Extraction at ingestion is the wrong primitive for agent memory: content discarded before the query is known cannot be recovered at retrieval time. We propose True Memory, a six-layer architecture that shifts the center of the system from a storage schema to a multi-stage retrieval pipeline operating over events preserved verbatim. The full system runs as a single SQLite file on commodity CPU with no external database, vector index, graph store, or GPU. On LoCoMo (1,540 questions across 10 multi-session conversations), True Memory Pro reaches 93.0% accuracy (3-run mean) against 61.4% for Mem0, 65.4% for Supermemory, approximately 71% for Zep, and 94.5% for EverMemOS under a matched gpt-4.1-mini answer model. On LongMemEval (500 questions), True Memory Pro reaches 87.8% (3-run mean). On BEAM-1M (700 questions at the 1-million-token scale), True Memory Pro reaches 76.6% (3-run mean), above the prior published result of 73.9% for Hindsight. A 56-configuration ablation shows a 1.3-percentage-point spread within the top-performing configuration family.

翻译：摄入时抽取是代理记忆的错误原语：查询前丢弃的内容在检索时无法恢复。我们提出True Memory——一种六层架构，将系统中心从存储模式转向基于逐字保留事件的多阶段检索流水线。该完整系统作为单个SQLite文件运行于通用CPU之上，无需外部数据库、向量索引、图存储或GPU。在LoCoMo（涵盖10个多轮对话的1,540个问题）上，True Memory Pro在匹配的gpt-4.1-mini答案模型下达到93.0%的准确率（三轮均值），而Mem0为61.4%、Supermemory为65.4%、Zep约为71%、EverMemOS为94.5%。在LongMemEval（500个问题）上，True Memory Pro达到87.8%（三轮均值）。在BEAM-1M（百万token规模下的700个问题）上，True Memory Pro达到76.6%（三轮均值），高于此前公布的Hindsight方法73.9%的结果。一项包含56种配置的消融实验显示，最优配置族内的性能波动幅度仅为1.3个百分点。