ReAct agent implementation in LangChain - Mini Project: Build & Apply

ReAct Agent Implementation with Langchain

📖 Scenario: You are building a smart assistant that can think and act step-by-step to answer questions by using tools and reasoning. This is called a ReAct agent.We will use the Langchain library to create this agent that can reason and call tools to get information.

🎯 Goal: Create a ReAct agent using Langchain that can use a search tool and a calculator tool to answer questions step-by-step.

📋 What You'll Learn

Create a list of tools with a search tool and a calculator tool

Set up the OpenAI LLM with temperature 0

Create a ReAct agent using Langchain's initialize_agent function

Run the agent on a sample question to see the step-by-step reasoning

💡 Why This Matters

🌍 Real World

ReAct agents help build smart assistants that can reason and use tools to answer complex questions, useful in customer support, research, and automation.

💼 Career

Understanding how to implement ReAct agents with Langchain is valuable for AI developers, chatbot engineers, and software developers working with intelligent systems.

1

Set up the tools list

Create a list called tools containing Tool(name='Search', func=search.run, description='useful for answering questions about current events') and Tool(name='Calculator', func=calculator.run, description='useful for math calculations'). Assume search and calculator are already imported and initialized.

LangChain

# Your code here to create the tools list

Need a hint?

Use a Python list with two Tool objects exactly as described.

2

Configure the OpenAI LLM

Create a variable called llm and assign it to OpenAI(temperature=0) to set up the language model with no randomness.

LangChain

tools = [
    Tool(name='Search', func=search.run, description='useful for answering questions about current events'),
    Tool(name='Calculator', func=calculator.run, description='useful for math calculations')
]
# Your code here to create the llm variable

Need a hint?

Use the OpenAI class with temperature set to 0.

3

Create the ReAct agent

Create a variable called agent and assign it to the result of initialize_agent(tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION, verbose=True) to build the ReAct agent with verbose output.

LangChain

tools = [
    Tool(name='Search', func=search.run, description='useful for answering questions about current events'),
    Tool(name='Calculator', func=calculator.run, description='useful for math calculations')
]

llm = OpenAI(temperature=0)
# Your code here to create the agent variable

Need a hint?

Use initialize_agent with the tools list, llm, AgentType.ZERO_SHOT_REACT_DESCRIPTION, and verbose=True.

4

Run the agent on a question

Call agent.run with the string 'What is the square root of the population of New York City?' to see the agent's step-by-step reasoning and tool usage.

LangChain

tools = [
    Tool(name='Search', func=search.run, description='useful for answering questions about current events'),
    Tool(name='Calculator', func=calculator.run, description='useful for math calculations')
]

llm = OpenAI(temperature=0)

agent = initialize_agent(tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION, verbose=True)
# Your code here to run the agent on the question

Need a hint?

Call agent.run with the exact question string.