arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2604.20158cs.AI
收藏

无状态决策记忆用于企业AI代理

Stateless Decision Memory for Enterprise AI Agents

Vasundra Srinivasan

Vasundra Srinivasan

AI总结:

本文提出无状态决策记忆(DPM),在企业监管领域中通过无状态设计提升决策精度和可审计性,同时提高效率。

摘要:

在受监管领域中,长期视界决策代理的企业部署主要依赖检索增强管道,尽管已有十年的日益复杂的状态ful记忆架构。我们主张这反映了隐藏的需求:受监管部署对四个系统属性(确定性回放、可审计性理由、多租户隔离、无状态性以实现水平扩展)具有负载承受能力,并且状态ful架构通过构造违反这些属性。我们提出确定性投影记忆(DPM):一个只追加事件日志加一个任务条件化的投影在决策时间。在十个受监管决策案例上,三个内存预算下,DPM在慷慨预算下与基于总结的记忆相当,在预算绑定时表现显著更好:在20倍压缩比下,DPM提高了事实精度+0.52(Cohen's h=1.17, p=0.0014)和推理一致性+0.53(h=1.13, p=0.0034),配对排列,n=10。DPM在绑定预算下还快7-15倍,使决策时间只需调用一个LLM而不是N次。对10个案例每个进行10次回放的确定性研究显示,两种架构都继承了残余的API级非确定性,但这种不对称是结构性的:DPM暴露一个非确定性调用;总结暴露N次叠加调用。审计表面遵循相同的一对N模式:DPM每个决策记录两个LLM调用,而总结记录LongHorizon-Bench上83-97次。我们最后提出TAMS,一个从业者启发式方法用于架构选择,以及在企业运营条件下状态ful记忆的失败分析。贡献是论证无状态性是解释企业偏爱较弱但可回放检索管道的负载承受属性,并且DPM展示了这种属性在无决策惩罚检索下是可实现的。

英文摘要:

Enterprise deployment of long-horizon decision agents in regulated domains (underwriting, claims adjudication, tax examination) is dominated by retrieval-augmented pipelines despite a decade of increasingly sophisticated stateful memory architectures. We argue this reflects a hidden requirement: regulated deployment is load-bearing on four systems properties (deterministic replay, auditable rationale, multi-tenant isolation, statelessness for horizontal scale), and stateful architectures violate them by construction. We propose Deterministic Projection Memory (DPM): an append-only event log plus one task-conditioned projection at decision time. On ten regulated decisioning cases at three memory budgets, DPM matches summarization-based memory at generous budgets and substantially outperforms it when the budget binds: at a 20x compression ratio, DPM improves factual precision by +0.52 (Cohen's h=1.17, p=0.0014) and reasoning coherence by +0.53 (h=1.13, p=0.0034), paired permutation, n=10. DPM is additionally 7-15x faster at binding budgets, making one LLM call at decision time instead of N. A determinism study of 10 replays per case at temperature zero shows both architectures inherit residual API-level nondeterminism, but the asymmetry is structural: DPM exposes one nondeterministic call; summarization exposes N compounding calls. The audit surface follows the same one-versus-N pattern: DPM logs two LLM calls per decision while summarization logs 83-97 on LongHorizon-Bench. We conclude with TAMS, a practitioner heuristic for architecture selection, and a failure analysis of stateful memory under enterprise operating conditions. The contribution is the argument that statelessness is the load-bearing property explaining enterprise's preference for weaker but replayable retrieval pipelines, and that DPM demonstrates this property is attainable without the decisioning penalty retrieval pays.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑