Having an AI strategy is not the same as having a strategy. A strategy answers where the company competes, why it wins, and what it sacrifices; AI is a capability that may serve that answer or have nothing to do with it. When the strategic document starts with the technology, what is being built is an adoption plan with an ambitious name.
The difference becomes clear in one question: if the technology were to change its name tomorrow, would the plan still make sense?
Why do our customers choose us? Price, speed, specialization, trust, coverage. The answer determines which capabilities are worth strengthening and which are irrelevant, however good they may be.
What part of our operation explains the margin? Not all processes are equally important. Two or three explain profitability; the rest are necessary but not differentiating.
What aren't we going to do? A strategy without compromises is a wish list. It's the part that's most often omitted and yet requires the most discipline.
With those three answers, the conversation about technology takes shape: invest where it reinforces the reason why you win, buy standard where you only need to operate, and discard the rest without guilt.
Level | Question that answers | Who decides |
|---|---|---|
Strategy | Where we compete and why we win | General management |
Capabilities | What do we need to know how to do very well? | Business Management |
Processes | How do we do it and at what cost? | Operational Managers |
Technology | What tools do we use to support it? | Technology, with business |
The common mistake is not making bad technological decisions: it is making level 4 decisions without having answered levels 1 and 2. When that happens, investments are justified by comparison with the competition rather than by their relationship to one's own advantage.
Four steps that fit in a quarter:
A company that does this will have two improved and demonstrable processes by the end of the year. One that launched eight initiatives will have eight pilot programs and an awkward conversation during the budget review.
None of this implies that technology is irrelevant in the strategic conversation. It contributes three concrete things:
Expand the realm of possibility. There are business models that weren't viable three years ago because the cost of certain operations was prohibitive. Stanford HAI documented a drop in the cost of inference of more than 280 times between the end of 2022 and the end of 2024, and that reopens options worth revisiting.
It changes the cost structure of competitors. Even if you do nothing, they will, and that can alter the market price.
It turns latent assets into exploitable ones. Data that you already have but couldn't use.
All three are legitimate starting points for a strategic discussion. None of them is, in and of itself, a strategy.
Before any technology plan, a final check: If this initiative were to go perfectly, what would change in the company's competitive position?
If the answer describes an operational improvement, it's a good efficiency project and should be presented as such. If it describes a new reason why a customer would choose this company over another, then it is part of the strategy.
Both things deserve a budget. Only one deserves to be called a strategy.
No: it's a capability. A strategy answers where the company competes, why it wins, and what it gives up. AI can serve that answer or be irrelevant to it. If a plan loses its meaning when the technology's name changes, it was an adoption plan, not a strategy.
Because it has become widespread. McKinsey documented in 2025 that 881% of organizations use AI in some function, but only 61% see a significant impact on EBIT. When everyone adopts the same thing, the advantage has to come from somewhere else.
Three: why customers choose us, what part of the operation explains the margin, and what we've decided not to do. With those answers, the technology decisions take care of themselves: we invest where we strengthen our competitive advantage and buy standard solutions where we simply need to operate.
Looking for what has been decided not to do. A plan with five fronts open simultaneously is a catalog: with limited resources, it produces five half-finished projects. S&P Global estimated in 2025 that 421,000 companies had abandoned most of their AI initiatives.
In four steps: write the competitive advantage in a sentence without adjectives, identify the two processes that support it, measure them, and choose a single intervention per process, taking it to production before opening the next one.
Three legitimate entry points: expand the space for viable business models as costs fall, alter competitors' cost structures even without taking action, and make existing data exploitable. None of these, on its own, constitutes a strategy.
Does your technology plan reinforce your competitive advantage or just follow it? We help connect technology investment with the processes that explain your margin, and discard what doesn't. Let's talk → |
Español Español 
English
Français
العربية
Português
Deutsch
Italiano
