I've spent most of my career helping organisations create value with technology. Start-ups. FTSE businesses. PE-backed companies. Large transformations. Product businesses. M&A. Across all of it, technology has rarely been the limiting factor.
Time is. Investment is. Complexity is. The organisation's appetite for change and innovation is. And above all, the cost of finding out whether an idea works.
Which is why the last twelve months have caught me off guard. I've been around technology long enough to have lived through most of the major hype cycles of the past twenty years, and very little surprises me anymore. This has.
Most ideas died in PowerPoint, because proving them cost too much, whether or not they were any good.
The surprise is not that AI can write emails or generate images. It is what AI has done to the economics of building: products, businesses, and solutions to problems that have sat untouched in backlogs for years.
The friction has disappeared
Historically, taking an idea from concept to working reality was expensive. You needed product managers, architects, engineers, infrastructure, test environments, security reviews, budget approval, and months of coordination before anyone could tell you whether the idea was any good.
Most ideas died in PowerPoint, because proving them cost too much, whether or not they were any good.
That equation has collapsed. A single person with domain expertise and product thinking can now take an idea to a working prototype in days. The thing itself, rather than a slide describing it: workflows, internal tools, automations, entire applications.
I know because I've been doing it. Over the past year I've spent evenings and weekends building with these tools: working applications, automations and products that, not long ago, would have required funding approval, specialist teams and months of delivery effort. Some of it worked. Some of it broke in instructive ways. All of it changed how I think.
When the cost of experimentation collapses, thousands of problems that were previously "too small to solve" become economically viable.
The opportunity is leverage
Many people are still looking at AI through a technology lens. I increasingly see an economic one: leverage, productivity, business model.
When the cost of building falls this far, you do more than the same things more cheaply. You do things that were never worth doing before: more experiments, more products, more swings at more problems.
The organisations that internalise this first will operate at a pace that will look completely unreasonable to their competitors.
Expertise matters more, not less
A misconception is taking hold that because AI can generate code, technology expertise matters less. I think the opposite.
The internet is already filling up with AI-generated applications that look impressive and are badly built. Poor security. Weak architecture. No governance, no resilience, no operational thinking, no understanding of enterprise reality. AI can produce software extraordinarily quickly. It can produce technical debt even faster.
The differentiator is judgement: understanding architecture, risk, scale and operating models. Knowing what good looks like. Knowing when to build, when to buy, and when to partner. AI has made those instincts more valuable, because they are now the scarce ingredient.
The constraint is moving
Throughout my career I've found myself obsessing over the same questions when attempting to innovate.
What's the job to be done? Where's the pain and gain? What happens immediately before this problem, and immediately after it? Why does this exist at all? Keep asking why.
Those questions matter more now than they ever have, because implementation is getting easier by the month while knowing where the value sits remains as hard as it's always been.
The constraint is shifting away from technology and towards imagination, judgement and execution.
Which is why I think every senior leader should be building
They do not need to become engineers, and their organisation's next product should probably not be built this way. But you cannot understand what is happening from a strategy paper, a vendor demo or a conference keynote.
You have to experience how fast an idea can move, see where it breaks, where the risks sit, and what these tools still do badly. That is how you develop your own sense of what is possible, and that sense is what changes how you look at your business.
Once you've built something, anything at all, you start seeing differently: processes that shouldn't exist, products that are suddenly viable, business models that can be rebuilt. And you start questioning assumptions about how digital products get built that have held for twenty years and may no longer be true.
And none of this means building everything yourself. Often the smartest move is still to buy, or to partner with people who do it brilliantly. But you make that call far better once you've built a few things and felt where these tools are strong and where they fall apart. That kind of judgement is earned, not read.
Final thought
I don't think AI is replacing product leaders, architects or technology executives. I think it is making the best ones far more effective.
Technology has always been an input. Value creation has always been the outcome. AI has rewritten the economics of getting from one to the other, and I suspect we are still at the beginning.
So if you're a senior leader and you haven't yet built anything with these tools, pick a problem that annoys you and give it a weekend. The output matters less than what it does to how you see.
In the spirit of the piece, a disclosure: AI helped me write this. The thinking is mine.