Search papers, labs, and topics across Lattice.
This paper introduces a novel graph-native bitemporal memory store for conversational AI agents that enables persistent memory without exhausting context budgets or compromising user data privacy. By utilizing a Neo4j property graph with HNSW vector indexes, the system supports semantic retrieval of memories linked to versioned content nodes, which are timestamped with both valid and transaction times. Evaluated on the LongMemEval benchmark, the memory store achieves a high retrieval rate on knowledge-update questions, revealing critical insights into the limitations of retrieval-based approaches for various question types.
A memory store design that enables conversational AI agents to recall information across sessions without sacrificing user privacy or context limits, achieving up to 80% retrieval accuracy on knowledge-update questions.
Conversational AI agents commonly lack persistent memory across sessions. The obvious fixes like injecting full chat histories into the context window, or delegating to a third-party memory service, either exhaust the model's context budget or send personal data through infrastructure the user does not control. We describe a memory store that avoids both problems: an agent-local Neo4j property graph augmented with HNSW vector indexes and a full bitemporal data model. Each memory is stored as an immutable identity node linked to versioned content nodes carrying two closed-open time intervals: valid time (when the fact was true in the world) and transaction time (when the database recorded it). This design supports point-in-time semantic retrieval without physically overwriting history. Semantic edges between related memories are maintained automatically at write time using cosine similarity over 1024-dimensional embeddings. We evaluate the system on LongMemEval, a 500-question benchmark spanning six question types designed to stress long-term memory. Across 60 sampled questions, the current-state semantic search path achieves 46.7% R@10 overall, rising to 80% on knowledge-update questions. The time-travel path yields 80% R@10 on knowledge-update but decreases recall on temporal-reasoning questions (50% to 37.5%), a consequence of post-filter dilution that points directly to a concrete design improvement. We discuss what these results reveal about the limits of pure retrieval for different question types and what each failure mode suggests for future work.
