Index scope

No limit! You can add as much as you want to an index. But with that many documents, you might want to look into using a vector store like pinecone or qdrant to be more efficient

Some people have had luck organizing documents into topics, and then creating an index per topic.

Then, each index could be a tool in a langchain agent, or you can wrap them all with a top level vector or keyword index to help route queries to the right data

This sounds like I'm building a massive graph data-structure

vector store 🤔

Yea kinda, just one approach though. A single vector index could also work well, just need to increase similarity_top_k in the query

Here's a guide on building the graph structure though
https://gpt-index.readthedocs.io/en/latest/how_to/index_structs/composability.html

Here's another on using langchain https://gpt-index.readthedocs.io/en/latest/guides/tutorials/building_a_chatbot.html

as I read through this I'll have more questions. Mind if I ping you in a few hours with questions in this thread @Logan M ?

or rather. I'll ask self contained questions outside of it. The Scope question has kind of been answered.

Yea ask wherever haha I have to step out but I'll be back in a few hours to check in