Paper 2511.18423
General Agentic Memory Via Deep Research
- Published
- Nov 2025
- Research lab
- Independent
- Citations
- 34
- GitHub
- 859 stars
01 In brief
Summary
The paper introduces General Agentic Memory (GAM), a memory framework for AI agents that follows a just-in-time (JIT) compilation principle, contrasting with traditional ahead-of-time (AOT) static memory systems that suffer from information loss.
GAM uses a dual-agent design: a Memorizer that compresses key historical information into lightweight memory while preserving complete history in a page-store, and a Researcher that performs deep research at runtime to retrieve and integrate relevant information for client requests.
This approach enables high-fidelity, task-adaptable memory, domain generalizability, and optimization via reinforcement learning.
Experiments on benchmarks including LoCoMo, HotpotQA, RULER, and NarrativeQA show GAM consistently outperforms memory-free (long-LLM, RAG) and memory-based (A-Mem, Mem0, MemoryOS, LightMem) baselines.
Ablations show the importance of both modules, multiple search tools, and test-time scaling (reflection depth, retrieved pages).
Efficiency analysis shows GAM offers competitive time costs with superior answer quality.
02 From the paper
Abstract
Memory is critical for AI agents, yet the widely-adopted static memory, aiming to create readily available memory in advance, is inevitably subject to severe information loss. To address this limitation, we propose a novel framework called \textbf{general agentic memory (GAM)}. GAM follows the principle of "\textbf{just-in time (JIT) compilation}" where it focuses on creating optimized contexts for its client at runtime while keeping only simple but useful memory during the offline stage. To this end, GAM employs a duo-design with the following components. 1) \textbf{Memorizer}, which highlights key historical information using a lightweight memory, while maintaining complete historical information within a universal page-store. 2) \textbf{Researcher}, which retrieves and integrates useful information from the page-store for its online request guided by the pre-constructed memory. This design allows GAM to effectively leverage the agentic capabilities and test-time scalability of frontier large language models (LLMs), while also facilitating end-to-end performance optimization through reinforcement learning. In our experimental study, we demonstrate that GAM achieves substantial improvement on various memory-grounded task completion scenarios against existing memory systems.