Our app allows users to converse with multiple llm, switching mid conversation if desired. I’m curious if we could include mockingbird as the grounded, professional llm, would it be superior at the end of a long conversation if asked to determine the action items and format a summary?
The flow in our app currently would allow you to use vectara search API to pull in your corporate knowledge at any point of the conversation.
I’d mockingbird optimized for RAG so tightly that it needs to be at the start of a conversation or only one shot RAG LLM responses?