I feel like if I wanted to read LLM output I could ask it myself