Is there an LLM AI we can use from prow?

I was thinking it would be very useful if we could use an LLM from prow and prowjobs. I don't think we know what people will come up with, but one idea to make it concrete: we could test our documentation, giving the LLM _only_ the docs and seeing whether it can answer key questions we define. Do we have this already? If we wanted to use GCP, I think this would look like a GCP project with the vertex APIs enabled.