AI Agent Chatbot

A multi-model conversational AI chatbot built with Streamlit, LangChain, LangGraph, and FastAPI. This project allows users to chat with various LLMs (Groq, Gemini, etc.) and enable advanced tools like Web Search, Wikipedia, and Python REPL.

✨ Features

Feature	Description
Multi-Model Support	Native support for Groq LLaMA3, Google Gemini, and more.
Advanced Tools	Integrated Web Search (Tavily), Wikipedia, and Python REPL.
Smart Orchestration	Powered by LangGraph for robust tool-use decision logic.
Optimized Performance	History sliding window & recursion limits for cost/speed efficiency.
Customization	Easily tailor system prompts and add custom tools in tools.py.

🛠️ Tech Stack

Tool/Library	Purpose
Python	Core logic and agent framework.
Streamlit	Sleek, interactive frontend for the chat interface.
FastAPI	High-performance backend API for model orchestration.
LangChain	The backbone for LLM integration and chaining.
LangGraph	State-of-the-graph logic for complex agentic workflows.
Docker	Containerization for easy deployment.

>_ Setup Instructions

1. Clone the Repository

git clone https://github.com/Balaji-R-05/ai-agent-chatbot.git
cd ai-agent-chatbot

2. Configure API Keys

Your .env should look like this:

# API Keys
GROQ_API_KEY=your_groq_key
TAVILY_API_KEY=your_tavily_key         # For Web Search

# Configuration
MAX_HISTORY=10
RECURSION_LIMIT=10                    # Max steps before the agent stops

3. Run with Docker (Recommended)

The easiest way to run the application is using Docker Compose:

docker-compose up --build

This will start both the backend (FastAPI) and the frontend (Streamlit).

4. Local Setup (Alternative)

Environment Setup

python -m venv venv
.\venv\Scripts\activate
pip install -r requirements.txt

Run the Application

Start both services using the batch script:

.\run.bat

Alternatively, start them separately:

# Terminal 1: Backend
uvicorn main:app --port 8000

# Terminal 2: Frontend
streamlit run client/app.py

---

• API Documentation: http://127.0.0.1:8000/doc
• Streamlit App: http://127.0.0.1:8501

---

Developed by Balaji-R-05