AI Agents can write and modify base Python / C++ / Rust / whatever pretty well, and thus users aren't limited by "sorry the building blocks only go together in this one particular manner".

It's like the difference between an EZ-Bake oven and a fully furnished kitchen. The EZ-Bake oven can get some stuff done but its limits are much more severely obvious than the kitchen's, and the kitchen's first limiting factor in what can be produced is usually the human cooking in it

There is a paper [0] that shows that LLMs, in fact, cannot write and modify ‘whatever’ pretty well.

[0] A Survey on Large Language Models for Code Generation: https://dl.acm.org/doi/epdf/10.1145/3747588