Exactly. Llms can output code, but they don’t understand long-term intent. As llm usage grows, engineers who truly understand the system become more valuable, not less. Knowing which changes are safe and why decisions were made is now an even more critical skill. It’s the backbone of any resilient system.