How AI Is Changing Software Development Part 4 -
The CTO´s Perspective

Why the Best Developer of the Future Might Be the One Who Produces the Least Code

For many years, a developer’s performance was relatively easy to assess: Good developers delivered features reliably, solved complex problems, and were able to quickly familiarize themselves with new technologies. Code quality naturally played a role as well. Nevertheless, productivity was often measured by how much a developer could implement themselves. This measure, however, is likely to change radically.

With Agentic Coding, pure implementation is becoming less important: In the future, a significant share of code will no longer be written directly by developers, but generated by agents. As a result, the question of where the real value of an experienced engineer lies is shifting. It will no longer primarily be about generating the best code, but about making the right decisions – both when it comes to requirements and when it comes to code review.

The easier implementation becomes, the more valuable all the skills that cannot be automated will become. Architectural decisions are just as important as product understanding, security, maintainability, and the ability to explain complex relationships clearly. A Senior Engineer will spend less time writing code themselves in the future. Instead, they will invest more time in reviews, technical guardrails, coaching, and evaluating different approaches. Their role will increasingly be to ensure that many locally good decisions ultimately result in a product that delivers value to the customer, as already described in Part 3 of the series *How AI Is Changing Software Development*.

This development will also have consequences for the career paths of many developers. Until now, technical excellence has been considered the most important prerequisite for professional growth. In the future, that alone will probably no longer be enough. Anyone who wants to take on greater responsibility will increasingly need to understand broader interdependencies, set priorities, and evaluate technical decisions in the context of the overall product. The ability to write good code will remain important. The ability to determine which code should be written in the first place, however, will become increasingly important.

For organizations, this development also has consequences: Many recruiting processes have traditionally focused heavily on algorithms, programming languages, or frameworks. At the same time, other skills are becoming increasingly important: Systems thinking, communication, architectural understanding, and sound judgment will increasingly determine the contribution a developer can make to a product organization.

From my perspective, Agentic Coding therefore changes not only the work of our developers, but also our understanding of what makes a good developer in the first place. Perhaps in a few years, we will find that the engineers who made the greatest contribution were precisely those who wrote the least code themselves. Their most important task was to turn many good individual decisions into a successful overall product.

To conclude the series *How AI Is Changing Software Development*, I would like to provide a brief summary: The four parts of this series have shown from different perspectives that Agentic Coding changes far more than just the way code is created. The daily work of developers is changing because they are increasingly becoming orchestrators of individual work steps. Organizations are changing because the bottleneck is shifting from implementation toward requirements, quality assurance, and decision-making. Products are changing because additional development capacity can quickly lead to greater complexity rather than greater customer value without a clear strategy. And finally, the definition of a good developer is changing because judgment, systems thinking, and the ability to make the right decisions are becoming increasingly important.

As CPTO at eperi, I see this less as a technological challenge than as an organizational transformation. Agents and Spec-Driven Development will further increase our development capabilities. The decisive factor, therefore, will be whether we as an organization are able to use this additional speed effectively. In my view, the successful software companies of the future will not be those that generate the most code, but those that use their additional capacity to make better decisions, keep their products understandable, and create lasting value for their customers and employees.