Bring the Tool to the Work
There is a visible gap between what AI tools can do and what most leaders are doing with them. The gap isn’t knowledge. Most leaders I know are very AI-curious. They read the posts, attend the sessions, listen to the experts, watch the demos. They know what the tools are and that others are doing great things with them.
Most people have read something about prompt templates, custom instructions, context files. Knowing what these are and knowing how to interact with AI to design and build them in a way that transforms how the tool works on your specific problems, are different capabilities entirely. Capability develops through doing. It always has.
The gap between knowledge and capability is compounded by pace. The tools are changing on a cycle measured in months. The “ten steps to better prompting” post from six months ago is already partially obsolete. LinkedIn is saturated with this content: build a context file, add skills, try this template. A lot of it is very competent. Almost all of it is functional description or a set of instructions to follow, rather than a demonstration of value. Functional descriptions of a rapidly changing tool age out fast. Capability is different. Someone who has developed the judgement to structure their interaction with AI, who understands the principles behind the current implementation, retains that capability even as the tools evolve underneath them. The principles are portable through time and can be built upon.
Over the past few months I have been working with several senior people on uplifting their AI capability. Different industries, different countries, different starting points, different problems to solve. One wanted to transform job searching into structured career support and a thinking partner. Another needed to move from basic conversational use to a genuinely capable, advisory working environment. A third had hit practical walls that were entirely solvable but invisible: burning through capacity, no awareness of session management and avoiding the tools that felt too technical. In each case the request was essentially the same: help me get real value out of this thing. All had read all the posts we have all read. All had commented “Guide” to get the latest playbook.
In each case the approach was the same. Start with the problem they already have. Build something that solves it, with them, using AI. Show them how solving one problem with AI builds the capability to solve the next. Load the context. Put the right structure in place. The output changes immediately and visibly. Every time. They didn’t need to be told this was better. They could see it working on their problem, in their language, with their constraints already understood. The next question is never “explain how that works.” The next question is “what else can we set up like that?” That is the moment capability starts to build. Not when someone explains the tool. When someone experiences the tool doing something that matters to them and can then imagine it doing something else.
That sequence is consistent across every engagement. The curiosity follows the experience. The learning follows the curiosity. The capability follows the learning. And because it was built on a real problem with real value, it holds. It is anchored in something the person actually uses, something that produces value, visible every day.
These are smart, capable, experienced people, who face practical barriers with AI: they have not seen what good looks like on their own work. No one has shown them the discipline layer that separates productive use from frustrating use and the available guidance describes the tools without demonstrating the value. Busy, senior people do not change how they work because of a LinkedIn post or a workshop demo. They change because they had a genuine moment of recognition on something that mattered to them. Real problems, real solutions, real learning. That is what sticks.
The leader who has stepped back from AI has not made a deliberate decision to opt out. They have made a series of reasonable micro-decisions: skipped the workshop because the diary was full, delegated the pilot because the team seemed keen, deferred the learning because nothing felt urgent. All of these ask them to add something to an overloaded calendar, without a clear use-case and obvious benefit. The approach that actually works does the opposite. It takes a problem already on their plate and shows them what AI looks like inside that problem. The tool comes to the work.
And when it does, something shifts beyond the individual. A leader who has built genuine capability, who has felt AI working on their own problems and developed a calibrated sense of what it does well and where it falls short, becomes a different kind of sponsor for AI across the organisation. They stop delegating from a distance and start directing from experience. That changes what gets funded, what gets prioritised and what the rest of the organisation believes is possible.
The leaders who develop genuine AI fluency are the ones who use AI on problems they already have. The problem is the curriculum. Building the solution is the course.