The author of the viral "LLMs are eroding my career" post doubles down on their warnings, arguing that AI is a compounding, cross-domain threat that will inevitably commoditize software engineering and other knowledge professions.

Hot take: The belief that software engineering has an infinite demand curve is a dangerous coping mechanism; AI is set to commoditize software development in the same way it decimated copywriting and UX writing.

• –**Documentation is Agent-Friendly**: With files like `AGENT.md`, agents can quickly master internal codebases and domain-specific rules, destroying the traditional information moat held by senior developers.
• –**Vibecoding Juggling**: The author reveals they use workarounds like writing generic design documents and adding extra test tickets to buy time for careful review, a survival strategy against the chaotic rush of AI-native engineering.
• –**The Fallacy of Infinite Demand**: Contrasting with Jevons Paradox, the author points out that the demand for copywriting and software has actual ceilings, meaning 10x developer productivity will translate to smaller team sizes and lower wages rather than endless new roles.
• –**Reinforcement Learning Destroys Moats**: Labs are actively training models to write high-quality code and learn engineering principles, meaning even the most complex architectures and clean-coding practices will eventually be automated.