Back to AI News
arxiv

ReTrace: Distilling Tool-Using Agents into Planner-Free Workflows

ReTrace converts a natural-language SOP and a few logged agent runs into a reusable workflow where code, not an LLM, selects each next step. It extracts the conditions behind actions from the agent's written reasoning, merges them into a compact rule document, and compiles that into an executable workflow for different orchestration platforms. On 79 production procedures, ReTrace workflows reached 93.8% completion versus 96.8% for the agent at 24% lower cost per task. On τ²-Bench, workflows reached 75.8% task success versus 80.6% for the agent while running in 1.3s instead of 170s; repair lifted k=2 workflows from 66.7% to 81.2%.

AI Brief

The paper offers a concrete method and measured trade-offs for cutting LLM-agent latency and cost by compiling repeated SOPs into deterministic workflows.

arXiv:2611.14492v1 Announce Type: new 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.
Be the first to rate this article.

Comments & Questions (0)

Captcha
Please be respectful — let's keep the conversation friendly.

No comments yet

Be the first to comment!