For most of the history of software, building was expensive.
I don’t just mean financially expensive. Building software required time, specialized skills, coordination, planning, and significant engineering effort. Development capacity was a scarce resource, which meant that much of product management was focused on deciding where that limited capacity should be invested.
The emergence of AI-powered development tools is beginning to change that dynamic.
Today, it is possible to generate prototypes, build features, create documentation, write tests, and even develop entire applications in a fraction of the time that would have been required only a few years ago. The speed of software creation is increasing dramatically, and there are strong reasons to believe this trend will continue.
It is tempting to view this primarily as an engineering story. I suspect the more significant impact may be on product management.
When building was difficult, many organizations were constrained by execution capacity. There were always more ideas than available engineering resources. In that environment, any increase in productivity produced obvious benefits.
But as building becomes easier, a different challenge emerges: the number of things we can build far exceeds the number of things we should build.
The scarce resource is no longer implementation.
The scarce resource is judgment.
Over the years, I have seen products fail for reasons that had little to do with technology. Well-built solutions that addressed unimportant problems. Features that nobody used. Initiatives that consumed months of effort without creating meaningful value for customers or the business. In most of those cases, the issue was not execution quality. The issue was that the organization answered the wrong question.
Not “Can we build this?”
But “Should we build this at all?”
Artificial intelligence does not answer that question.
It can help develop a solution faster. It can generate alternatives. It can accelerate experimentation. It can dramatically reduce implementation costs. But someone still needs to understand users, identify meaningful problems, validate assumptions, and determine how a solution fits into a broader strategy.
In fact, the easier it becomes to build, the more important it becomes to have clarity about the problem being solved.
For years, product management has sometimes been viewed as a coordination function between business and technology. I believe that perspective was always incomplete, but it becomes even less accurate in a world where execution is increasingly abundant.
The real value of product management was never backlog administration or writing user stories. Its value lies in reducing uncertainty. In discovering which problems are worth solving. In understanding customers better than competitors. In making decisions about where limited resources should be focused to maximize impact.
None of that disappears with AI.
If anything, it becomes more important.
When an organization can build ten times faster, it can also waste resources ten times faster. Speed amplifies both good decisions and bad ones.
This is why I suspect some of the most important competitive advantages of the next decade will not come from access to the best AI models. They will come from understanding customers better, identifying more meaningful opportunities, and translating those opportunities into products that create real value.
The history of technology is full of examples where new tools reduced the cost of production. What rarely changes is the value of knowing what deserves to be produced.
As AI continues to make building easier, the discipline of product management may become more relevant than ever.
Not because organizations need more people managing roadmaps.
Because they need more people capable of deciding what is worth building in the first place.