Frame
Define the intelligence question, decision relevance, scope, and the distinction the assessment must resolve.
Adversary Analysis Division
Authored intelligence. Evidence before publication.
AAD turns structured holdings into reviewed intelligence products. It is designed to keep collection, assessment, approval, and publication distinct—and to refuse claims the evidence does not carry.
Division mandate
Judgment remains human, attributable, and reviewable.
AAD does not collect independently or create the evidence used to clear its own products. Analysts author judgments against structured holdings, record assumptions and alternatives, and route products through defined review gates.
ScopeOperational and strategic adversary assessment across kinetic,
strategic, information, economic, and cyber domains.
InputExisting entities, relationships, hypotheses, and source
records maintained by the wider architecture.
ControlRead-only analytical access is the intended design
boundary.
OutputAuthored assessments, collection requirements, corrections, and
publication records.
Analytical method
Every judgment carries its supporting structure.
Assess
State key judgments with confidence, reasoning type, assumptions, and explicit alternatives.
Grade
Evaluate source structure and independence without treating corroboration as proof of factual truth.
Challenge
Preserve competing hypotheses and identify the evidence most likely to discriminate between them.
Publish
Release only after the applicable evidence, review, and approval conditions are satisfied.
Product forms
Structured outputs for different analytical decisions.
Operational assessment
Focused judgment
A bounded assessment of a defined adversary question, with key judgments, evidence grades, assumptions, and alternatives.
Comprehensive package
Integrated picture
A consolidated actor, organization, event, relationship, and hypothesis record supporting deeper review.
Collection requirement
Discriminating evidence
A prioritized statement of what is unknown and what additional collection would change the assessment.
Correction record
Visible revision
A durable record of substantive changes, why they were made, and how the resulting judgment differs.
Representative assessment
Tren de Aragua organizational structure
A current division package examines U.S. legal designations, cell-level hierarchy, transnational command, and the decision consequences of centralized versus franchise models.
EstablishedConcurrent U.S. designations under separate legal
authorities.
SupportedDifferentiated functional roles and hierarchy within examined
cells.
UnresolvedWhether Venezuela-based leadership operationally directs
foreign cells.
PriorityEvidence of documented cross-border tasking by central
leadership.
This example summarizes a draft internal product dated August 9, 2026.
It is presented to demonstrate analytical structure, not as a released
intelligence judgment.
Current research
Behavioral signatures: testing structure against response, not description.
Method
Pairs a stress event — a legal designation, a prosecution wave, a leadership loss — with the response a named organizational archetype predicts: directed, federated, or franchised.
Standard
An archetype is admitted only if it passes an identifiability test: at least one stress kind must produce an observation that could not also be true of a rival archetype. Without that, it is recorded as undistinguished rather than typed by preference.
Finding
The original stress vocabulary had no category for legal designation — the most consequential shock in the corpus fit none of the existing kinds. The gap was added as a distinct category rather than forced into a neighboring one.
Finding
The evidentiary window assumption failed: the first post-shock record for each tracked subject arrived 230 and 902 days after the shock, both outside the window the model was using. Absence read as unassessed for months — not because collection failed, but because the record did not yet exist.
Correction
A grading review found two arms of one government counted as independent sources. Correcting it took the division's highest-grade findings from 24 to 7.
Applied to Tren de Aragua and the 'Ndrangheta under active
tracking. No archetype has yet been assigned sufficient evidence
to stand — every recorded stress event remains single-origin. That
is stated as the current limit of the method, not omitted.
The adjudication interface itself. Case identifiers and exact shock dates are redacted for this public view; the underlying verdicts and evidence counts are shown as recorded.