Detailed Analysis
Boris Cherny, a prominent engineer at Anthropic, has renewed his publicly stated argument that traditional "software engineering" as a professional discipline is effectively dead, pointing to conditions at Anthropic itself as evidence of the transformation underway. Cherny, who previously gained recognition in the developer community as the creator of Flow (Facebook's JavaScript type checker) and as a TypeScript expert, has been an increasingly vocal figure in the debate over how large language models and AI coding assistants are fundamentally restructuring the nature of software development work. His comments represent a continuation of a broader internal and public conversation at Anthropic about how the company's own tools are reshaping the profession that built them.
The argument Cherny is advancing reflects a position gaining traction among a subset of AI-adjacent technologists: that the discrete, skill-based craft of writing, debugging, and architecting code — the traditional definition of software engineering — is being rapidly subsumed by AI systems capable of performing those tasks autonomously or near-autonomously. At Anthropic specifically, which develops the Claude family of models, engineers reportedly work in an environment where AI handles an increasing share of what would previously have required dedicated human coding effort. This creates a self-referential dynamic in which the company building AI that transforms software development is itself a living demonstration of that transformation.
The statement carries particular weight coming from within Anthropic because the company occupies a unique position at the frontier of AI capability research. Unlike technology companies that merely adopt AI tools, Anthropic both produces and is subject to the displacement effects of its own models. Claude's coding capabilities — including agentic coding through products like Claude Code — have been cited repeatedly by Anthropic leadership, including CEO Dario Amodei, as accelerating toward a point where AI agents could conduct the majority of AI research itself. Cherny's remarks fit within this framing, suggesting the transition is not theoretical but already operational within the company's own engineering culture.
The broader industry context for Cherny's claim is one of accelerating disruption. GitHub Copilot, Amazon CodeWhisperer, Google's Gemini-integrated development tools, and Anthropic's own Claude Code have all demonstrated measurable productivity gains for developers, with some studies citing 30–55% increases in coding speed. However, critics of the "software engineering is dead" framing argue that the discipline is evolving rather than disappearing — that engineering judgment, system design, security reasoning, and product intuition remain deeply human contributions even as mechanical code generation becomes automated. The debate ultimately centers on whether AI is a force multiplier for engineers or a replacement for the role itself, a question that Cherny and Anthropic appear to be answering, at least provisionally, in favor of the latter.
Read original article →