Welcome to Day 3! Yesterday, we built our first chain. It worked, but we did something that professional AI engineers try to avoid: we used a simple string for our prompt.
In a real-world app, prompts need to be dynamic, reusable, and structured. Today, we’re diving into Prompt Templates—the secret to making your AI reliable and scalable.
🛑 The Problem: Hardcoded Strings
Imagine you’re building a travel assistant. If you write your prompt like this:
"Tell me a 3-day itinerary for Paris"
...your code is stuck. Every time the city changes, you have to rewrite the code.
Prompt Templates solve this by using placeholders (like {city}), turning your instructions into a reusable function.
🏛️ The Two Main Types of Templates
The official documentation breaks prompts into two categories based on the model you're using.
1. PromptTemplate (Standard)
Best for "Completion" models that just want a single block of text.
from langchain_core.prompts import PromptTemplate
template = PromptTemplate.from_template("You are a helpful assistant. Explain {topic} to a 5-year-old.")
# We can reuse this for 'Space', 'Economics', or 'Cooking'!
2. ChatPromptTemplate (The Gold Standard)
Most modern models (GPT-4o, Claude, Gemini) are Chat Models. They don't just want text; they want a conversation history with specific Roles.
from langchain_core.prompts import ChatPromptTemplate
chat_template = ChatPromptTemplate.from_messages([
("system", "You are a professional {industry} consultant."),
("human", "How can I improve my {business_process}?"),
])
# When we run this, LangChain formats it perfectly for the AI
formatted_prompt = chat_template.invoke({
"industry": "Real Estate",
"business_process": "lead generation"
})
🎭 Understanding the "Roles"
Why bother with "System" or "Human" tags?
- System: Sets the "vibe" or rules (e.g., "Don't use emojis," "Answer in bullet points").
- Human: The actual user input.
- AI (or Assistant): You can even provide "example" responses from the AI to guide its behavior (this is called Few-Shot Prompting).
⚡ Pro Tip: Partial Formatting
Sometimes you know some information early, but not all of it. For example, your System Prompt might always include the current date, but the user's question comes later.
# Create a template with a fixed 'name' but a dynamic 'question'
partial_template = chat_template.partial(industry="Tech Operations")
# Now you only need to provide the 'business_process' later!
final_chain = partial_template | model
This keeps your code incredibly clean and modular.
🎯 Day 3 Summary
Today, we upgraded from "shouting at a bot" to "designing a system." You learned:
Why templates are better than strings.
The difference between PromptTemplate and ChatPromptTemplate.
How to use Roles to control AI personality.
How to Partial a prompt to save time.
Your Homework: Create a ChatPromptTemplate for a "Code Reviewer" agent. Give it a System role that tells it to be "strict but encouraging," and a Human role for the code snippet.
See you tomorrow! ☕
Top comments (0)