Most organisations are still treating AI as a project. It is not a project.

A project has a scope, a team, a budget and an end date. When it finishes, the organisation returns to its normal operating state. AI does not work this way. It is a restructuring force — one that is rewriting the economics of labour, software, services and competition simultaneously. Treating it like a project is one of the most expensive mistakes an organisation can make, not because the project will fail, but because the project frame guarantees you will address only a small fraction of what is actually happening.

The organisations that will be in the best position in three years are not necessarily the ones that move the fastest. They are the ones that are moving with a coherent view of where they are going. Speed without direction is just expensive drift. A strategy is not a plan for every contingency. It is a set of settled decisions that stops your organisation from relitigating the same questions every time AI comes up in a meeting.

Without a strategy, AI decisions get made anyway. A procurement manager adopts a tool that creates compliance risk. A department head pilots three platforms with no integration path. A board approves an AI ethics policy without a view of what the business is actually doing. Leadership announces an AI transformation without knowing which processes are candidates and which are not. These decisions compound. Every uncoordinated choice narrows the options for the next one.

With a strategy, those decisions have a frame. Not a bureaucratic approval process, but a settled answer to three questions. Where will we apply AI, and where will we not. How will we evaluate and prioritise AI investments when the business case is uncertain. And what are we building or protecting — what is the competitive position we want to be in, and how does AI help us get there rather than just making us busier.

The hardest part of an AI strategy is the third question. Most organisations stop at capability — what can AI do, what tools are available, what does the industry benchmark say. Fewer organisations work backwards from the competitive position. What are customers actually paying you for. Which parts of that value will AI make easier to replicate. Which parts will become more valuable precisely because they cannot be automated. The organisations that answer those questions before their competitors do are the ones that will define the reference points for their sector, and reference points matter enormously because they shape what buyers expect and what they are willing to pay.

The cost of waiting is not symmetric. If you develop a strategy and execute it well, you earn a sustainable advantage. If your competitor develops a strategy and executes it well while you are still forming a committee, you spend the next several years playing catch-up against a moving target. The gap does not stay fixed while you deliberate. It compounds.

This is not an argument for panic. Most AI disruption happens over years, not quarters. The organisations that make bad AI decisions in haste — deploying tools without data governance, cutting headcount before the replacements are reliable, outsourcing judgement to models that are not ready for it — tend to create more problems than they solve. A rushed strategy is worse than no strategy in a narrow set of circumstances. But a considered strategy developed over four to six weeks is better than eighteen months of uncoordinated experimentation, and the time to develop one is not whenever the board gets around to it. It is now, because the competitive frame is being set now.

The sector that waits does not freeze in place. It drifts. Suppliers adopt AI at different rates. Customers recalibrate expectations based on what the early movers demonstrate is possible. Talent realigns toward organisations that are using AI in ways that are interesting and visible. By the time the late mover decides it is time to act, the environment has already shifted around them, and they are making strategy in response to a landscape they did not help shape.

An AI strategy does not need to be long. It does not need to be perfect. It needs to settle the three questions — where to apply AI, how to make decisions about AI, and what competitive position you are building toward — and it needs to be owned at the leadership level rather than delegated to a technology team. Technology teams can implement a strategy. They cannot own the competitive judgements that belong to strategy.

The question is not whether AI will affect your organisation. It already is — in the tools your people are using without telling you, in the platforms your competitors are shipping, in the expectations your customers are forming. The question is whether what is happening is happening by design or by default. A strategy is not a guarantee of success. It is the difference between leading the change and inheriting it.

Published by High Impact Group on myAIstrategy.