Retrieval

The links between what you know

Your knowledge isn't a list; it's a web. The graph layer is an early step toward making the connections — who, what, when, and how things relate — first-class.

ExperimentalConnections

The relationships between what you know

DanaMeeting · Mar 3PricingQ3 LaunchRoadmap#marketingattendedaboutpart oftaggedmentionsowns
Evidence-backed nodes & typed edgesExperimental
Who, what, and how it connects — typed, evidence-backed relationships across your archive.
01

Search finds the thing. It doesn't show you what the thing connects to. The relationships between people, messages, and threads are knowledge too.

02
  • A graph built from import metadata — people, messages, threads, tools, and the edges between them.
  • Read operations to find a node, walk its neighborhood, or trace the shortest path between two.
  • Cross-source identity resolution that recognizes the same person across different sources.
  • Evidence-backed edges: every relationship traces to where it came from.
03
  1. Project the graph for a vault.
  2. Find a person, thread, or entity.
  3. Walk its neighborhood or trace a path to another.
  4. Follow any edge back to its evidence.
04
  • Relationships are derived from your sources, not invented.
  • Every match traces to its evidence — identity resolution you can inspect.
  • An early, opt-in layer — not yet wired into the assistant.