Structural Taxonomy
A repeatable method for operational judgement.
Broker Intelligence Framework gives operators and assessors a common method for examining exceptions across fragmented brokerage environments. It does not replace firm procedures or authoritative systems. It disciplines how evidence is used when the answer is not obvious.
01 — THE CYCLE
The Intelligence Cycle
01
Observe the event without prematurely explaining it.
02
Classify the available information.
03
Identify the authoritative source for each material question.
04
Reconstruct the relevant workflow and state changes.
05
Test contradictions, missing evidence and alternative explanations.
06
Determine operational significance.
07
Decide action, monitoring, escalation or no action.
08
Assign the correct owner.
09
Retain the evidence and reasoning.
10
Close the immediate review.
02 — THE LANGUAGE
Evidence Language
| Confirmed | Traceable to live evidence, official documentation or a named validator with authority for the point. |
|---|---|
| Observed | Directly visible in the available evidence, without yet establishing the cause or full significance. |
| Inferred | A reasoned explanation supported by facts but not yet validated. |
| Unknown | The answer is not established by the evidence currently available. |
| Gap | Evidence, control, ownership or documentation required for a reliable conclusion is absent. |
Control principle: Plausibility, repetition, seniority and AI-generated explanations do not convert an inference into confirmed evidence.
03 — THE ASSESSMENT
What is Assessed
- Evidence discipline and traceability
- Source hierarchy
- Workflow reconstruction
- Operational judgement
- Scope and time control
- Ownership and escalation quality
- Closure discipline
- Operational continuity
04 — AI BOUNDARIES
AI may support the investigation. It does not establish the operational truth.
- The framework teaches disciplined use of AI where it can improve structure, consistency and review efficiency, while preserving the authority of live systems, official documentation and named validators.
- AI may structure an evidence register, identify missing fields, surface contradictions, test whether a conclusion is supported and draft feedback for human review.
- Material AI assistance must be disclosed where it affects the investigation or assessment output.
- Every AI-derived explanation remains Inferred until verified against an authoritative source.
- Client, employee, firm-confidential and vendor-confidential data must not be entered into an uncontrolled model.
- Operators are assessed on whether they challenge, verify and correctly classify AI output rather than accepting it uncritically.
- Final capability classifications, critical-failure decisions and deployment recommendations remain assessor-controlled.
Methodology boundary:AI is an analytical support layer within the method. It is not evidence, an assessor of record, regulatory assurance or a substitute for accountable human judgement.