Cheap Code, Expensive Confusion

AI has made it absurdly easy to generate software-like artifacts. But, that is not the same as generating value.

Daily we are flooded with shallow POCs, demos and agent integrations that solve little, but create the appearance of innovation. The barrier to producing something that looks like software has collapsed. The barrier to sound judgment not. In fact, judgment matters more than ever because the lever becomes more powerful by the day.

Democratized software development can easily become a distraction. It can attract people to spend time building unnecessary solutions outside their core responsibilities. Instead of thinking critically about whether software is needed at all, whether a problem already has a solution, or whether the work creates measurable customer value.

Much of this activity is internally focused and feeds bureaucratic self-justification rather than external value creation. Visibility is often more rewarding than steady operational excellence. Flashy AI initiatives, crisis management, and demonstrative activity can advance careers. Even when they create little value or even contribute to the underlying mess.

That is a real problem for management. Discerning value from noise. Deciding what is worth building and how to focus company attention is becoming much harder, because common empowerment does hold opportunities, of course.

Yet many organizations are doing the opposite of what is a proven recipe against uncertainty. Instead of falling back on first principles, they surrender to hype and FOMO. Everything has to be “AI-enabled,” whether it creates customer value or not.

The result is predictable: wasted time and money, duplication, security risk, and rising cognitive debt. People build things outside their competence, ignore existing solutions, and feed bureaucracy rather than customers.

Meanwhile, professionals are told that software has become easy because someone vibecoded a prototype in an afternoon. That is nonsense. The hard part was never typing code. The hard part is understanding complexity, context, trade-offs, integration, maintenance, and risk.

Giving everyone an LLM no more replaces engineers, than giving everyone a saw replaces carpenters.

The managerial answer should not be blanket democratization. It is disciplined focus. Focused deliberate experiments close to your value creation, strong ownership, and relentless attention to actual value.

The scarce resource is the ability to say: This is worth building, and that is not.

But of course, this requires the capability to assess and courage to make calls.

Good managers know, where value creation happens: Where product meets customer.