跳到正文
arXiv cs.MA· Jiangyun Zhang, Zhou Tang, Kristen Surrao, Zhiyu Chen, Albert Guan, Tor Nitayanont, Yupei Zhang, Roopali Singh, Yunkai Xiao, Shreya Pandit, Julia Huang, Shayan Ali Akbar, Erwin Cornejo, Yuan Li, Yi Zhang·· 23 小时前AI 评分40

ARISE:面向企业智能体的自进化记忆架构,无需更新模型权重即可提升准确率

ARISE: Practical Self-Evolving Memory for Accurate and Cost-Efficient Enterprise Agents

AI 导读

ARISE 是一种企业 LLM 智能体记忆架构,让记忆从自身失败中进化,无需更新模型权重、工具或工作流逻辑。在 SOP-Bench 和 τ²-bench 的 14 个企业工作流域、5 个模型上,用 Claude Opus 4.6 做归因与整合的 ARISE-Teacher 分别比基线提升 10.9 pp 和 5.8 pp,并超过 GEPA 4.5 pp 和 1.4 pp。

正文 · 原文

Authors:Jiangyun Zhang, Zhou Tang, Kristen Surrao, Zhiyu Chen, Albert Guan, Tor Nitayanont, Yupei Zhang, Roopali Singh, Yunkai Xiao, Shreya Pandit, Julia Huang, Shayan Ali Akbar, Erwin Cornejo, Yuan Li, Yi Zhang

View PDF HTML (experimental)

Abstract:Enterprise LLM agents operate in dynamic environments where foundation models evolve rapidly, workflows update frequently, and deployments need to satisfy accuracy and cost requirements. We propose ARISE (Agent Refinement through Iterative Self-Evaluation), a practical memory architecture in which the agent's memory evolves from its own failures without updating model weights, tools, or workflow logic. ARISE uses the model itself or a teacher model to attribute the failures and propose corrections. Then, ARISE keeps the corrections as retrievable micro memory, consolidates them into macro rules, and accepts a rule update only if held-out accuracy rises. Across 14 enterprise-workflow domains from SOP-Bench and $\tau^2$-bench using 5 models, ARISE-Teacher, which uses Claude Opus 4.6 for attribution and consolidation, gains 10.9 pp over the baseline on SOP-Bench and 5.8 pp on $\tau^2$-bench. ARISE-Teacher also outperforms GEPA by 4.5 pp on $\tau^2$-bench and 1.4 pp on SOP-Bench. With the deployment model as its own attributor, ARISE-Self still gains 10.0 pp on SOP-Bench. ARISE-Teacher uses 4.0% fewer model calls and 5.2% fewer output tokens at comparable per-task cost (-1.0%).
Comments: 16 pages (8 main), 7 figures, 12 tables
Subjects: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
ACM classes: I.2.7
Cite as: arXiv:2611.23588 [cs.AI]
  (or arXiv:2611.23588v1 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2611.23588

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Jiangyun Zhang [view email]
[v1] Tue, 17 Nov 2026 22:38:58 UTC (370 KB)

来源:arXiv cs.MA · arxiv.org