Unverified50% confidenceFactExact time
LangChain经历了从以链式调用（Chain）为核心到以智能体（Agent）为核心的范式转变
1
Sources
50%
Confidence
Long-term
Relevance
7/17/2026
First Seen
Sources
Related Claims
UnverifiedLangChain的核心设计理念是将大模型调用抽象为可组合的「链」（Chain），并引入ReAct（Reasoning + Acting）范式79% similarUnverifiedLangChain introduced two paradigms—chain invocation (Chain) and agents (Agent)—where the former is suited for fixed sequential process tasks and the latter is suited for open-ended tasks that dynamically infer the next step.77% similarVerifiedLangChain is based on the ReAct (Reasoning + Acting) think-act loop pattern and provides a general-purpose tool-chain orchestration framework.76% similarUnverifiedLangGraph 引入了「图」的抽象概念，将 Agent 工作流建模为节点和边，区别于传统 LangChain 的链式调用74% similarUnverifiedLangChain and CrewAI can be used to build custom Agent workflows.74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/541492API
curl https://kongchang.com/api/v1/knowledge/claims/541492MCP
get_claim(id=541492)