MemForest: An Efficient Agent Memory System with Hierarchical Temporal Indexing
cs.DB, cs.AI, cs.MA
Submitted: 2026-05-16
Updated: 2026-09-06
Comments: 12 pages. Extended version with appendix as supplemental material. Submitted to VLDB
Code: https://github.com/Concyclics/MemForest
License: http://creativecommons.org/licenses/by/4.0/
The gist: Memory is a fundamental component for long-context LLM agents, supporting persistent state across interactions through a continuous serve-and-update lifecycle.
Terminology
Abstract
Memory is a fundamental component for long-context LLM agents, supporting persistent state across interactions through a continuous serve-and-update lifecycle. Despite substantial prior work, many stateful systems retain sequential autoregressive extraction or state-dependent maintenance on the write path, delaying when new evidence becomes queryable. To address these challenges, we present MemForest, a memory framework that reformulates agent memory as a write-efficient temporal data-management problem. MemForest breaks the sequential bottleneck via parallel extraction, decoupling memory construction into concurrent, independent operations. We further introduce MemTree, a hierarchical temporal index that organizes memory as time-ordered trees and replaces global rewrites with localized dirty-path refresh. Dirty summaries can be refreshed in parallel across nodes and trees. End-to-end work remains proportional to incoming content; the logarithmic bound applies only to structural insertion and level-dependent refresh depth in balanced trees. We evaluate MemForest on two long-context benchmarks, LongMemEval-S and LoCoMo. Experiments use Qwen3-4B, Qwen3-30B, and Gemma-4-12B-IT. With Qwen3-30B, MemForest reaches 81.8 percent pass at 1 on LongMemEval-S, while its input-normalized build rate is 6.0 times that of EverMemOS. On LoCoMo categories 1 to 4, it reaches 84.09 percent, within 0.13 percentage points of EverMemOS; on a matched conversation, its build rate is 9.5 times higher. These results show that MemForest reduces memory-freshness latency while retaining strong answer quality.
Sources
- Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory
- From Local to Global: A Graph RAG Approach to Query-Focused Summarization
- Decentralizing AI Memory: SHIMI, a Semantic Hierarchical Memory Index for Scalable Agent Reasoning
- EverMemOS: A Self-Organizing Memory Operating System for Structured Long-Horizon Reasoning
- Choosing How to Remember: Adaptive Memory Structures for LLM Agents
- Zep: A Temporal Knowledge Graph Architecture for Agent Memory
- Collaborative Memory: Multi-User Memory Sharing in LLM Agents with Dynamic Access Control
- Gemma 4 Technical Report
- DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence
- Qwen3 Technical Report
- HiMem: Hierarchical Long-Term Memory for LLM Long-Horizon Agents
- Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models
- Are We Ready For An Agent-Native Memory System?
Related papers
- Vibe Coding on Trial: Operating Characteristics of Unanimous LLM Juries
- Human-Level Text-to-SQL via Reinforcement Learning on Verified Data, Without Pipeline Engineering
- Bridging Business Intent and Data: A Benchmark for Automatic Relational Data Product Generation
- DexterSQL: Deep Schema Exploration and Rule-based Correction for Text-to-SQL Generation
- MaDI-Bench: An End-to-End Data Integration Benchmark
- Eigenius: A Typed Knowledge-Graph DBMS with Epistemic Stratification and Institution-Mediated Reasoning