See, what we do is run an LLM, and when it makes mistakes, we will have another LLM supervise and correct it. Of course, that LLM will also make mistakes, but it’s okay because will just have another LLM to supervise the 2nd one! Of course that LLM…