I mostly enjoy the problem solving and a challenge.

Before LLMs often code wasn't actually the hard part, it was more just a procedural break I got between the design or trying to figure out the best problem to solve, rather than just solving the obvious one.

So I always was looking to write the library that encoded conventions for the tradeoffs we want once and reuse it instead of writing bespoke code for every product, which seems to be the road more travelled.

So a multiplicative effect, so I can refocus on the hard parts sooner.

LLMs sure don't help the tendency to solve the obvious problem and writing bespoke code every time though.