yeah, you can create a pipeline : extract transcript => summrize/structure the transcript using a LLM => add to the KB using API