The end of the classic developer or a new evolution?

Until recently, we were debating whether AI would one day help programmers. Today, we are no longer talking about assistance, but about AI independently writing production-ready code, running applications, testing systems, and fixing bugs at a speed that would have been unimaginable just two years ago. The changes are not happening slowly and gradually, but almost explosively, right before our eyes, month after month.

In that context, I received an email from a programmer colleague. Without much introduction, he advised me to install an AI coding assistant in my editor and subscribe to at least one of them. But the point of his message was not the tool itself. The point was much deeper — the way we develop software today has already changed radically.

What truly surprised me, however, was not the recommendation itself, but the sentence that followed — he says he almost no longer programs.
 

From developer to orchestrator

Today, as he claims, he mostly assigns tasks and defines what needs to be done, while AI writes the code, tests it, runs commands, fixes errors, and suggests optimizations. His role increasingly resembles that of a project manager communicating with an extremely capable, tireless developer who never sleeps.

According to his experience, Claude is very strong when working with concrete commands and analyzing terminal output, while ChatGPT offers more room for continuous work without interruptions. He uses both tools in parallel and says that the monthly subscription pays for itself many times over through time savings. When someone who has been writing code professionally for years says this, it is hard to ignore.
 

Descriptive programming as a new paradigm

The concept of descriptive programming is being mentioned more and more often. Instead of typing line by line for hours, we now describe what we want to achieve. We define the context, business logic, and constraints, and the model generates the implementation. Our role is shifting from micromanaging syntax to making architectural decisions and ensuring quality.

The programmer thus becomes someone who asks the right questions and gives precise instructions, rather than someone who mechanically writes loops, validations, and basic CRUD operations.
 

What does this mean for the market and web agencies?

The changes are already visible. Platforms like Shopify are integrating AI assistants into their systems to allow users to customize themes and functionality more easily. If a similar level of automation is deeply integrated into other web platforms, a large portion of simple web projects could become almost fully automated.

For many web agencies, this would mean a serious shift in the business model. Projects that once took weeks of work could be completed in hours. The focus would need to move from technical implementation toward strategy, design, branding, and business consulting.
 

The limits of artificial intelligence

Still, it is important to remain realistic. AI is not aware of business context in the way humans are. It can make mistakes, generate insecure code, and sometimes convincingly offer solutions that are not actually optimal. There is still a need for someone who understands the bigger picture, knows how to assess risk, and takes responsibility for the final product.

AI is an extremely powerful tool, but it is still just a tool.
 

Evolution, not disappearance

Perhaps the truth lies somewhere between euphoria and fear. Programming as we know it will probably not disappear completely, but it will transform. Those who adapt and learn to use AI tools effectively will have a huge advantage. Those who ignore them could quickly fall behind.

Maybe in a year or two we will no longer even talk about “using AI,” because it will be a standard part of every development environment. Until then, it seems wise to do what my colleague suggested — seize this moment, experiment, build personal projects, and learn how to communicate with models that are becoming more capable every day. Because if one thing is certain, it is that the way we create software is changing irreversibly.