@smerfj - I'm currently researching same problem. You can find some information in LlamaIndex project docs. What you probably need is so called composite index with both vector database and knowledge graph that links particular knowledge bits or text paragraphs together. Alternatively you can try restricting vector search to chunks computed from one particular document.
I suspect that knowlege graphs are "the shit" because you can keep and query really small but highly relevant pieces of data without overflowing LLM context and slowing it down.
In my company we've been investigating 1Gbps ethernet link between Cisco switches having only 150 kbps data rate.
It was a big ISP with engineers who could decipher IP packets from screen hex dump in their mind (I shit you not) and best Cisco support money can buy.
After few weeks it turned out that disabling auto speed negotiation and forcing 1G rate fixed the issue. And yes - all interfaces claimed that it was always 1G negotiated.