> if it can be provided fast enough

That's my biggest gripe with it. If I ask a quick throwaway question, I'd really rather not wait for the LLM to build a test framework for its composable principles-driven components framework and WebGL/WebGPU abstraction first.