From Direct API Calls to LangChain Abstraction

Beyond the Raw Request

When starting with Large Language Models (LLMs), developers typically use direct API calls (like the OpenAI Python library) to send a prompt and receive a completion. While functional, this approach becomes unmanageable as applications scale.

The Problem of Statelessness

Large Language Models are inherently stateless. Every time you send a message, the model "forgets" who you are and what you previously said. Each interaction is a blank slate. To maintain a conversation, you must manually pass the entire history back to the model every single time.

The LangChain Solution

LangChain introduces the ChatOpenAI model wrapper. This isn't just a wrapper for the sake of it—it is the foundation for modularity. By abstracting the model call, we can later swap models, inject memory, and use templates without rewriting our entire codebase.

The Pirate Scenario

Imagine a customer email written in "Pirate-style" slang. To translate this into a formal corporate response, a direct API call requires hardcoding the instructions. With LangChain, we separate the "Style" (Pirate vs. Formal) from the "Content" (The Email) using abstraction.

TERMINAL bash — 80x24

> Ready. Click "Run" to execute.

Question 1

Why do we say LLMs are "stateless"?

They do not have access to the internet.

They cannot generate the same response twice.

They do not inherently remember previous messages in a conversation.

They are only capable of processing text, not data states.

Challenge: Initialize ChatOpenAI

Solve the problem below.

You are building a creative writing assistant and need to initialize your first LangChain model.

Your task is to create a ChatOpenAI instance named my_llm with a temperature of 0.7 to allow for more creative (non-deterministic) responses.

Task

Write the Python code to import and initialize the model.

Solution:
from langchain_openai import ChatOpenAI
my_llm = ChatOpenAI(temperature=0.7)