Defining the New AI–Engineer Relationship
The debate over whether AI will replace software engineers centers on how tools like Claude AI coding tools automate repetitive programming tasks while expanding what small teams and individuals can build. This debate asks if AI job displacement tech will erase traditional software engineering jobs or change their shape, shifting from manual coding to higher-level problem framing, product thinking, and orchestration of AI agents. Anthropic executive Boris Cherny, creator and head of Claude Code, uses his own work as an example of this shift. He says coding is effectively “solved” for the kind of work he does and that he has not written a line of code in more than six months. For him, the question is less whether AI will replace software engineers and more which parts of their current job description will survive.
Boris Cherny’s Warning: The Title ‘Software Engineer’ May Vanish
Cherny offers a stark view on the future label and scope of software engineering jobs as AI spreads through development workflows. He predicts that the title “software engineer” could start to disappear, replaced by broader roles closer to “builder,” where designers, product managers, and leaders ship code with the help of agents. As head of one of the fastest-growing AI coding tools, he argues that coding is getting solved for more of the code we write, and he is actively automating his own role. This raises a concrete risk for entry-level engineers whose work focuses on boilerplate coding: AI replace software engineers in those narrow tasks first. Yet even in that warning, Cherny frames the change as a shift in responsibilities and titles rather than the end of human involvement in software creation.
A Golden Age for Founders, Not the End of Coding
Alongside his caution, Cherny describes what he calls a golden age for young computer science graduates willing to found startups. He tells new graduates that entry-level roles still exist, but if they are entrepreneurial, “go start a startup.” With Claude Code and similar AI coding agents, small teams can act like far larger engineering organizations, shrinking the gap between an idea and a working product. In a recent talk with Y Combinator founders, Cherny asked how many let Claude Code write 100% of their code; about half the hands went up, while only one founder said they do not use the model at all. This pattern suggests not that coding disappears, but that AI shifts who can build software and how fast they can move, opening a wider door for nontraditional founders.
Which Software Engineering Tasks Will AI Replace?
The divide in opinion about software engineering jobs AI centers on which tasks AI will handle entirely and which it will only augment. Cherny points to coding as an area where AI agents now cover a large share of production work, especially for repetitive or well-defined problems. That threatens some entry-level positions where value comes from typing out routine code rather than understanding users or systems. At the same time, the rise of AI job displacement tech creates demand for people who can specify requirements, design architectures, and judge whether agent-produced code is correct. Claude AI coding tools move power toward those who can frame problems clearly, wire together tools and APIs, and continuously refine prompts. Specialized engineers may see their roles split: some low-level tasks automated, while new work emerges in oversight, agent orchestration, and safety-critical review.
From ‘Engineer’ to ‘Builder’: More People Writing Code-like Instructions
Cherny forecast that the number of people writing code or using agents to write code could grow dramatically, even as the old job title fades. He predicts that there will be “100 times more” such people than today, as AI cuts the skill barrier to shipping software. This implies a world where product managers, domain experts, and founders issue detailed instructions to tools like Claude Code, handling logic in natural language rather than by hand-coding every function. Entry-level software engineers may compete with these new builders, but they also gain a chance to move up the stack into product, data, or AI agent supervision. Instead of a simple story where AI replace software engineers, Cherny’s view points to a messy transition: fewer traditional coding-centric jobs, but many more humans engaged in code-adjacent problem solving.