Agent SystemsEXPERIMENTexperiment · memory graphs

Graph Memory

Relationship memory between agents, tools, and knowledge.

The Thesis & Bottleneck

The Real-World Bottleneck

Vector recall answers what is similar. Agents also need what is related: who decided, with which tool, toward what outcome.

The Architectural Insight

Memory should be a graph — agents, memory, and knowledge as connected nodes — not a pile of embeddings.

How I Built It

A bench for relationship memory: entities and decisions linked across sessions, retrievable by relation as well as similarity.

Technical Stack & Protocols
PostgrespgvectorHermes

Real Outcomes & Lessons

Observed Production Outcome

Open research: relation-aware recall for multi-agent coordination.

What I Learned The Hard Way

“The interaction itself becomes documentation when memory is traversable.”

System Telemetry

  • System TrackAgent Systems
  • Current StatusEXPERIMENT
  • Architectureexperiment · memory graphs
  • Launch StageBench experiment feeding Raven and the production agent setup.

System Access

Interested in how this architecture works under load?

Book a 30-min discovery call →