arXiv cs.AI· Kaiyuan Hou, Yang Ping Kuo, Indranil Bhattacharya, Rohith Nama, Arushi Gupta, Carlos D. Castillo, Prathap Ramachandra, Xiaofan Jiang, Francesco Carbone·· 2 小时前AI 评分48
ReTrace：将使用工具的 AI 智能体蒸馏为免规划器工作流
ReTrace: Distilling Tool-Using Agents into Planner-Free Workflows
AI 导读
ReTrace 将 SOP 和少量已记录的智能体运行日志转化为可复用工作流，由代码而非 LLM 选择下一步骤，规则文档可针对不同编排平台编译。在其生产平台的 79 个流程上，ReTrace 工作流完成率 93.8%（智能体 96.8%），单任务成本低 24%；在 τ²-Bench 上任务成功率 75.8%（智能体 80.6%），运行时间从 170 秒降至 1.3 秒。
正文 · 原文
Abstract:Business process automation runs the same Standard Operating Procedure (SOP) again and again on new inputs. Large Language Model (LLM) agents can follow an SOP written in natural language, but they call an LLM to choose every step of every run, which adds latency and cost at scale. We present ReTrace, which turns an SOP and a few logged agent runs into a reusable workflow in which code, not an LLM, selects the next step. The agent's reasoning, the explanation it writes with each tool call, often states why it took an action; the tool calls alone do not. ReTrace extracts the condition behind each step, merges these conditions across runs into a compact rule document, and generates an executable workflow from it. Only this last stage depends on the platform, so rule documents can be compiled for different orchestration platforms. On $79$ procedures from our production platform, ReTrace workflows reach $93.8\%$ completion, against $96.8\%$ for the agent, at $24\%$ lower cost per task. On $\tau^2$-Bench, they reach $75.8\%$ task success (agent: $80.6\%$) and run in $1.3$s instead of $170$s; three rounds of repair on training failures lift $k{=}2$ workflows from $66.7\%$ to $81.2\%$. On eight SOP-Bench procedures, ReTrace without repair trails few-shot generation ($80.6\%$ vs. $83.6\%$), but repair from labeled training failures lifts it to $87.8\%$ at $0.40$s per task, compared to $85.5\%$ in $46.6$s for the agent.
| Subjects: | Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multiagent Systems (cs.MA) |
| Cite as: | arXiv:2611.14492 [cs.AI] |
| (or arXiv:2611.14492v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.14492 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Indranil Bhattacharya [view email]
[v1]
Tue, 10 Nov 2026 18:49:47 UTC (544 KB)
来源：arXiv cs.AI · arxiv.org