Your AI Strategy Has a Shelf Life

I remember when companies used to write five-year digital strategies.

Beautiful things.

Eighty slides. Competitive landscape. Transformation roadmap. Technology architecture. Three horizons. Perhaps a messaging house somewhere.

Then everyone went away feeling very strategic.

The problem was that by year two, half the assumptions had changed.

AI has taken that problem and put it on steroids.

Stanford's AI Index found that the cost of querying a model performing around GPT-3.5 level on MMLU fell more than 280-fold between November 2022 and October 2024. At the same time, increasingly capable models have become dramatically more accessible. (Stanford HAI)

That's the ground moving underneath your PowerPoint.

Stop treating AI strategy like a destination

I see organisations asking which AI platform they should standardise on, which model to use and which tools employees should access.

Reasonable questions.

But they're often being treated as strategic decisions when many of them are increasingly temporary ones.

Today's best model may not be next quarter's. A workflow that isn't economically viable today might become trivial when inference costs collapse.

Check it. 👇

Your AI strategy probably shouldn't tell you exactly what your organisation will be using in three years.

It should help you decide what to do when whatever you're using today inevitably changes.

Strategy needs a faster heartbeat

The OECD's 2026 Digital Government Outlook makes a useful point: organisations need governance, investment and skills that fit the iterative nature of digital technologies, with staged funding, learning, evaluation and the ability to adjust as they go. (OECD)

You cannot remove uncertainty by writing a longer strategy.

You design for it.

That means knowing which business problems matter before choosing tools. Understanding which workflows are worth redesigning. Having principles for what AI can and cannot do.

And deciding when you will reconsider your decisions.

The most dangerous AI strategy may not be having no strategy. It may be having one that nobody remembers to question.

So perhaps every major AI decision should come with another field beside owner, budget and deadline:

Review date.

Six months from now, is this still the right model? Is this workflow still worth automating? Has the cost changed? Has the risk changed?

I think the sequence is actually quite simple.

First, understand where you are.

Then build the AI strategy.

Then keep challenging it.

That's one of the reasons I built my AI Communications Strategy Scorecard.

It gives communications and marketing teams a starting point: an assessment of where they currently stand across strategy, workflows, people, governance and measurement.

From there, you can build an AI strategy around the organisation you actually have, rather than the technology everyone happens to be talking about this month.

And then comes the bit that's easy to forget.

Review it.

Because AI readiness isn't something you achieve once and stick on the wall.

The technology will change. Your workflows will change. Your people will learn. New risks will appear.

So the strategy needs to move too.

If you don't know where to start, take the Scorecard. It's free and takes about 12 minutes.

We spent years telling organisations they needed to move faster.

AI has created a slightly different challenge.

Now they need to become better at changing their minds.