Google 论文:把工作流写成可编辑程序图,24 组评测 21 组拿第一
rohanpaul_ai · x · 2026-09-21
A new Google paper introduces Procedural Graphs: keeping an LLM agent's long-task workflow in an editable procedure graph instead of burying it in chat history.
Problem: agents running long tasks forget where they are, repeat tool calls, or execute steps out of order.
Method:
- The graph gives the agent a small map of possible next steps while leaving the LLM free to reason
- After runs, another LLM compares successes and failures and edits the map — edits are kept only if they don't hurt held-out tasks
Results:
- Ranked 1st or tied 1st in 21 of 24 model-benchmark combinations
- Repaired a flawed human-designed workflow: MultiChallenge success went from 58.93% to 92.86% after refinement
Caveats: the guidance costs extra tokens, so it fits cases where long workflows are the bottleneck. Core recommendation: move critical procedures out of chat history into an explicit, execution-improving workflow.
所属事件:谷歌等提出程序图方法提升长任务智能体表现(2 条相关)→
「编程与Agent」频道最新
- 两人小公司用 46 个 AI Agent 运营：审批门禁、哈希锁文件与翻车复盘 — dkovl · 2026-11-12
- LlamaIndex 让 AI Agent 渗透测试自家 LlamaParse，四天暴露四条经验 — llama_index · 2026-11-12
- Manhattan 字体 v0.0.1：每个字母都是一个 Python 函数 — allanyu_ · 2026-11-12
- 用 Backplane「提示词造出」任意扩展卡,网友呼吁扩展口回归标配 — _Stocko_ · 2026-11-12
- Agent 会话跑了 51 分钟烧 24.8 万 token,网友直呼还在干活 — _Stocko_ · 2026-11-12
- 可视化 AI Agent 团队协作：每个 Agent 一间「房间」，共享知识库 — LiorAloni11 · 2026-11-12