Your board chair wants to see an “AI action plan”? This is it.

Your board chair wants to see an AI action plan.

You could hand over a list of forty use cases. Or you could answer the question the list is avoiding: not where can we use AI, but which work should AI actually do.

A small fraction of automation projects carry most of the returns, so more pilots isn't progress. The discipline is task-level: find where variation defeats rules because that's the place where AI earns its keep. Everywhere else, conventional automation is cheaper to build and easier to audit.

Discover, pick, deliver — the three steps are in the carousel. The third is where most programmes quietly stall, and rarely for lack of engineers, but mostly because product leaders accountable for the results are just not there.

Strategy is also a lot of what not to do. Most of our work on these is talking clients out of things.

Send us your AI ambition. We will tell you how to deliver it.