Direction is a leadership decision — before it is a technology one.

Most leadership teams are not short of AI ideas. They are short of an agreed basis for choosing between them. Proposals arrive from departments, vendors and the board at the same time — each reasonable on its own terms, none measured against a shared view of what the business is trying to strengthen. The result is motion without direction.

Pilots, demonstrations and vendor proposals are visible, and visibility is easily mistaken for progress. But a demonstration proves that something is possible — not that it is worth doing. Activity is not evidence of strategic value. When leadership has not first established what AI is intended to strengthen, more activity simply raises the cost of eventually choosing.

The more expensive mistake is rarely a technical one. It is solving the wrong problem well — committing capability, capital and organisational attention to work that does not strengthen what makes the business distinctive. Direction comes first because it is the only thing that tells you which questions are worth answering, which evidence is worth gathering, and which specialists are worth engaging.

Sometimes the appropriate direction is not AI at all. Where existing software, conventional analytics, process improvement or automation provides the better answer, that is the answer — and saying so is part of the work.