The first stage decides what Mandate can honestly answer.
A context-isolated pre-screening model examines every incoming question. It catches prompt injection attempts, off-scope queries, and category misroutes before they reach the response model. Security and scope enforcement happen here, first — not after the system has already started forming an answer.
The response model never answers from general knowledge alone.
Before generating a response, the system retrieves relevant source material from a controlled corpus — bill text, Supreme Court precedent, and constitutional text. This is Retrieval-Augmented Generation: look it up first, then reason. It's the architectural choice the research on legal-AI hallucination points to most directly — see the research behind the pipeline below.
Supreme Court opinions sourced from CourtListener, a project of the Free Law Project. Executive orders sourced from the Federal Register, via the National Archives and Government Publishing Office.
That coverage runs across all three branches of government — Congress, the Supreme Court, and the White House are each tracked on their own tab on the Floor.
Majority Leader
Minority Leader
Legislative
John G. Roberts Jr.
Appointed by Pres. George W. Bush
Sworn in Sep 29, 2005 · 20 years serving
Lifetime appointment
Judicial
Donald J. Trump
47th President of the United States
Inaugurated Jan 20, 2025 · Term 2 of 2
1 year serving (current term)
Executive
Legal reasoning as scaffolding. No conclusion at the end.
The response model builds the answer using a legal-reasoning framework called IRAC, with one deliberate modification: no conclusion. The structure forces each component — issue, provision, both cases, context — into its own slot, at equal length, before the response can be finalized. See the IRAC deep-dive below ↓
Cross-check before confident hallucination becomes a problem.
For constitutional questions specifically, a cross-model advisor is invoked on demand to cross-check the generated response against enumerated powers, the Bill of Rights, and relevant precedent. It exists for one reason: confident hallucination of constitutional interpretation is the failure mode that makes legal AI dangerous. Routing a second model at this point catches contradictions the response model can't see on its own.
Three dimensions. All three must pass.
A separate cross-model auditor reads the final response and scores it on three dimensions: argument balance, editorializing, and source grounding. Responses that fail are rewritten before the user sees anything.
