Six layers. One operating discipline.

What the layers are

The 6-Layer Architecture disaggregates institutional AI use into six operating functions: Strategy, Governance, Execution, Measurement, Optimization, and Intelligence. Unlike a maturity model (which describes where a function sits on a scale), the layers describe what the function must do — all six, always. A function operating only Strategy and Execution without Measurement and Governance accumulates risk without evidence. A function operating Governance without Intelligence is reactive rather than positioned. The 59-module Module Library maps each module to one primary layer; modules are selected by layer to fill specific operating gaps, not by category alone.

How the layers work

Concurrent, not sequential.

The 6-Layer Architecture is not a project plan. The layers do not gate each other — a function does not complete Strategy before starting Governance, or complete Governance before starting Execution. Every AI capability the function deploys operates across all six layers simultaneously: there is a strategy decision behind each deployment, a governance posture around each tool, an execution artefact tracking each pilot, a measurement cadence against each live capability, an optimization loop addressing each drift, and an intelligence posture that was informed by market scanning before the selection.

What varies by maturity band is the quality and formality of each layer's outputs, not whether the layer is active. A Foundational-band function has implicit strategy (the GC's working sense of what AI is for) rather than a written Strategy Canvas. A Defensible-band function has an explicit Strategy document, a governance committee, a live Risk Register, a quarterly Measurement cadence, a structured Optimization loop, and a continuous Intelligence feed. Both are operating all six layers; the gap is in the artefact quality and the compounding effects of formal documentation.