Graph Database for Agent Systems
GraphSub is a graph database and control substrate for AI agents that need persistent memory, tool traces, replayable outputs, and low-latency graph operations.
Definition
A graph database for agent systems stores people, tasks, tools, memories, approvals, events, and outputs as connected state. GraphSub extends that model with agent-native primitives for memory, replay, capability tokens, and operational proof.
Why it matters
Autonomous workflows fail when state is trapped in logs, chat transcripts, or isolated tools. A graph substrate gives agents a durable map of what happened, which dependencies changed, and which outputs should be recomputed.
GraphSub approach
GraphSub combines sub-millisecond graph operations with hierarchical memory, stigmergic coordination, workflow replay, and capsule packaging for real-world enterprise workflows.
Code example
from graphsub import Client
client = Client(base_url="https://graphsub.com")
memory = client.memory.remember(
tier="working",
content="limit breach on EUR rates book",
source="risk-run-2026-07-03",
importance=0.92,
)
trace = client.capsules.execute(
capsule="trading-control-graph",
workload="risk-recompute",
evidence="source-linked",
)
print(trace.replay_id, trace.affected_outputs)Write agent memory, launch a capsule, and inspect affected outputs.
FAQ
- What is an agent graph database?
- It is a graph database optimized for AI agents, tool traces, memory, approvals, and replayable workflow state.
- How is GraphSub different from a traditional graph database?
- Traditional graph databases focus on storage and query. GraphSub adds agent memory, replay, workflow capsules, and proof-oriented APIs.