AI strategy is a business decision, not a technology one

Every week, a new capability makes headlines and a new set of vendors promises to modernise your business. The temptation is to treat AI as a technology decision: which model, which platform, which partner. The evidence suggests the leaders are asking a different question entirely — what does this change about the economics of a process we already run?

Start with the process, not the model. Pick the three or four workflows where your cost, cycle time, or quality has the most headroom. Estimate what a modest improvement is worth in cash terms. Only then evaluate the technology.

Strategy first, architecture second

The organisations that fail treat AI as a project. The ones that succeed treat it as a change in how work happens — which makes it an operating model decision, a risk decision, and a people decision. Technology is the last and easiest part.

The competitive advantage is rarely the model. It's the speed at which you can test a real workflow, measure it honestly, and decide to scale it.

A practical starting point

• Map the workflows where a 20–30% improvement is material in cash terms.
• Estimate the cost of building, running, and governing each one.
• Sequence the three that clear an honest return threshold.
• Assign an owner whose bonus is attached to the outcome, not the deployment.

When AI strategy is framed as a business decision, it stops being a technology risk and becomes a return on a portfolio — which is a conversation every leadership team already knows how to have.