AI is making it cheaper to build — so why are teams still building the wrong things?

The cost of building software is collapsing. What once required a team of engineers for six months can now be prototyped in days. GitHub Copilot, Claude, and a dozen other tools have fundamentally changed the economics of software development.

And yet, most organisations are still building the wrong things.

This is the central paradox of the AI era: the technology that should make organisations more effective is, for many of them, simply accelerating their existing dysfunction. They are moving faster — but in the wrong direction.

The real constraint was never building

For most of the history of software, the binding constraint was execution. Good ideas died because teams were too slow, too expensive, or too distracted to build them. The backlog was long, the team was small, and the question was always: which of these important things do we do next?

AI is removing that constraint. But it is revealing a deeper problem that was always there: most organisations do not have a reliable process for figuring out which problems are actually worth solving.

Discovery is now the competitive advantage

When building is cheap, the value shifts upstream. The organisations that win are not those with the fastest engineers — they are those with the clearest thinking about which problems to solve.

This means:

• Investing in customer research and continuous discovery
• Getting better at problem framing (not just solution definition)
• Building leadership capability to make difficult prioritisation decisions
• Resisting the pressure to build just because building is now easy

What this looks like in practice

We work with organisations who are making this mistake right now. They have adopted AI tools, their velocity has increased, and their backlog is getting shorter. But customer satisfaction is flat. NPS is not moving. Revenue is not growing.

When we dig in, the pattern is almost always the same: they are building features that customers do not care about, faster than before.

The answer is not to slow down. It is to invest at least as much in discovery as in delivery. To ask “should we build this?” before “how do we build this?” Every sprint.

The organisations that will win

The organisations that will win in the AI era are those that combine fast, cheap execution with rigorous, evidence-based discovery. They use AI to prototype and validate quickly — not to ship faster without thinking.

The question is not “what can we build?” It is “what is worth building?”

That is a harder question. It always was. AI has not made it easier — it has just made the cost of getting it wrong much higher.