Zep: A Temporal Knowledge Graph Architecture for Agent Memory

We introduce <PRE_TAG>Zep</POST_TAG>, a novel memory layer service for AI agents that outperforms the current state-of-the-art system, <PRE_TAG>MemGPT</POST_TAG>, in the <PRE_TAG>Deep Memory Retrieval (DMR)</POST_TAG> benchmark. Additionally, <PRE_TAG>Zep</POST_TAG> excels in more comprehensive and challenging evaluations than DMR that better reflect real-world enterprise use cases. While existing retrieval-augmented generation (RAG) frameworks for large language model (LLM)-based agents are limited to static document retrieval, enterprise applications demand dynamic knowledge integration from diverse sources including ongoing conversations and business data. <PRE_TAG>Zep</POST_TAG> addresses this fundamental limitation through its core component <PRE_TAG>Graphiti</POST_TAG> -- a <PRE_TAG>temporally-aware</POST_TAG> <PRE_TAG>knowledge graph engine</POST_TAG> that dynamically synthesizes both <PRE_TAG>unstructured conversational data</POST_TAG> and <PRE_TAG>structured business data</POST_TAG> while maintaining <PRE_TAG>historical relationships</POST_TAG>. In the DMR benchmark, which the <PRE_TAG>MemGPT</POST_TAG> team established as their primary evaluation metric, <PRE_TAG>Zep</POST_TAG> demonstrates superior performance (94.8% vs 93.4%). Beyond DMR, <PRE_TAG>Zep</POST_TAG>'s capabilities are further validated through the more challenging LongMemEval benchmark, which better reflects enterprise use cases through complex temporal reasoning tasks. In this evaluation, <PRE_TAG>Zep</POST_TAG> achieves substantial results with accuracy improvements of up to 18.5% while simultaneously reducing response latency by 90% compared to baseline implementations. These results are particularly pronounced in enterprise-critical tasks such as cross-session information synthesis and long-term context maintenance, demonstrating <PRE_TAG>Zep</POST_TAG>'s effectiveness for deployment in real-world applications.