Mapping

Overview

What mapping covers and why it is two disciplines

• Mapping sits upstream of pattern selection. A workflow that has not been mapped well cannot have a pattern allocated to it with confidence, and the most common failure mode in early AI adoption is choosing a pattern from the dashboard description of the work rather than from its mapped reality. Mapping is what turns "the underwriting workflow" from a label into a thing the framework can act on. It is also where the consulting practice does its most concrete work: strategy, training and piloting all depend on a map that holds together. The discipline operates in two distinct modes that have to be chosen between deliberately rather than treated as a single activity.
• Mapping for cognition renders cognitive shape: the patterns of attention the practitioner applies, the context they hold in their head, the analogues they reach for, the moves that feel routine versus the ones that demand slowing down, the questions they ask of the work before they make any visible move. Its artefact is measured by how faithfully it renders that texture. Mapping for specification bottoms out the work as a system would need to receive it: inputs, outputs, decision rules, edge cases, failure modes, contracts with adjacent systems. Its artefact is measured by how completely it bottoms out. The two modes are not exclusive; most workflows benefit from both, and each cognitive pattern needs them in a different mix.
• Partnership benefits most from cognition-mode mapping because it runs on shared mental model and depends on the AI being inside the texture of the work. Utility demands the deepest specification-mode mapping because runtime is policy execution and any context not specified at design time is unavailable. Agency needs both: cognition mapping to author a mandate that respects how the work actually moves, and specification mapping to bottom out the cases the mandate must cover. Delegation usually needs lighter versions of each. The iterative reality is that a light first-pass map suggests a hypothesis pattern, the hypothesis tells the practitioner which mode of deeper mapping to invest in, and the deeper map confirms or refutes the call. Mapping and selection pass through each other rather than running in strict sequence.