I work with a lot of organisations on cloud and technology strategy, and there’s a phrase I’ve been hearing with increasing regularity over the last three years. It comes from the boardroom, from the CEO, occasionally from a CFO who’s just come back from a conference looking slightly panicked…
“We need an AI strategy.”
It sounds urgent, and of course it is urgent in a way, but I’ve come to realise it almost always means two completely different things depending on who’s saying it and why. Getting your head around that gap is probably the most useful thing you can do when it lands on your desk.
When a board or CEO says “we need an AI strategy,” they’re almost never asking for a technical roadmap. What they’re really after is reassurance; that the organisation isn’t falling behind, that investors won’t ask an awkward question at the next AGM and get a blank stare in return, that somebody somewhere in the building has thought about this. They want something credible and coherent, ideally fitting on a slide that nobody has to squint at.
When an IT leader hears it, though, they hear something quite different. They hear “investment in data infrastructure.”, and they’re absolutely right to, because without a solid data foundation, you’re building your AI solution on quicksand.
I’ve made a silly analogy of this a few times over the years, something along the lines of how you can’t build anything massive on sand, apart from maybe the pyramids. Then I was in Egypt with my family last year and discovered something I never knew before… even the pyramid builders knew you can’t just stack stones on sand! The pyramids actually have foundations of bedrock and mahoosive limestones buried up to 6-8 metres deep! I’d been proving my own point for ages and I hadn’t even realised…
Two documents are better than one…
If you’re the IT leader with this request, my suggestion is that you probably need two documents rather than one, because the audience for each is completely different and trying to serve both at once almost never works.
The first is a exec board narrative. Short, strategic, and confident, focused on [most importantly] business outcomes, then perhaps competitive positioning, and risk. This is the information the CEO walks into an investor meeting with, and it needs to tell a story about where the organisation is heading without getting into how the plumbing works. Nobody on the board wants to hear about the optimised data lake architecture, and frankly they shouldn’t have to.
The second is the working strategy. This is the real one, with the investments, the data programme, the build-vs-buy decisions, the governance model, and probably a spreadsheet or twelve attached. It’s the document your teams will actually deliver against, and it has to be brutally honest about what state your data is in right now and what needs to happen before any of the exciting AI stuff becomes real.
Mashing these two together won’t work. Either it ends up so vague, it can’t actually drive decisions, or the working strategy gets presented to the board and you watch the room quietly die somewhere around slide seventy-eight [we’ve all been there!]. The CEO needs to calm investors, the IT team needs to know what they’re actually building, and one strategy document genuinely can’t do both.
Start With the Bedrock
Before writing any document, the most valuable thing you can do is take an honest look at your data. Not a new procurement exercise, not an AI pilot, just a straightforward answer to whether you could actually build something real on top of what you’ve got today.
In most organisations, the answer is “it’s complicated,” and that’s absolutely fine as a starting point, indeed nobody expects perfection [or indeed the Spanish inquisition!]. The outcome you really want to avoid is cracking on with the strategy without asking the question, then discovering six months later that the foundations aren’t there and having to retrofit them while everything else is already going up around you.
The ancient Egyptians figured that bit out about four and a half thousand years ago. There’s probably a lesson in there somewhere… 🔺
RSS – Posts