Unverified90% confidenceFactTime unknown
在MCP出现之前，大模型与外部工具的交互主要依赖各家厂商自定义的Function Calling机制，OpenAI和Google Gemini各有不同的函数调用格式，互不兼容
1
Sources
90%
Confidence
Long-term
Relevance
6/3/2026
First Seen
Sources
LangGraph+MCP实战：智能体与MCP协议核心原理深度解析
bilibili码士集团-踢踢
Related Entities
Related Claims
Unverified在 MCP 出现之前，大语言模型与外部工具集成方式高度碎片化，包括 OpenAI 的 Function Calling、LangChain 的 Tool 体系及各厂商插件机制互不兼容85% similarVerified在MCP出现之前，AI调用外部工具主要依赖OpenAI的Function Calling机制或各框架自定义的插件系统80% similarUnverified在MCP出现前存在LangChain Tools、OpenAI Function Calling等多套互不兼容的集成规范，造成生态碎片化77% similarUnverifiedMCP出现之前，业界常用OpenAI提出的Function Calling机制，但该机制是厂商特定实现，工具集成面临N×M的复杂度问题；MCP将集成复杂度从N×M简化为N+M75% similarUnverifiedBefore MCP, the industry had Function Calling mechanisms but they lacked cross-platform standardization, with each AI platform implementing them differently74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/38148API
curl https://kongchang.com/api/v1/knowledge/claims/38148MCP
get_claim(id=38148)