Github CEO Thomas Dohmke has published a strong message to the global software engineering community. In a recent blog post entitled “Developers, Reinvented,” Dohmke stressed that the future of software development depends on developers who learn to work with AI rather than against it.
He shared insights drawn from detailed conversations with 22 software experts who have already integrated AI tools like GitHub Copilot into their everyday routines. Interestingly, many of these developers doubted the usefulness of AI. But over time, they realized that they relied on it to amplify their skills rather than replacing them.
According to Dohmke, these AI-savvy experts have moved from traditional programmers to what they call “code enablers” or “Creative Directors of Code.” Their roles have evolved from simply writing lines of code to setting goals, deriving structures, and meticulously reviewing AI-driven systems.
“The change is more about doing different jobs, not less work,” Dohmke said, adding that the scope of development has expanded significantly.
He emphasized that those who start recruiting AI early are likely to be competitive. Boring or massive tasks once thought to be overwhelming, such as building new features, such as comprehensive codebase rebuilding and tuning multiple AI agents, are now possible with the help of AI systems.
However, this shift is not without challenges. Dohmke warned that non-adapted developers might find themselves overtaken in a rapidly changing industry. He pointed out that some developers believe AI can generate up to 90% of all code within the next 2-5 years. In such a scenario, the most highly regarded skills go far beyond writing syntax. Instead, system design, understanding AI mechanics, managing intelligent workflows, and ensuring output quality will be the central stage.
This feeling is not isolated. Julia Liuson, senior leader at Github's parent company Microsoft, reflected a similar view, saying, “Using AI is no longer an option.” While some critics see such statements as pressure tactics, they reflect the broader industry consensus on the urgency to adapt to the reality driving this new AI.
Dohmke also acknowledged that not everyone is keen on this transformation. “For some people, the idea of managing AI systems instead of writing code personally can be less satisfying,” he admitted. However, he argued that abstraction is always part of the developer's work, whether using a high-level programming language or framework. Today, that abstraction simply exceeds one level, with AI agents working on the directions of developers.
In an optimistic note, he concluded that he frawned AI as a partner rather than as a threat. “Using AI as a collaborator allows developers to travel faster, aim higher, and tackle challenges that once seemed impossible.”
Dohmke's message is clear. The future of coding is not just about writing code. It's about telling your intelligent system to build:
