Last updated

Unified Graph

Papr connects chat messages, documents, and structured data into one memory graph so retrieval can span across sources.

Why It Matters

Without a unified graph, context is fragmented:

  • chat in one store,
  • docs in another,
  • structured records elsewhere.

With Papr, these sources can be connected and retrieved in one query flow.

Input Paths That Converge

  1. Messages (/v1/messages)
  2. Documents (/v1/document)
  3. Direct memory and structured ingest (/v1/memory)

Retrieval Layer

Use /v1/memory/search with enable_agentic_graph=true to retrieve:

  • relevant memories,
  • connected entities,
  • relationship-aware context.