Structural discovery for the federal mission

The LEF SI 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.

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.

What the engine does

  1. Ingest live public federal sources, alone or woven together.
  2. Map structure — typed relationships, voids the record requires but no source fills, temporal kinetics across snapshots.
  3. Self-falsify — every candidate is tested against the records before it stands; one the records refute stays on record as refuted, never dropped.
  4. Brief the human — downloadable landscape artifact, with its methods block.

Analyst lenses, chosen for each run's records

Findings are readable as:

  • Intelligence — linkage or directional signal
  • Investigation — what live evidence substantiates or refutes
  • Program Assessment — performance vs accountability framework
  • Policy / SOP — rule with no implementing record
  • Compliance / Audit — potential gap against governing requirements
  • Adjudication — contested findings against the controlling rule
  • CQI — process weakness with a corrective-action loop

These seven are the base set. Four more come in when a run's records call for them: IG / Oversight, Budget / Appropriations, Legislative / Statutory and Risk / Threat. A lens with no finding stays empty; none is filled to fill a section.

Sample artifacts

Prior demonstration runs. Full set via 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 →

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.