Domain expertise has always been the real moat
Agentic AI makes code generation cheap, raising the value of domain expertise over software engineering skill.
The author argues that software engineering's core difficulty has always been understanding a domain, not writing code itself. Agentic AI severs this link by generating working software without requiring the developer to build a mental model of the domain.
Two personas now emerge: domain experts without software background (dispatchers, clinical coders, actuaries) can use agents effectively because they know what correct outputs look like. Strong generalist engineers unfamiliar with a domain cannot verify whether generated code is correct, only whether it's well-built. Pre-AI, engineers could climb into domains by learning; domain experts had no equivalent path to coding skill.
The most valuable person is now someone with both skillsets: they can verify generated code is sound and its outputs are true. The author recommends experienced engineers invest years learning a specific domain, regulatory regime, or physical process rather than refining mechanical coding ability, since agents now provide that transcription cheaply.
What HN community is saying
Commenters split between skepticism and reinforcement of the thesis. Top-ranked dissent argues AI models are already trained on similar implementations and that domain expertise itself is a target for LLM replacement, not a safe moat. One commenter directly challenges the article's framing as self-delusion comparable to textile workers denying automation.
Defenders of the domain-expertise view note that LLMs encode knowledge without genuine understanding and that querying an LLM does not make one a domain expert. A real estate title insurance consultant provides concrete counter-evidence: domain knowledge in that field is so deep and multivariate that neither AI nor short-term learning creates competence. Others draw parallels to self-driving cars and chess, where machines surpass humans but human demand persists. The thread's center of gravity is skepticism toward the article's premise that domain expertise remains scarce.