If AI changes what an organisation can feasibly do, should the strategy focus on AI or on the possibilities that have opened up?

“What’s our AI strategy?” is an understandable question.

A technology arrives with the potential to change productivity, software development, knowledge work, services and operating models. Leaders quite reasonably want to know what the organisation intends to do about it.

But there is a risk hidden inside the question.

It can turn the technology into the objective.

We need an AI strategy, so we need AI projects.

We need AI projects, so we need to identify AI use cases.

Success can then become evidence that the organisation is doing enough AI.

That is almost the reverse of the interesting strategic question.

The interesting thing about AI isn’t that we can now do AI. It’s that the boundaries around what an organisation can do have moved.

Some things that were previously too expensive may now be affordable.

Some things that took months may take days.

Some services that could only provide standardised interactions may become capable of responding much more specifically to individual circumstances.

Some software that could never justify a conventional development team may now be worth building.

Some analysis that was impractical at scale may become routine.

Some experiments that required a business case before anybody could learn whether the idea worked may now be cheap enough to try first.

Those are strategic changes because they alter the feasible set of choices available to the organisation.

The technology matters because the constraints moved.

That suggests a different starting question.

What would we do differently if this constraint were no longer true?

If software were much cheaper to create, what would we build?

If analysing unstructured information were dramatically easier, which decisions or services would change?

If individual teams could test ideas without waiting for a large delivery programme, what would we want them to learn?

If routine execution required less human effort, where would we deliberately put the human judgement that became available?

The answers may involve AI.

They may also involve changing a process, redesigning a service, stopping something, reorganising work or deciding that a newly possible thing is still not worth doing.

This is why AI strategy and organisational strategy should not drift too far apart.

A specialist capability may be necessary to understand models, platforms, risks and emerging possibilities.

But the questions worth answering belong to the organisation.

What are we trying to achieve?

Which constraints have historically stopped us?

Which of those constraints are now moving?

What becomes possible as a result?

There is a connection to AI adoption here too.

If using AI becomes the objective, measuring AI usage makes sense.

If the objective is to exploit a changed boundary, the measure should be the thing that became better, faster, cheaper or newly possible.

AI is strategically important.

But perhaps the strongest evidence of a mature AI strategy is that the conversation eventually becomes less about AI and more about what the organisation can now do.