Commentary: Developers, wake up – or lose your job

Contents

Just a few years ago, AI-powered programming tools were considered a nice gimmick. That has fundamentally changed. Copilot, Claude, Codex, and the like now write code that works – not always perfectly, but usually quite usable. They generate unit tests, create documentation, and suggest refactorings. Development is accelerating rapidly. And anyone who believes they are not affected is mistaken.

Golo Roden english

Golo Roden is the founder and CTO of the native web GmbH. He works on the design and development of web and cloud applications and APIs, with a focus on event-driven and service-based distributed architectures. His guiding principle is that software development is not an end in itself, but must always follow an underlying technical expertise.

What AI can do – and what it can't

Important note: AI systems operate at the level of syntax and patterns. They recognize recurring structures and reproduce them with impressive speed. This is often sufficient for standard tasks, for example, to set up a REST API, write a database connection, or validate a form.

However, AI does not understand why something is the way it is. It does not know about trade-offs between performance and maintainability. It doesn't know which decisions will become costly in six months. It has no feel for how an architecture fits the organization. And it doesn't know the unwritten rules of the business domain, the logic that is not in any requirements specification but decides the success or failure of a project.

The uncomfortable truth

This means that value creation in software development is shifting away from pure implementation towards conception. This is not a grim prophecy, but already a reality. Those who today primarily churn out code without a deeper understanding of the overall system are making themselves replaceable in the medium term.

But what remains is architecture. However, architecture in this context does not mean drawing UML diagrams or similar. Architecture means making the decisions that are difficult to change later: How do we slice our services? What dependencies do we enter into? Where do we draw the boundaries between modules? These questions require contextual knowledge, experience, and judgment. These are all things that cannot be derived from training data.

Act now

And this means: The time to deal with architecture is not „sometime when I have time.“ It is now. Those who wait until AI tools become even better are waiting too long.

Specifically, this means: develop abstraction skills, practice systems thinking, build domain knowledge instead of just processing tickets. Developers must learn why one decision was made, not just what the code looks like.

Developers who possess these skills will become more valuable, not less. They will become those who can effectively use AI tools and critically evaluate their results. They will be the architects who design systems while pure coding is increasingly automated.

In other words, if you want to keep your job as a developer, you must engage with architecture now. A good place for this is the Software Architecture Conference, OOP 2026. It teaches how developers can strategically design software projects instead of just coding.

(mki)