(Photo by Dariusz Sankowski on Unsplash)

In a recent post, I shared ideas to smooth a company's AI transformation. One of those ideas was:

Develop a real AI strategy (with real use cases)

Longtime readers will note that this is nothing new; I say it quite a bit. But then Kelwin Fernandes dropped this question in a comment:

What does 'a real AI strategy' look like in your experience when you strip away the buzzwords?

I started to type out an answer, then realized I could share an extended version with everyone here. So that's what this piece is about.

What is an AI strategy?

The term "AI strategy" means different things to different people. To understand my definition, we can start with the notion of strategy as a practice, and then see it as a noun.

Based on Michael Porter's seminal 1996 HBR piece "What Is Strategy?", strategy-the-practice is about deciding what to build or what service to provide. It's the company's direction.

From there, I say that strategy-the-noun is an artifact that comes out of strategy-the-practice. The leadership team has made their decision on where to go, so they write it all down such that everyone has a concrete reference.

AI strategy(-the-noun), then, is a strategy specific to the company's AI efforts. A company goes through the act of defining a strategic direction for how they'll use AI – why they'll use it, what they'll use it for, what boundaries determine where they will not use it, what specific use cases are on tap – and then they write it down.

What is it not?

Based on that, you can see that an AI strategy is not just a mission statement or a decree to use AI. Nor does it mean handing everyone keys to Claude or Gemini and saying "have at it." Those acts may serve as motivation (convincing team members to use AI), but they don't set direction (where to go).

In fact, when a company issues a blanket AI usage mandate but does nothing else to guide it, they're encouraging people to wander aimlessly in the hopes that they stumble onto some meaningful use cases. It's the anti-strategy. And it gets anti-strategy results: wasted time, wasted money, and damaged team morale.

But how does it look?

We've finally made it back to Kelwin Fernandes's question about what a "real AI strategy" looks like.

My answer is: The format or appearance doesn't matter. What matters is what's in the document.

If you squint just right, you'll see that "AI strategy" is a fancy term for "a plan." That's exactly what it is. Like any good plan, it helps you understand where you're going and shows you when you're veering off-course.

So long as that plan maps out specific details on what to do and why to do it (going back to Michael Porter's definition of strategy) then that's what counts.

Can your AI strategy have fancy formatting? Sure. Can it also be an informal document? Absolutely. If team members can review it to understand the what and the why behind this company's use of AI, you're in good shape.

I'll close out by noting three key points:

1/ An AI strategy, like any other strategy, is not set in stone. It's a forward-looking view of the world, and as such, it requires review and adjustment as the future becomes the present.

2/ While the AI strategy document holds value, the greater value is in the exercise of developing it. That brings key players to the same understanding and allows everyone to hold the business to account.

3/ For more details on AI strategy, I'll point you to my other write-ups:

• "What is a data strategy, and why do I need one?" – This is my original blog post on data strategy from 2016. (Yes, 2016. Ten years ago.) Swap "data" for "AI" and it still holds.
• "The value of plans and planning" – This explains the realities of an AI strategy changing over time.

And if you need guidance on your AI transformation, reach out. I take a practical, risk-focused approach to defining strategy and surfacing use cases. I outline my approach in my latest book, Twin Wolves: Balancing risk and reward to make the most of AI.