ChatPromptTemplate for conversations in LangChain - Step-by-Step Execution

Concept Flow - ChatPromptTemplate for conversations

Define Template Variables

↓

Create ChatPromptTemplate

↓

Input User Data

↓

Format Prompt with Variables

↓

Send Prompt to Language Model

↓

Receive and Use Response

This flow shows how you define variables, create a ChatPromptTemplate, input data, format the prompt, send it to the model, and get a response.

Execution Sample

LangChain

from langchain.prompts import ChatPromptTemplate, SystemMessagePromptTemplate, HumanMessagePromptTemplate

chat_prompt = ChatPromptTemplate.from_messages([
    SystemMessagePromptTemplate.from_template("You are a helpful assistant."),
    HumanMessagePromptTemplate.from_template("Hello, {name}!")
])

formatted = chat_prompt.format_prompt(name="Alice")
print(formatted.to_messages())

This code creates a chat prompt template with a system and human message, fills in the name variable, and prints the formatted messages.

Execution Table

Step	Action	Input/Variable	Result/Output
1	Define system message template	Template: 'You are a helpful assistant.'	SystemMessagePromptTemplate created
2	Define human message template	Template: 'Hello, {name}!'	HumanMessagePromptTemplate created
3	Create ChatPromptTemplate	Messages: [system, human]	ChatPromptTemplate instance created
4	Format prompt with name='Alice'	name='Alice'	Prompt filled with 'Hello, Alice!'
5	Convert formatted prompt to messages	Formatted prompt	List of messages ready to send to model
6	Print messages	Messages list	[SystemMessage('You are a helpful assistant.'), HumanMessage('Hello, Alice!')]

💡 All steps complete, prompt formatted and ready for model input

Variable Tracker

Variable	Start	After Step 4	Final
chat_prompt	None	ChatPromptTemplate instance	ChatPromptTemplate instance
name	None	"Alice"	"Alice"
formatted	None	FormattedPrompt with messages	FormattedPrompt with messages
messages	None	None	[SystemMessage, HumanMessage with 'Alice']

Key Moments - 3 Insights

Why do we use {name} inside the human message template?

The {name} is a placeholder variable that gets replaced with actual data ('Alice') during formatting, as shown in step 4 of the execution_table.

What does format_prompt() do exactly?

format_prompt() fills all template variables with provided values and prepares the messages for the language model, as seen in step 4 and 5.

Why do we convert the formatted prompt to messages before sending?

The language model expects a list of message objects, so formatted.to_messages() converts the filled template into that format, shown in step 5.

Visual Quiz - 3 Questions

Look at the execution_table, what is the output after step 4?

A ChatPromptTemplate instance created

B Prompt filled with 'Hello, Alice!'

C List of messages ready to send

D SystemMessagePromptTemplate created

💡 Hint

Check the 'Result/Output' column for step 4 in the execution_table

At which step do we create the ChatPromptTemplate instance?

A Step 3

B Step 2

C Step 4

D Step 5

💡 Hint

Look for 'ChatPromptTemplate instance created' in the execution_table

If we change the name variable to 'Bob', what changes in the execution_table?

A Step 5 no longer converts to messages

B Step 3 creates a different template

C Step 4 output changes to 'Hello, Bob!'

D Step 1 system message changes

💡 Hint

Variable 'name' affects the formatted prompt output at step 4

Concept Snapshot

ChatPromptTemplate lets you create chat message templates with placeholders.
Use from_messages() to combine system and human templates.
Fill variables with format_prompt() to get ready-to-send messages.
Convert with to_messages() before sending to the language model.
This helps keep conversations dynamic and reusable.

Full Transcript

This visual execution trace shows how to use ChatPromptTemplate in langchain for conversations. First, you define message templates with placeholders like {name}. Then you create a ChatPromptTemplate combining these messages. When you have actual data, you call format_prompt() with variables to fill placeholders. This produces a formatted prompt object. You convert it to a list of messages with to_messages(), which you can send to a language model. The execution table walks through each step, showing how templates become ready messages. The variable tracker shows how variables like 'name' change from None to 'Alice'. Key moments clarify why placeholders are used and how formatting works. The quiz tests understanding of each step and variable effects. This method keeps chat prompts flexible and easy to reuse.