业务侧 RAG+Agent 落地实践：基于 LangChain 构建简易工具调用 Agent

from langchain_openai import ChatOpenAI
from langchain.agents import AgentExecutor, create_openai_tools_agent
from langchain_core.tools import tool

# 复用上文已经构建好的vectordb向量库
retriever = vectordb.as_retriever(search_kwargs={"k":3})

@tool
def company_knowledge_search(query:str)->str:
    """用于查询企业内部制度知识库，仅业务相关问题调用此工具。入参为用户问题。"""
    docs = retriever.invoke(query)
    return "\n".join([d.page_content for d in docs])

tools = [company_knowledge_search]
llm = ChatOpenAI(model="gpt‑4o‑mini", temperature=0, api_key="your-api-key")

agent = create_openai_tools_agent(llm, tools)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True, max_iterations=3)

resp = agent_executor.invoke({"input":"公司的外勤报销流程是什么？"})
print(resp["output"])