An AI agent's memory is not a database. A database stores facts that are either current or deleted. An agent's memory stores beliefs that evolve: a user says they live in Boston, then later mentions moving to Portland. The first statement was true when stored. It is now superseded but not wrong — the history matters for understanding the trajectory, even though only the latest state is actionable.
Kumiho (arXiv:2603.17244) treats agent memory as a graph with formal belief revision semantics. Each piece of knowledge is a node. Relationships between knowledge are edges. When new information arrives, the graph doesn't just add a node — it applies AGM belief revision: the new belief is integrated, contradictions with existing beliefs are resolved, and the resolution is recorded as a version transition. The history of revisions is preserved as a version graph, not just a snapshot.
The dual use is the architectural insight: the same graph structure that handles belief revision also handles versioning of agent-produced assets — documents, code, plans, analyses. When an agent revises a belief, the downstream artifacts that depended on that belief can be traced through the graph and flagged for update. The knowledge structure and the work product are unified in a single versioned representation.
Performance: 93.3% accuracy on long-conversation memory benchmarks, roughly doubling the next best baseline. The improvement comes not from better retrieval (finding relevant memories) but from better revision (knowing which memories to update and how). Most memory systems fail not because they forget but because they remember too much — contradictory facts coexist without resolution, and the agent generates responses that reflect the unresolved contradiction.
Memory is not storage. Memory is the ongoing negotiation between what was believed and what is now known.