AI Agents Development
From 0 to production

# react_agent.py — a basic ReAct Agent with LangChain
# from the AI Agents Development course | Automation4MI

from langchain.agents import create_react_agent, AgentExecutor
from langchain_openai import ChatOpenAI
from langchain.tools import tool
from langchain import hub


# ---------------------------------------------------------
# Defining Tools — the heart of every Agent
# each @tool turns a regular Python function into a Tool the LLM can call
# ---------------------------------------------------------

@tool
def search_web(query: str) -> str:
    """Search the web for current information."""
    # in the course you connect here to the Tavily Search API or SerpAPI
    # for demo purposes we return a mock result
    return f"Search results for: {query} — [data from the API]"


@tool
def calculate(expression: str) -> str:
    """Evaluate a mathematical expression safely."""
    # eval with empty builtins — prevents running malicious code
    # important tip: do not use eval without restriction in production!
    try:
        result = eval(expression, {"__builtins__": {}}, {})
        return str(result)
    except Exception as e:
        return f"Error: {e}"


# ---------------------------------------------------------
# Defining the LLM — the Agent brain
# temperature=0 for consistent and predictable results
# ---------------------------------------------------------

llm = ChatOpenAI(model="gpt-4o", temperature=0)
tools = [search_web, calculate]

# load a ready-made Prompt from the LangChain Hub — includes a ready ReAct format
# hwchase17/react includes the Thought/Action/Observation template
prompt = hub.pull("hwchase17/react")


# ---------------------------------------------------------
# Building the Agent
# create_react_agent = connects LLM + Tools + Prompt into an Agent object
# AgentExecutor = runs the Agent in a loop until a final answer
# ---------------------------------------------------------

agent = create_react_agent(llm, tools, prompt)
executor = AgentExecutor(
    agent=agent,
    tools=tools,
    verbose=True,      # verbose=True prints every Thought/Action — useful for debugging
    max_iterations=5   # limits infinite loops — always recommended!
)

# ---------------------------------------------------------
# Running — a question that requires two tools: search + calculation
# ---------------------------------------------------------

result = executor.invoke({
    "input": "What is the current market cap of Nvidia and what is 15% of it?"
})
print(result["output"])