Appendix A. Trying out LangChain

This appendix offers a quick, hands-on introduction to experimenting with LangChain in a Jupyter Notebook. It focuses on getting started fast with OpenAI’s REST API as the default backend, while noting that local, open‑source engines like Ollama can be explored later. The walkthrough centers on simple sentence completions and basic prompt engineering, highlighting how LangChain’s abstractions make it easy to prototype LLM-driven tasks with minimal setup and cost (using a small GPT model for affordability).

Readers are guided through prerequisites—Python 3.11+, an OpenAI API key, and a working Jupyter environment—followed by practical steps to create a virtual environment, install dependencies, and launch a notebook. After importing the necessary LangChain components, the appendix shows how to securely provide an API key (via prompt or environment variable), instantiate a ChatOpenAI client with a budget-friendly model, and issue a simple completion prompt. The immediate result demonstrates how LangChain returns structured messages that capture the LLM’s output and metadata, confirming the end‑to‑end setup is working.

With the basics in place, the appendix illustrates prompt engineering by moving from a direct prompt to a reusable PromptTemplate that parameterizes tone and length, then applies it to a short summarization task. It introduces chains and the LangChain Expression Language (LCEL), composing components with a pipeline operator to create readable, declarative workflows (for example, a prompt template feeding directly into an LLM). By showing how components implement a common runnable interface, the appendix emphasizes how LCEL simplifies building, reusing, and scaling prompt‑driven pipelines for common LLM applications.

FAQ

C:\>cd Github\building-llm-applications\ch01
C:\Github\building-llm-applications\ch01>python -m venv env_ch01
C:\Github\building-llm-applications\ch01>.\env_ch01\Scripts\activate

(env_ch01) C:\Github\building-llm-applications\ch01>pip install -r requirements.txt

(env_ch01)> jupyter notebook

(env_ch01)> jupyter notebook 01-langchain_examples.ipynb

import getpass
OPENAI_API_KEY = getpass.getpass("Enter your OPENAI_API_KEY")

set OPENAI_API_KEY=your_key

import os
api_key = os.getenv("OPENAI_API_KEY")

from langchain_openai import ChatOpenAI
llm = ChatOpenAI(openai_api_key=OPENAI_API_KEY, model_name="gpt-5-nano")
llm.invoke("It's a hot day, I would like to go to the…")

prompt_input = "Write a short message to remind users to be vigilant about phishing attacks."
response = llm.invoke(prompt_input)
print(response.content)

from langchain_core.prompts import PromptTemplate
prompt_template = PromptTemplate.from_template(
    "You are an experienced copywriter. Write a {num_words} words summary "
    "of the following text, using a {tone} tone: {text}"
)
prompt_input = prompt_template.format(text=some_text, num_words=20, tone="knowledgeable and engaging")
print(llm.invoke(prompt_input).content)

chain = prompt_template | llm
response = chain.invoke({"text": some_text, "num_words": 20, "tone": "knowledgeable and engaging"})
print(response.content)