Picture a board meeting. The Chief Digital Officer presents a slide with twelve AI pilots. There's a chatbot for customer service, a document classifier for legal, a demand forecasting model for supply chain, a handful of generative AI experiments. The board nods. Progress is being made.
Six months later, two of those pilots delivered marginal improvements. Eight were quietly shelved. Two are "still in progress," which in most organisations means nobody wants to admit they've stalled. The total investment: about £1.2 million. The strategic impact: unclear.
Nobody in that board meeting asked the hard question: "What is all of this for?"
Projects Are Not a Strategy
This is the most common failure mode I see in AI adoption. Organisations confuse activity with direction. They have AI projects. They don't have an AI strategy.
The distinction matters. An AI strategy answers a specific question: what new value will AI create for our customers and our business over the next three to five years? AI projects answer a different question: what can we automate next?
One is about direction. The other is about tasks.
A collection of AI experiments is not a strategy. It's a to-do list with a technology label.
The pilot graveyard that plagues so many organisations is a symptom, not the disease. The disease is the absence of strategic intent. When every team picks its own AI project based on local needs, you get a portfolio that looks busy but goes nowhere. No compounding. No shared learning. No path from efficiency to value creation.
I wrote about the AI Value Gap recently: the distance between where organisations stop with AI (automation and cost savings) and where AI could actually take them (new products, new revenue, new competitive positions). The pilot-graveyard pattern is one of the main reasons that gap exists.
What an AI Strategy Actually Looks Like
A genuine AI strategy starts with business outcomes, not technology capabilities. It doesn't begin with "we should use large language models" or "let's build a data lake." It begins with three questions:
1. Where does AI shift our competitive position?
Not "where can AI save us money" but "where does AI change the game in our market?" This forces you to think about customers, competitors, and value creation rather than internal processes.
2. What capabilities do we need to build, not buy, to get there?
Every organisation can buy AI tools. The strategic question is which capabilities become core to your differentiation. Those you build internally. Everything else you partner or procure.
3. How do we sequence this so each step funds the next?
This is what I call the funding ladder. Quick wins in automation generate cost savings. Those savings fund the budget and credibility for augmentation plays. Augmentation successes unlock the political capital and organisational confidence for genuine value creation. Each stage finances the next.
The funding ladder is critical because it solves the board's favourite objection: "Show me the ROI." You don't need to justify the full transformation upfront. You need to show that step one pays for step two, and step two pays for step three. The early wins aren't the destination. They're the funding mechanism.
The Capability Gap Nobody Talks About
Most organisations have plenty of technical AI capability. They can build models, deploy APIs, fine-tune large language models. The data science team is talented. The engineering infrastructure is solid.
What's missing is strategic AI capability. The ability to look at AI's possibilities and translate them into business strategy. To answer the question: "Given what AI can now do, what should our business become?"
This is not the CTO's job. The CTO ensures the technology works. It's not the Chief Data Officer's job. The CDO ensures the data is governed and accessible. It's not even the Chief Digital Officer's job, though it often gets dumped there by default.
It's a strategic leadership function. Someone needs to sit at the intersection of business strategy and AI possibility and connect the two. In some organisations that's the CEO. In others it's an emerging role. But in far too many, nobody owns it. And that's why the projects multiply while the strategy stays blank.
I work with leadership teams where the technical capability is genuinely impressive. They can build almost anything. The problem is nobody has decided what's worth building. The question "can we?" has been answered. The question "should we, and in what order?" remains open.
What This Means for You
Try this exercise. Pull up your current AI project portfolio and sort every initiative into one of two columns:
Column A: Connected to a clear strategic outcome (new revenue, new market, competitive repositioning, customer value creation).
Column B: Interesting experiment, useful efficiency gain, or "someone thought this was a good idea."
Be ruthless. If the connection to strategic value requires more than one sentence to explain, it belongs in Column B.
If more than half your portfolio sits in Column B, you have projects. You don't have a strategy.
The fix is not more AI. It's not a bigger budget or a fancier platform. The fix is better questions about AI. Specifically:
If that last question doesn't have a clear answer, start there. Everything else follows from it.
Going Deeper
This is exactly the gap the AI Value Institute exists to close. If you haven't already, take the AI Value Gap Assessment to see where your organisation sits across the three stages of AI adoption. It's a five-minute benchmark that cuts through the noise.
My upcoming masterclass on 2 May covers this in depth: how to build a genuine AI strategy, sequence your investments, and move from projects to strategic transformation. If you're a business leader who knows your organisation is doing AI but isn't sure it's doing AI strategically, that session is built for you.
And every week in the AI Value Institute newsletter, I share practical thinking on exactly these themes. No vendor pitches, no hype cycles. Just honest strategy for leaders navigating AI.