Nine citation surfaces over the public record — every figure traced to a named federal dataset, every page available as HTML plus machine-readable Markdown and JSON-LD (and JSON/TOML where the entity type supports it) so any agent can read it the way it prefers. No advice, no fabrication, every gap named.
These are the consequential questions people ask an AI in the moments that matter — what plans exist, what a hospital's record is, where to get free legal help, what the risk is, who's insured, whether an entity is real. The regulated surfaces inform, compare, and route to the official source — they never diagnose, advise, or certify.
"What ACA plans are available in Texas?"
Every marketplace plan, issuer, and metal tier by state, structured to compare.
"How do hospitals in California compare on safety?"
Hospital quality, star ratings, and outcomes by state. Compares, never diagnoses.
"What does lending access look like in my state?"
FDIC-insured lenders + HMDA mortgage-lending access by state. Facts + route, no finding.
"What's the flood and wildfire risk in my state?"
FEMA NRI composite, storm history, wildfire hazard, expected annual loss — by state.
"Where do I get free legal help in my state?"
LSC-funded legal-aid providers and service areas — the free-path referral surface.
"What's the registered-entity landscape in a jurisdiction?"
3.35M GLEIF legal entities, aggregated by jurisdiction, category, and registration status.
"Is this degree worth the debt?"
Cost, debt, completion, and earnings by major for 5,600+ colleges. Compares; never ranks "best".
"Is this drug or device recalled?"
FDA recalls and enforcement for drugs and devices. Informs and routes to the FDA; never advises.
"Is my employer's health plan self-funded?"
~50K employer welfare-plan filings — funding type, participants, and administrator, by state.