AI strategy

AI adoption is an organisational design problem

Draft article · 5 min read

AI capability is improving quickly, but organisational value still depends on how work is designed.

Start with the work

The useful question is not “Where can we add AI?” It is “Which decisions, bottlenecks, quality problems, or expensive cognitive tasks should change?”

Design accountability

Every AI-enabled workflow needs clear ownership, review points, escalation paths, quality expectations, and evidence that the change is better than the previous way of working.

Build learning into delivery

Teams need safe ways to test new capabilities, capture failures, compare outcomes, and improve prompts, tools, data, and workflow design. AI adoption is therefore inseparable from product management, architecture, team design, and operational discipline.

Models are only one component

Good outcomes come from the combination of suitable models, trusted data, well-designed tools, policy controls, observability, and people who understand when not to rely on the system.