Federal demonstrator Unclassified · not FedRAMP

Structural discovery for the federal mission

The LEF Ai Engine reads multi-source public federal data as structure — voids, cross-corpus gaps, indications of change — and falsifies its own findings before a human analyst decides. Presence-based search cannot do this: a gap has no document to return.

What the engine does

  1. Ingest live public federal sources, or a multi-snapshot upload.
  2. Map structure — typed relationships, voids the record requires but no source fills, temporal kinetics across snapshots.
  3. Self-falsify — a contradiction pass suppresses weak candidates before they surface.
  4. Brief the human — downloadable landscape artifact, with its methods block.

Seven analyst lenses (every run)

Findings are readable as:

Sample artifacts

Prior demonstration runs. Full set via Federal Reports and Civic Reports in the header.

Cross-corpus Census × FDIC × NIH × USGS

Where structure lives in one federal corpus but never another — multi-source weave.

Open landscape PDF → Temporal Velocity & moving gaps

Indications of change: emerging silence vs decaying silence across snapshots.

Open landscape PDF → Acquisition / finance USAspending × Census × Congress × FEC

Cross-corpus read of awards, legislation, and finance structure on public data.

Open landscape PDF →

Deployment posture

The invite demo is an unclassified demonstrator in a containerized Google Cloud service — isolated and access-gated, but not an accredited or FedRAMP-authorized boundary. Run unclassified / public data only. An accredited enclave deployment is available as a funded effort.

Have an invitation code

Unlocks the live showcase (public federal sources, alone or woven).

Request an invitation

For TPOCs and program evaluators. Reviewed personally. Enclave inquiries welcome on the same address.

licensing@livingedenframeworks.com

Outputs are structural arguments for a human analyst. The engine recommends; it does not author the decision.