Storage Is Not Memory: A Retrieval-Centered Architecture for Agent Recall
TLDR
True Memory introduces a retrieval-centered architecture for agent recall, achieving high accuracy by preserving verbatim events and outperforming existing systems.
Key contributions
- Proposes "True Memory," a six-layer architecture centered on a multi-stage retrieval pipeline.
- Preserves events verbatim, shifting from ingestion-time extraction to improve recall.
- Runs efficiently as a single SQLite file on commodity CPU without external databases or GPUs.
- Achieves 93.0% accuracy on LoCoMo, outperforming Mem0, Supermemory, and Zep.
Why it matters
This paper introduces a novel agent memory architecture that significantly improves recall accuracy by prioritizing retrieval over storage-centric approaches. Its efficient, self-contained design makes it practical for real-world applications, setting a new standard for agent memory systems.
Original Abstract
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.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.