And good agentic AI usage seems to come down to two things: being able to judge AI output quality, and making sound architectural decisions, like not defaulting to React just because the LLM is most comfortable with it, or not picking a stack based on what the model prefers rather than what the project actually needs.
So my question is: should I be pivoting my learning toward books and courses on code quality and best practices (testing, edge cases, defensive patterns), and on system design (knowing when to reach for React vs Vue vs Astro, Cloudflare vs Firebase, etc.)?
Curious to know what you think about this