Your First LangChain App

You have your environment, your library, and your API key. Now it's time to run the "Hello World" of the agentic age. We will initialize a chat model and ask it a simple question.

1. The Code

Create a file named hello_langchain.py:

from dotenv import load_dotenv
from langchain_openai import ChatOpenAI

# 1. Load your API key
load_dotenv()

# 2. Initialize the model
# We set temperature=0 for consistent, non-creative results
model = ChatOpenAI(model="gpt-4o-mini", temperature=0)

# 3. Invoke the model
response = model.invoke("Hello! What are you capable of?")

# 4. Print the result
print(response.content)

---

2. Breaking Down the Parameters

• model="gpt-4o-mini": We use the 'mini' size because it is faster and cheaper for learning.
• temperature=0: This controls "Randomness." 0 means the model will give the same answer to the same question every time. 1 means it will be very creative.
• .invoke(): This is the standard method to send a message to any LangChain component.

---

3. The Response Object

In LangChain, a model doesn't just return a string of text. It returns an AIMessage object.

• response.content: The actual text.
• response.response_metadata: Information about token usage and model flags.

---

4. Troubleshooting Common Errors

"Authentication Error"

• Cause: Your API key is wrong or your .env file isn't being loaded.
• Fix: Check that .env is in the same folder as your script.

"Model Not Found"

• Cause: You misspelled the model name.
• Fix: Use standard names like gpt-4o-mini or gpt-3.5-turbo.

---

Key Takeaways

• Initialization requires a model name and temperature.
• invoke() is the primary way to interact with models.
• The model returns a Response Object, not just text.
• Understanding the Temperature setting is the first step in control.