_aleph2c_

I’ve noticed a trade-off in my daily workflow. An LLM is basically a "Senior Dev in my pocket" for almost any technical niche. It’s an amazing resource for clearing blockers, but I’m worried about the long-term cognitive cost.

There is a fundamental difference between having a senior dev tell you the answer and "drilling" the keystone-concepts yourself until they become muscle memory. One gets the things done in a hurry; the other builds self-sufficiency.

I’m starting to feel the "brain-softening" effect where I reach for the prompt before I’ve even tried to orient myself in the problem.

How are you guys handling this? What are your personal guardrails, disciplines, or "hard-mode" rules to ensure you’re using AI as an accelerant rather than a crutch that rots out your technical intuition?