Use LLMs to identify bugs that are hard to see with the human eye like mispelling a variable at one place or using the wrong variable in another. I also like to use it to decide the outline or direction a project should go