, none of them hidden.
- 1 Committee and floor advancement
- 2 A companion bill moving in the other chamber
- 3 Cosponsor count and growth velocity
- 4 Bipartisan support
- 5 Sponsor influence (leadership, committee chairs)
- 6 Lobbying registrations and media coverage
- 7 Hearing and amendment activity
- 8 Historical base rate by bill type and policy area
The demand side
Why Capitol Hill cannot close this gap alone
Congress's own staff face the same information problem from the inside, not just outside groups looking in. The Congressional Management Foundation's 2017 survey of senior Hill staff found a wide gap between how important staff called access to nonpartisan policy expertise and how satisfied they actually were with what they had.
POPVOX Foundation has described this as one of three separate pacing problems facing Congress: the pace of industry against the pace of Congress, the pace of Congress against the executive branch, and an internal pacing problem, Congress's own operations running slower than the information arriving at its door.
Appropriations bills show the same strain from the outside. No regular appropriations bill has been enacted on time since fiscal year 1997, and most now move as omnibus packages bundling many individual bills into one, a structure the Library of Congress notes requires disentangling the bill into its original parts to trace, and one a Brookings analysis argues is built in part specifically to limit scrutiny.
CMF 2017 Hill staff survey
81%
called nonpartisan policy expertise very important
24%
said they were actually very satisfied
A 57-point gap between the need and the reality.
- Constituent contacts grew 300%+, 1995 to 2004
- No regular appropriations bill on time since FY1997
Choosing a platform
Legislative intelligence vs. basic bill tracking
| A basic bill tracker | Apogee | |
|---|---|---|
| What a question returns | A status field and a link to the bill. | A synthesized, cited answer drawn from the bill text, its full action history, and everything connected to it. |
| Predicting outcomes | Not offered, or a single opaque score with no stated inputs. | A multi-signal model that names every factor driving the score, built because only 1.3 to 2.5 percent of bills are enacted per Congress, too few positive cases for a black-box model to learn reliably. |
| Reaching it | A dashboard with its own login, its own filters, its own learning curve. | The chat interface already open, whether that is Claude, ChatGPT, Gemini, or Copilot. |
| Coverage | Federal only, or state coverage sold as a separate, priced-up module. | Congress and all 50 states plus DC, searchable in the same conversation. |
| Pricing | Quote-gated, seat-priced, a sales call required to see a number. | Published pricing, self-serve signup, no contract. |
Who gets left out
Cost has historically decided who gets this kind of information
Published reviews of established vendors on sites like G2 and Capterra include recurring complaints about cost relative to actual use, disproportionately from smaller nonprofits and advocacy groups priced against the same enterprise tier as a large lobbying shop. One documented workaround: the League of Women Voters holds a single shared subscription at the national level and distributes tracking sheets across its state chapters rather than every chapter paying separately.
Published, self-serve pricing removes the need for that kind of workaround: the number is visible before a sales call is ever required.
Comparing specific platforms
A direct, named comparison for each incumbent.
Example queries
"Which bills have the most momentum right now?"
Try it →"What are the chances H.R. 1234 actually passes this Congress?"
Try it →"Compare the House and Senate versions of the reconciliation bill"
Try it →"Find CRS reports on Section 230 reform"
Try it →"What does the CHIPS Act say about semiconductor manufacturing subsidies?"
Try it →"Give me the full picture on HR 1: sponsors, lobbying, news, hearings"
Try it →Full capability documentation → See influence & coalition prediction →
Frequently asked questions
What is legislative intelligence software?
Software that combines bill tracking, full-text analysis, and predictive scoring into one system, instead of running a separate tool for each. It answers what a bill says, who backs it, how fast it is gaining ground, and how likely it is to pass, on top of tracking its status.
What is the difference between legislative tracking and legislative intelligence?
Most vendors draw the same line: tracking tells you what is happening, intelligence is supposed to tell you what it means and what to do next. In practice that means a status field and a link versus what a bill's text actually says and how likely it is to move. That framing comes almost entirely from vendors selling the newer term, not from an independent standard, so treat it as a real distinction in practice rather than a settled definition.
Does legislative intelligence require AI?
Not by definition, but every current vendor using the term builds it on AI, mainly for full-text search that matches meaning instead of exact keywords, and for passage prediction. A tool that only tracks status without either does not typically get called legislative intelligence.
Is legislative intelligence a recognized industry category?
No analyst firm defines it as one. There is no Gartner Magic Quadrant, Forrester Wave, or IDC Market Guide for legislative intelligence; G2 files every vendor in the space under the generic Public Affairs and Advocacy bucket. Vendors use the term because it describes what their products do, not because an outside body ranked or coined it.
Does "intelligence" mean prediction, analysis, or just aggregation?
Vendors do not agree. Some treat it as connected search across sources that used to require separate tools. Others treat it mainly as a prediction score. Others insist real intelligence is human judgment a dataset cannot replace. There is no single accepted definition.
Why is predicting whether a bill will pass difficult?
Only about 1.3 to 2.5 percent of introduced bills are enacted per Congress, which means a model that always predicts failure already scores roughly 96 percent accurate without learning anything. Marketing claims of 98 to 99 percent accuracy sit close enough to that naive baseline to deserve scrutiny; a transparent, signal-by-signal score is more defensible than a single accuracy number.
How is Apogee different from a basic bill tracker?
A bill tracker returns a status field and a link. Apogee returns a synthesized, cited answer built from the bill's full text, its complete action history, and everything connected to it, plus a transparent, signal-by-signal passage prediction, reached in the same chat already in use rather than a separate dashboard.
Does Apogee cover state legislatures, or just Congress?
Both. Bills and legislators are searchable across Congress and all 50 states plus DC. State legislative staff directories are live nationwide for signed-in users, and Massachusetts carries the deepest state-level coverage, with live bill text, committee detail, and roll-call votes. Momentum scoring and passage prediction currently run on federal data only and are not yet offered for state bills.
What sources does Apogee cite?
Congress.gov, the Government Publishing Office, the Congressional Research Service, the Federal Register, Regulations.gov, the FEC, CBO, GAO, the Library of Congress, and the Census Bureau, plus state legislative sources for the 50-state bill and legislator data. Every answer links back to the underlying filing.
How much does Apogee cost?
Pricing is published and self-serve, with no sales call required to see it. Current tiers are on the pricing page.
Ready for legislative intelligence that cites its sources?
Free to start. No sales call. Works with Claude, ChatGPT, Gemini and Copilot.