arXiv：cs.CL· Irene Weber (University of Applied Sciences Kempten, Germany)·· 27 天前
从 LLM 聊天到自主 AI Agent：LLM 集成应用的七种形态综述
Forms of LLM-Integrated Applications from LLM-Chats to Autonomous AI Agent System
AI 导读
一项综述系统梳理了 LLM 集成应用的七种常见形态：LLM 聊天、自定义 Agent、RAG、AI 增强工作流、Copilot、编码 Agent 以及部分 agentic RAG，并从架构模式、执行控制与用户介入点、每任务 Agent 调用次数、工具使用四个维度加以刻画。
正文 · 原文
Authors:Irene Weber (University of Applied Sciences Kempten, Germany)
Abstract:Large language models (LLMs) are increasingly embedded as components in software systems, marketed under labels such as chatbot, copilot, retrieval-augmented generation, workflow, coding agent and AI agent. Whether these labels denote genuine architectural forms or serve as branding has not been assessed systematically.
In the sources surveyed, labels do carry architectural content, most clearly in vendor usage: copilot denotes a router-worker architecture operating a host application under step-by-step user confirmation, while the more recent shift to the label agent coincides with AI-planned multi-step execution of which the user sees only the outcome. The coding agents of four major providers share one architecture, a reason-and-act loop delegating to subagents.
This survey describes seven recurring forms---LLM chats, custom agents, retrieval-augmented generation (RAG), AI-enhanced workflows, copilots, coding agents, and, in part, agentic RAG---in a common vocabulary of agents and tools. Each is characterized along four structural dimensions (agentic RAG only partially): the architectural pattern, the control of execution and the point of user intervention, the number of agent calls per task, and tool use. An illustrative corpus of 22 systems from research publications and vendor documentation grounds the descriptions and shows where they reach their limit.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Software Engineering (cs.SE) |
| ACM classes: | I.2.11; D.2.11; I.2.7 |
| Cite as: | arXiv:2610.11899 [cs.CL] |
| (or arXiv:2610.11899v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.11899 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Irene Weber [view email]
[v1]
Thu, 8 Oct 2026 13:05:33 UTC (101 KB)
来源：arXiv：cs.CL · arxiv.org