Query local data PLUS the main ChatGPT LLM

[interzonez]

I can setup Langchain to ingest data or text into a vector database. I can then ask questions of that data.

How can I call-out to the full ChatGPT model using a local memory state for my next query - to the core LLM.

Example:
"Based on the document that has just been analysed write some Python code to implement x,y,z."

OR
"Based off understanding the ingested data, describe other cases where there have been similar x,y,z results."

I guess this is about keeping a live memory state and building on top of it?