🛠️ LangChain ReAct Agent: Using Tools to Answer Questions

LangChain's ReAct agent framework allows language models to make decisions and interact with external tools. In this post, we'll walk through how to create a simple agent that uses a custom tool to answer the question: “What time is it?”

🧠 What’s the ReAct Pattern?

ReAct stands for Reasoning + Acting. It enables an agent to:

Think step by step
Use tools when needed
Return a final answer

🔧 What We’re Building

A custom tool to get the current time
A ReAct agent using OpenAI’s GPT-4o
An executor that ties it all together

1from dotenv import load_dotenv
2from langchain import hub
3from langchain.agents import AgentExecutor, create_react_agent
4from langchain_core.tools import Tool
5from langchain_openai import ChatOpenAI
6
7load_dotenv()
8
9def get_current_time(*args, **kwargs):
10   """Returns the current time in H:MM AM/PM format."""
11  import datetime
12  now = datetime.datetime.now()
13  return now.strftime("%I:%M %p")
14
15tools = [
16  Tool(
17      name="Time",
18      func=get_current_time,
19      description="Useful for when you need to know the current time",
20  ),
21]
22
23# Pull the prompt template from the hub
24prompt = hub.pull("hwchase17/react")
25
26llm = ChatOpenAI(model="gpt-4o", temperature=0)
27
28agent = create_react_agent(
29  llm=llm,
30  tools=tools,
31  prompt=prompt,
32  stop_sequence=True,
33)
34
35agent_executor = AgentExecutor.from_agent_and_tools(
36  agent=agent,
37  tools=tools,
38  verbose=True,
39)
40
41response = agent_executor.invo
42print(response)

🧱 How It Works

1. Define the Tool

This tool just returns the current time using Python's datetime module:

1def get_current_time():
2   ...

2. Register the Tool

1Tool(name="Time", func=get_current_time, description="Useful for when you need to know the current time")

3. Use ReAct Prompt

We use a ReAct-style prompt from LangChain’s hub:

1prompt = hub.pull("hwchase17/react")

This allows the LLM to reason and decide whether to use a tool.

4. Execute with AgentExecutor

We create an agent executor that manages the interaction:

1agent_executor = AgentExecutor.from_agent_and_tools(...)

✅ Example Output

When asked:

"What time is it?"

the agent might respond:

The current time is 03:42 PM.

🚀 Why Use ReAct + Tools?

Extend LLMs with real-time data
Support tool use for calculations, APIs, or databases
Create autonomous agents with decision-making skills

← Previous: Simple Agent with Tool

Next: ReAct Agent with Memory →