Skip to main content

Prompt engineering with LangChain

In this tutorial we'll use LangChain to call OpenAI from a Modelbit deployment.

Add your API key to Modelbit​

Make an OpenAI API Key to load into Modelbit. In Modelbit's Settings, click Integrations and then click the OpenAI tile. Add your OpenAI API key to this form.

Install LangChain​

We'll develop this deployment in a Python notebook. Make sure LangChain is installed, authenticate your Python notebook with Modelbit, and set your API key for local development:

pip install langchain langchain_openai

Authenticate with Modelbit:

import modelbit
mb = modelbit.login()

And set your API key in the notebook:

import os
os.environ['OPENAI_API_KEY'] = mb.get_secret("OPENAI_API_KEY")

Create a LangChain chat prompt​

We'll make a simple chatbot to assist with customer support:

from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser

def chat_with_langchain(input: str) -> str:
    llm = ChatOpenAI()
    prompt = ChatPromptTemplate.from_messages([
        ("system", "You are a tech support agent at a financial services company."),
        ("user", "{input}")
    ])
    output_parser = StrOutputParser()
    chain = prompt | llm | output_parser
    return chain.invoke({"input": input})

Let's test that our function works as expected:

chat_with_langchain("Why won't you give me a refund?")

Success, it responds with:

I'm sorry to hear that you're experiencing difficulties with our refund process. As
a tech support agent, I...

Deploy to Modelbit​

We can now deploy chat_with_langchain to Modelbit:

mb.deploy(chat_with_langchain)

This will create a REST API and a SQL function that uses LangChain to call OpenAI.