Unverified50% confidenceFactExact time
多Agent系统的思想起源于1980年代的分布式人工智能研究，在LLM时代每个Agent本质上是具有特定角色设定的LLM调用单元
1
Sources
50%
Confidence
Long-term
Relevance
7/25/2026
First Seen
Sources
Related Claims
Unverified在LLM时代，多Agent的含义已从传统基于规则的软件代理转变为由大型语言模型驱动的智能体协作71% similarVerifiedMulti-Agent System（多智能体系统）是当前AI领域最前沿的研究方向之一，不同专业Skills的Agent可以像企业职能部门一样分工协作68% similarUnverifiedSince 2023, projects like Stanford's Generative Agents, Microsoft's AutoGen, and Andrew Ng's Agentic Workflow have demonstrated that LLM-based agents can handle complex cognitive tasks including planning, execution, and reflection68% similarUnverifiedLLM兴起后，多智能体系统中的Agent从刚性规则执行器转变为具备自然语言理解与生成能力的柔性智能体，能够动态调用外部工具68% similarUnverified在大模型时代，Agent架构中LLM充当大脑负责推理决策，工具充当四肢负责与外部世界交互，记忆模块保存对话历史和中间状态67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/611290API
curl https://kongchang.com/api/v1/knowledge/claims/611290MCP
get_claim(id=611290)