01 |ONE GRAPH
One graph for turns and documents. Every fact cites its source.
Conversational turns and parsed documents write to the same entity and relation graph. Each fact keeps a source object, so a fact from a turn or a document resolves to the passage it came from.
02 |WHAT YOU GET
Entities, relations and passages. Joined in one graph.
Directed relations
Turns and documents together
Documents parsed
Linked by section and keyword
Cited to the passage
Vectors beside the graph
03 |WHY IT MATTERS
What a vector store cannot say. Which entity a fact is about, and where it came from.
One graph across sources
A fact learnt in conversation and a fact extracted from a document attach to the same entity, so recall can walk from one to the other.
A source object on every fact
Each fact stores its source as a kind and a trust level, with a reference and span back to the passage, or the facts it was derived from, so a citation points at its origin.
One transaction per fact
Each fact, its edges and its history are written in one transaction, so a partial write never lands.
Entities are matched by name
Differently named references to the same entity are not merged automatically; that is a deliberate trade-off.
GET STARTED
Give your agents connected memory. One graph for conversations and documents, with every fact carrying its source.
