I am trying to structure outputs from a Large Language Model (LLM) and came across the instructor package, which is a true gem. However, I don’t fully understand how instructor works “under the hood.” Here are two simple examples that achieve the same result: