What AI Amplifiesframing, boldness, systems thinking
The clearest look I have had at this compression up close came from an associate we had recently promoted. He came up through our year-long trainee programme, meticulous and consistent, the kind of junior a manager builds a clean promotion case around. Early in 2026, as we were bringing in Claude Code, Codex, and Antigravity, I set him a simple arrangement on an in-house project, a 360-review tool built over our existing HRIS: he would frame the problem and let the models carry the layout exploration.
He turned out approaches fast enough that management had real options to weigh, sat with the PM to pressure-test what was buildable, and somewhere in the middle wrote a small MCP tool that pulled the open and resolved comments off his Figma files into a Google Sheet, so he always knew which threads were live before a review. His seniors started asking how he had done it. His craft did not thin under the tools. It sharpened.
A model can fill an afternoon with acceptable layouts. It cannot yet sit in the meeting where one of them is chosen and defended. When AI executes mid-spectrum work competently, two things become proportionally more valuable: the inputs that determine what should be built, and the judgement that organises execution into coherent wholes. Compression of one dimension forces value upstream, into framing, and outward, into systems. Three capabilities sit at those edges, and each is developable through practices earlier parts of the book describe.
Problem framing (4.2) converts the brief as received into the brief as diagnosed, moving from symptom to condition to root cause. As AI can execute anything a team specifies, value shifts toward knowing what to specify. A practitioner who can frame an ambiguous organisational concern into a precise, investigable problem is worth disproportionately more, because the alternative, AI executing the wrong specification competently, is now high-cost and fast. In the pre-AI era, a team that specified the wrong problem paid in wasted sprints. Now it pays in a week, ships the wrong product, and discovers the problem at scale.