[用Portkey增强Langchain：高效构建与管理AI应用的终极指南]-优快云博客

本文链接：https://blog.youkuaiyun.com/tt_jishu/article/details/142857035

# 用Portkey增强Langchain：高效构建与管理AI应用的终极指南

## 引言

在构建AI驱动的应用程序时，开发者通常需要面对模型选择、API管理、性能监控等多重挑战。Portkey作为AI应用的控制面板，通过其AI Gateway和可观测性套件，帮助数百个团队构建可靠、高效、快速的应用。本文将探讨如何通过Portkey与Langchain集成，实现生产级的AI应用准备。

## 主要内容

### 1. 通用API接入

Portkey支持通过一个统一的API与150多个模型连接。只需更新Langchain中的`ChatOpenAI`模型，即可轻松切换不同的AI提供商。以下是实现步骤：

1. **安装Portkey SDK**

   ```bash
   pip install -U portkey_ai

连接Portkey AI Gateway

from langchain_openai import ChatOpenAI
from portkey_ai import createHeaders, PORTKEY_GATEWAY_URL

PORTKEY_API_KEY = "..." # 不需要时可以自行托管网关
PROVIDER_API_KEY = "..." # AI提供商的API密钥

portkey_headers = createHeaders(api_key=PORTKEY_API_KEY, provider="openai")

llm = ChatOpenAI(api_key=PROVIDER_API_KEY, base_url=PORTKEY_GATEWAY_URL, default_headers=portkey_headers)

response = llm.invoke("What is the meaning of life, universe and everything?")

2. 高级路由控制

Portkey AI Gateway提供负载均衡、回退机制及自动重试功能。以下配置展示了如何在gpt-4和claude-3-opus-20240229之间进行负载均衡：

config = {
    "strategy": {
         "mode": "loadbalance"
    },
    "targets": [{
        "virtual_key": "openai-25654",
        "override_params": {"model": "gpt4"},
        "weight": 0.5
    }, {
        "virtual_key": "anthropic-25654",
        "override_params": {"model": "claude-3-opus-20240229"},
        "weight": 0.5
    }]
}

portkey_headers = createHeaders(api_key=PORTKEY_API_KEY, config=config)

llm = ChatOpenAI(api_key="X", base_url=PORTKEY_GATEWAY_URL, default_headers=portkey_headers)
response = llm.invoke("What is the meaning of life, universe and everything?")

3. 完整的可观测性与代理追踪

Portkey为Langchain集成提供了完整的可视化追踪能力。通过简单的修改，可以在Portkey的仪表盘上查看请求日志。

from langchain import hub  
from langchain.agents import AgentExecutor, create_openai_tools_agent  
from langchain_openai import ChatOpenAI
from portkey_ai import PORTKEY_GATEWAY_URL, createHeaders

prompt = hub.pull("hwchase17/openai-tools-agent")

portkey_headers = createHeaders(
    api_key=PORTKEY_API_KEY,
    virtual_key="OPENAI_VIRTUAL_KEY",
    trace_id="uuid-uuid-uuid-uuid"
)

@tool
def multiply(first_int: int, second_int: int) -> int:
    return first_int * second_int

tools = [multiply]

model = ChatOpenAI(api_key="X", base_url=PORTKEY_GATEWAY_URL, default_headers=portkey_headers, temperature=0)
agent = create_openai_tools_agent(model, tools, prompt)
agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)

response = agent_executor.invoke({"input": "Multiply 3 by 36"})