🧮 LangChain Math Assistant

A simple AI-powered math assistant using LangChain and OpenAI, capable of handling basic arithmetic operations and natural conversation.

Features

• Conversational AI: Chat with the assistant in plain English.
• Math Tools: Perform addition, subtraction, multiplication, division, square roots, exponents, and more.
• Real-Time Streaming: See responses as they are generated.
• OpenAI Integration: Powered by OpenAI’s API via LangChain for flexible language understanding.
• Environment Variables: Secure usage of API keys with .env file.

---

Table of Contents

• How It Works
• Quick Start
• Adding More Math Tools
• Understanding the Components
• Lowering OpenAI Costs
• FAQ / Further Reading

---

How It Works

• The assistant listens to your input.
• It uses LangChain agents to reason about your question.
• If your input matches a tool (like math operations), it calls the corresponding Python function.
• The result is streamed to your terminal in real time.

---

Quick Start

1. Clone the repository:

   git clone https://github.com/rohanpandavv/MathBuddy.git
   cd MathBuddy

2. Install uv (if you don’t have it):

   pip install uv

3. Install dependencies with uv:

   uv pip install -r requirements.txt

4. Set up your .env file:

   OPENAI_API_KEY=sk-xxxxxx
   

5. Activate the virtual environment:

   source .venv/bin/activate  

   or

   venv\Scripts\activate.bat

6. Run the assistant:

   python main.py

7. Example Usage:

   You: What is 12 times 9?
   Assistant: The product of 12 and 9 is 108
   
   You: Square root of 16?
   Assistant: The square root of 16 is 4.0
   
   You: quit
   

---

Adding More Math Tools

You can add any math operation by creating a new function and decorating it with @tool.
Example:

import math
@tool
def factorial(a: float) -> str:
    """Useful for finding out the factorial of a number."""
    return f"The factorial of {a} is {math.factorial(a)}"

Add your newly created tools to the tools list.

tools = [addition, subtraction, division, multiplication, remainder, root, factorial]

---

Understanding the Components

• ChatOpenAI: A wrapper to access OpenAI’s language models. You can set the temperature parameter (randomness) and model (e.g., gpt-3.5-turbo for lower cost).

• HumanMessage: Used to represent user input as part of a message sequence for the agent.

• @tool Decorator: Converts a Python function into a “tool” that the agent can use, guided by the function’s docstring.

• create_react_agent: Instantiates a LangChain ReAct agent which can REason and ACT (call tools) based on user input.

• agent_executor.stream: Streams the assistant’s response in real time, providing a conversational experience.

---

Tips

• Keep prompts concise.
• Don't add unnecessary history to the prompt.
• NEVER PUSH .ENV FILES ON ANY VERSION CONTROL
• MAKE SURE YOU SAFEGUARD YOUR OPENAI SECRET KEY