AI adoption starts with better questions, not better prompts
For all the excitement around artificial intelligence, I believe the most useful place to start is not with the technology. It is with a better question.
Over the past few years, I have worked with founders, SMEs, leadership teams and public-sector partners who want to understand what AI could mean for them. Some arrive with a clear ambition; others simply feel they may be falling behind. Most have seen the headlines, tried a tool or two, and know that something significant is changing. But they are often unsure where to begin.
My experience is that organisations make the most progress when they resist the urge to start with, “Where can we use AI?” Instead, they ask: “What is taking too much time?” “Where are our customers or colleagues experiencing friction?” “Which decisions are difficult because we do not have the right information?” “What could we do better if we gave people more capacity to use their judgement, creativity and expertise?”
Those questions turn AI from an abstract opportunity into a practical conversation about work, people and outcomes.
I co-designed and delivered AI Empower, a pilot funded by the Department for Science, Innovation and Technology that helped SMEs explore how AI could support productivity and growth. The most valuable part was not demonstrating a particular platform. It was creating space for businesses to surface the real problems underneath the request for technology: repetitive processes, inaccessible knowledge, overloaded teams, inconsistent customer journeys or a lack of time to focus on higher-value work.
That distinction matters. A tool may be impressive, but if it is applied to the wrong problem, without a clear owner or measure of success, it can quickly become another unused subscription. Conversely, even a modest use case can create meaningful value when it is rooted in a genuine operational challenge and introduced thoughtfully.
I have come to see responsible AI adoption as a leadership and change challenge as much as a technical one. Leaders need to be clear about the purpose, the boundaries and the human role. Teams need permission to experiment, but also confidence that they can raise concerns and ask difficult questions. And organisations need to consider whose experience is represented in the data, who could be excluded by a new process, and where human judgement must remain firmly in the loop.
This is particularly important for smaller organisations. They may not have large innovation teams, dedicated data specialists or lengthy procurement processes. But they often have an advantage: they know their customers closely, can make decisions quickly and can test practical changes without excessive complexity. The key is to start small enough to learn, but with enough intent to measure what changes.
A simple approach I use is to identify one priority problem, map the current process, involve the people who do the work, and agree what a better outcome would look like. That might mean reducing time spent on admin, improving the quality of an early draft, helping a team find reliable information more quickly, or making a service more accessible. Only then is it time to consider whether AI is the right response and what safeguards are needed.
The organisations I most admire are not treating AI as a replacement for people or as a shortcut around strategy. They are using it to augment capability: helping people spend less time on routine tasks and more time solving problems, building relationships and making better decisions.
There is no single blueprint for adopting AI well. The technology will continue to develop at speed, but the fundamentals are stable: start with a real problem, bring people into the process, test responsibly, measure the outcome and keep learning.
Better questions will always be more valuable than a rush to have the newest answer.
Read More Productivity & Efficiency
