Redesign your org to be AI-first

The classic organisational redesign is a structural exercise: spans and layers, reporting lines, a new operating model on a slide. The work itself, the actual tasks that fill everyone's week, is assumed to carry over unchanged into the new boxes. For decades that assumption held well enough. AI breaks it, because AI changes what the work is before you have decided who does it.

Designing AI-first means inverting the sequence. Start from the task layer: which tasks AI absorbs, which it augments, which stay distinctly human, and only then ask what roles those remaining bundles of work add up to. Done this way, the answers stop resembling the old chart with fewer heads. Overlapping roles combine into hybrids. Routine-heavy roles compress. And entirely new roles emerge from the freed capacity, which is the upside case most redesigns never even model.

Design the structure around the work AI leaves behind, not around the work you used to have.

The other failure mode of the big-bang restructure is certainty. A design is announced as a decree, the organisation discovers its flaws in production, and the correction costs more than the design. Scenarios are the honest alternative: every proposed change modelled on the graph first, capacity impact, skills required, people affected, transition path, so leadership decides between evidenced options rather than defending a guess. Agents propose; your people decide.

What this programme produces is a structure you can defend line by line, to the board and to the workforce, because every change traces to task-level evidence rather than to a template from someone else's transformation.