arXiv cs.AI· Subhrangshu Nandi, Hitesh Arora, Yunfei Bai, Udita Patel, Chirag Shah, Hannah Kwon, Wei Zhang, Weian Chen, Chloe Ruoqi Li, Alberto Sconfienza, Ayush Goyal, Yasmine Lee, Ming Xiong, Prathap Ramachandra·· 2 小时前精选AI 评分60
arXiv 论文：企业 SOP 场景下四种 LLM Agent 架构的成本与成功率权衡
Agent Architecture Cost-Success Trade-offs: A Controlled Evaluation across Enterprise-Style SOPs
AI 导读
arXiv 论文 2611.12953 在固定模型、任务与工具的条件下，比较函数调用单智能体、技能增强智能体、中心化多智能体编排和 triage swarm 四种架构（八组臂）在 13 个企业 SOP 上的表现。
推荐理由
论文在固定模型与工具的受控条件下比较四种智能体架构，给出成本与成功率的量化证据，可用于企业选型参考。
正文 · 原文
Authors:Subhrangshu Nandi, Hitesh Arora, Yunfei Bai, Udita Patel, Chirag Shah, Hannah Kwon, Wei Zhang, Weian Chen, Chloe Ruoqi Li, Alberto Sconfienza, Ayush Goyal, Yasmine Lee, Ming Xiong, Prathap Ramachandra
Abstract:Enterprises increasingly use large language model agents to execute standard operating procedures (SOPs): long, branching, tool-intensive workflows. Yet there is little controlled evidence for choosing among competing agentic architectures. Holding the model, tasks, and tools fixed and varying only the architecture, we compare four designs (a function-calling single agent, a skill-augmented agent, centralized multi-agent orchestration, and a triage swarm; eight arms) across 13 enterprise SOPs, and introduce a skill-definition quality rubric to build matched high and low-quality skill corpora. Three takeaways emerge. (1)~\textbf{There is no universal best architecture; the right choice is SOP-dependent.} On a given SOP, task success ranges from under $10\%$ on the hardest to near $100\%$ on the easiest; no architecture is statistically separable in aggregate, and added structure only plausibly pays off on complex, branching SOPs (a directional, not yet significant, trend). (2)~\textbf{Added architecture cost must be justified by a task-success gain.} Every coordinated arm (multi-agent or skill-augmented) spends $1.4$--$4.7\times$ the tokens and $2.6$--$5.8\times$ the latency of the single agent for gains that are usually small and rarely significant, making the single agent the efficient default; high-quality skills do lift success (to $73.1\%$ from $67.4\%$) while degraded skills fall below it. (3)~\textbf{Implementation quality outweighs the single- versus multi-agent choice.} A plausible multi-agent implementation loses to the single agent, yet a targeted redesign of it becomes the only arm to significantly beat the baseline ($75.6\%$ vs.\ $67.4\%$).
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.12953 [cs.AI] |
| (or arXiv:2611.12953v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.12953 arXiv-issued DOI via DataCite |
Submission history
From: Hitesh Arora [view email]
[v1]
Sat, 7 Nov 2026 05:12:41 UTC (1,600 KB)
来源：arXiv cs.AI · arxiv.org