May 30, 2026
Ctrl+Alt+Elite
Domain expertise has always been the real moat
AI can write the app, but commenters say only insiders know when it’s secretly wrong
TLDR: The article says AI has made writing software easier, so the real value now is knowing a field well enough to tell when the result is wrong. Commenters split hard: some say that human insider knowledge is now priceless, while others insist AI can learn that too and nobody’s job is safe.
A spicy new argument lit up the community: the real power in software was never typing code — it was knowing what “correct” looks like. The article claims today’s AI tools can spit out working programs fast, which means the old edge of “I know how to build this” is getting cheaper. The new flex? Being the person who can spot when the answer is plausible nonsense — especially in messy real-world fields like payroll, medicine, transport, or insurance.
And oh, the comments did not stay calm. One camp basically said, “Nice theory, but the robot already swallowed the textbook.” They argued that large language models — AI systems trained on mountains of human writing and code — already contain loads of business know-how, so developers can just ask questions and speedrun their way to competence. Another commenter went even harder, warning that anyone treating their niche job as untouchable is living in denial.
But the backlash was just as loud. Skeptics rolled their eyes at what they saw as yet another identity crisis from software workers trying to explain why they still matter. One especially savage jab compared this whole debate to textile workers reacting to the machines during the Industrial Revolution — not subtle. Meanwhile, others sided with the article and said domain knowledge was always the interesting part anyway: if a spreadsheet solves the problem, who cares about fancy apps? In short, the crowd couldn’t agree on whether AI is killing expertise, exposing its value, or just giving everyone a new excuse to panic online.
Key Points
- •The article states that software development has traditionally depended on first understanding the underlying business or operational domain.
- •It argues that agentic AI now enables software generation without requiring the user to build the full domain model beforehand.
- •The article says the main constraint has shifted from the ability to build software to the ability to judge whether the output is correct.
- •It contrasts domain experts, who can validate correctness, with generalist engineers, who may build robust systems but miss domain-specific errors.
- •The article concludes that people who combine domain expertise with engineering ability will be best positioned because they can verify both implementation quality and domain correctness.