AI makes the distance shorter. It doesn’t make the decisions easier.
The biggest change AI has made to my work is not that code appears more quickly. It is that I can take a product decision and turn it into something real enough to test without losing weeks between the thinking and the build.
01
The handovers are getting shorter.
I can move from the proposition, user flow and operating logic into working software with far fewer handovers. That means I stay closer to the implementation, test assumptions earlier and remove ambiguity while the product is still cheap to change.
That is very different from writing a specification, passing it into a delivery process and waiting weeks to discover that three people interpreted it in three different ways.
02
Faster is not the same as better.
These tools will build the wrong thing just as enthusiastically as the right one. Someone still has to understand the customer problem, decide what evidence matters, shape the commercial model and notice when a plausible-looking feature is adding pointless complexity.
The real value comes from combining judgement with the ability to make a decision tangible quickly. AI can accelerate the build. It cannot tell you which business is worth building.
- Start from a real customer or operating problem.
- Build the smallest version that can answer the important question.
- Keep data, security and maintainability visible from the start.
- Use the working product to improve the decision, not merely demonstrate activity.
03
It changes what a small team can prove.
With SolarTyle, PikClub and Foremates, I have moved directly from product direction into working tools and customer experiences. It has meant proving much more of the proposition before putting a large delivery structure around it.
That is the opportunity I find interesting: a tighter loop between judgement, build and evidence. It lowers the cost of learning. But only if the team remains clear about what it is trying to learn in the first place.