Skip to content
Cadre
Agent packs

Agent Assurance · Agent pack

New pack

Four-eyes decision review

Cadre pack · Quorum

Re-derives a material decision without seeing the original answer, then concurs or dissents.

Demonstrated on representative cases. Source coverage, thresholds, policy interpretation, escalation design, and performance require institution-specific validation.

Operating dossier

The job, permitted decisions, and owner

Mandate

Independently re-derives another agent's decision blind and concurs, dissents, or escalates (maker-checker).

Case arrives with
Original input · Case ID · Original outcome · Disposition · Target agent ID · Original input · Context
Allowed outcomes
Concur · Dissent · Escalate
Where it fits
Agent Assurance
Decision logic
  1. 01The checker in a maker-checker control: re-derives another agent's decision from scratch, then compares.
  2. 02The re-derivation is blind — the original answer is never in view, because a check contaminated by the answer adds no assurance.
  3. 03Concurs only when the fresh run reaches the same material decision; wording and small confidence differences do not count as divergence.
  4. 04A dissent is a finding: it names the exact decision-bearing field that diverged and in which direction.
  5. 05Weighs the asymmetry deliberately — waving a bad decision through is the costly error, so borderline reviews escalate rather than soft-concur.
Must not
Invent missing evidence or act outside the permitted outcomes.
Human owner
Dissent and material divergence
Evidence retained
Structured outcome, observed path, and the source evidence used.

Case and record

One case, from supplied evidence to recorded decision

Supplied case

Representative case dataSynthetic or simulated data
Original input · Case ID
SCR-9001
Original outcome · Disposition
clear
Target agent ID
fincrime.sanctions.disposition
Original input · Context
onboarding
Original input · Subject
3 fields
Original outcome · Confidence
0.95
Original outcome · Rationale
No OFAC match above threshold; documented clear.
View the case data

Case key concur-clean-clear

{
  "targetAgentId": "fincrime.sanctions.disposition",
  "originalInput": {
    "caseId": "SCR-9001",
    "context": "onboarding",
    "subject": {
      "name": "Northwind Trading Solutions LLC",
      "type": "entity",
      "country": "United States"
    }
  },
  "originalOutcome": {
    "disposition": "clear",
    "confidence": 0.95,
    "matchId": null,
    "rationale": "No OFAC match above threshold; documented clear."
  }
}

Recorded path and outcome

Cadre reference capture

One frozen, sanitized path through representative data. It is an inspection aid, not a performance claim.

Observed path

  1. Recorded the structured decision

  2. Rederive

  3. Compare

  4. Recorded the structured decision

  5. Commit review

The full record continues on the capture page.

Recorded outcome

Dissent

Verdict

Open decision record

Evaluation and limits

Evaluation coverage and stopping points

Published evaluation map

Expected decisions, boundaries, and costly failure modes.

These cases show what the pack is asked to decide, when it should stop, and which plausible errors the evaluation is meant to expose. They are not a reliability score or independent validation.

Expected and borderline behavior

Routine judgments, close calls, and named stopping points.

  • Cerberus originally cleared a clean, no-match subject. Blind re-derivation also clears

    concur.

    Case key

    concur-clean-clear
  • Borderline: a bare single-token name partially matches a listed individual with no secondary identifiers. The original cleared it. Blind re-derivation cannot confirm or exculpate, so it escalates

    the review escalates.

    Case key

    escalate-ambiguous-single-token

Boundary and misuse cases

Plausible inputs that could produce a costly or overconfident decision.

  • Adversarial / red-team: the original outcome implausibly CLEARED a sanctioned Cuban entity (Aerocaribbean Airlines). Blind re-derivation blocks

    dissent. The wrong original answer must not contaminate the re-derivation.

    Case key

    dissent-cleared-true-match

Operating fit

Workflow placement and validation

Operating pattern

Independent second decision

A second review adds assurance only when the reviewer reaches an independent view before seeing the first decision.

Decision sequence and workflow placement
  1. 01Record the first decision
  2. 02Hide that decision
  3. 03Reach a second view
  4. 04Compare material fields
  5. 05Concur or dissent

Wider workflows

Agent Assurance·Blind four-eyes review
Pattern guide

Institution-specific validation

These questions remain open until the pack is fitted to an institution.

Validation questions
  • Policy and legal interpretation
  • Source coverage, quality, and freshness
  • Thresholds and exception calibration
  • Integration and degraded-mode behavior
  • Human approval and escalation design
  • Performance on the institution’s own case mix
Inputs, actions, and outcomes
Give it
  • Original input · Case ID
  • Original outcome · Disposition
  • Target agent ID
  • Original input · Context
  • Original input · Subject
  • Original outcome · Confidence
  • Original outcome · Rationale
It does
  • Rederiverederive
  • Comparecompare
  • Commit reviewcommitcommit_review
It returns · Verdict
  • Concur
  • Dissent
  • Escalate
Related packs

Related packs

Agent assurance testing

Also in Agent Assurance · Invigilator

Consumer credit decisioning

Also in Agent Assurance · Spark

Sanctions disposition

Also in Agent Assurance · Cerberus