Most organisations know AI matters. Boards are asking about it. Teams are curious. The pressure to "do something with AI" is real.
But the responses tend to fall into one of two traps.
The moonshot trap
Some organisations swing for the fences. A big AI strategy. A bold vision deck. An enterprise platform procurement. Maybe a partnership with a major consultancy to "reimagine" operations.
These look great in board papers. They generate excitement in town halls. And they almost always collapse under their own weight.
Why? Because they skip the foundations. You cannot deploy a sophisticated AI capability into an organisation that does not yet understand how to prompt effectively, has no governance framework, and whose teams view AI as something IT does.
Big bets without foundations are not strategy. They are expensive hope.
The dabbling trap
Other organisations take the opposite approach. A pilot here. A prompt training session there. A few ChatGPT licenses nobody is really using. Someone in marketing built a GPT once and it was "pretty cool."
This feels safer. Lower risk. Let people experiment.
But experiments without direction do not add up to anything meaningful. There is no shared capability being built. No compounding value. Each initiative exists in isolation, and when the person who championed it moves on, the experiment dies with them.
Dabbling is not cautious. It is wasteful.
What actually works
The organisations getting real value from AI are doing something different. They are advancing technology and people together. Every step delivers something measurable and deliberately sets up what comes next.
This means:
- Starting with clarity, not technology. Understanding where AI can genuinely create value before buying tools or building anything.
- Building capability alongside solutions. The team that will use and extend the AI is involved from day one, not handed a finished product.
- Designing for compounding. The knowledge assistant you build today uses architecture that makes the voice agent possible tomorrow. Nothing is throwaway.
- Measuring ownership, not just deployment. The real success metric is whether the organisation can keep building without you.
The principle
Every step should deliver value. Every step should unlock what comes next.
This is not a framework for the sake of frameworks. It is the pattern we see in every successful AI transformation we have been part of. The travel company whose Guest Operations team now handles enquiries in minutes, not hours, and is extending the capability independently. The property portfolio that reclaimed thousands of hours and delivered millions in benefits across the entire organisation.
Neither started with a moonshot. Neither dabbled. Both advanced deliberately, building capability that compounded.
If your AI strategy feels like it is stalling, the answer is probably not a bigger bet or another pilot. It is a clearer path where every step counts.