研究人员开发了 AutoRPA，一个将基于 LLM 的代理的决策逻辑转换为高效机器人流程自动化 (RPA) 功能的框架。该方法解决了为重复性 GUI 任务反复调用 LLM 推理的低效问题。AutoRPA 利用翻译器-构建器管道和混合修复策略来合成健壮的 RPA 功能，显著提高了运行时效率和可重用性，同时大幅减少了 token 使用量。 AI
影响 通过将 LLM 决策逻辑转换为高效 RPA 来自动化重复性 GUI 任务，减少了 token 使用量并提高了运行时效率。
排序理由 该集群描述了一篇详细介绍 LLM 驱动的代码合成新颖框架的新研究论文。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
arXiv:2605.21082v1 Announce Type: new Abstract: Large Language Model (LLM) based agents have demonstrated proficiency in multi-step interactions with graphical user interfaces (GUIs). While most research focuses on improving single-task performance, practical scenarios often invo…
Large Language Model (LLM) based agents have demonstrated proficiency in multi-step interactions with graphical user interfaces (GUIs). While most research focuses on improving single-task performance, practical scenarios often involve repetitive GUI tasks for which invoking LLM …
Large Language Model (LLM) based agents have demonstrated proficiency in multi-step interactions with graphical user interfaces (GUIs). While most research focuses on improving single-task performance, practical scenarios often involve repetitive GUI tasks for which invoking LLM …