Yeah experimentation is probably the best use case of LLMs

And at the end of the experiment I usually have my solution