🚀 Agentic AI System using MCP + LangGraph + FastAPI

🧠 Overview

This project demonstrates a real-world Agentic AI system that goes beyond traditional chatbots.

Instead of only generating text responses, this system can:

🧠 Reason → 🧰 Choose tools → ⚙️ Execute actions → 📡 Return real-world results

It integrates MCP (Model Context Protocol) for tool standardization, LangGraph for intelligent agent orchestration, and FastAPI for deployment.

🎯 Why This Project?

Traditional AI models:

❌ Only respond with text

This system:

✅ Understands user intent ✅ Selects the right tool automatically ✅ Executes real-world tasks (email, search, etc.) ✅ Works like an autonomous assistant

⚙️ Key Features

🧠 Intelligent Agent

Dynamically decides which tool to use
Multi-step reasoning with LangGraph

🌐 Web Search Capability

Real-time Google search using SERPAPI
Returns structured results

📧 Email Automation

Sends emails using Yagmail SMTP
Fully automated response system

🔗 MCP Tool Integration

Standardized tool calling architecture
Easily extendable system

⚡ FastAPI Backend

REST API endpoint for AI agent
Production-ready structure

🏗 System Architecture

User Request
      ↓
FastAPI (main.py)
      ↓
Assistant Layer (assistant.py)
      ↓
LangGraph Agent (agent.py)
      ↓
MCP Tool Server (mcp_tool_server.py)
      ↓
External Tools
   ├── 🌐 SERPAPI (Search)
   ├── 📧 Yagmail (Email)
   └── ⚙️ Math Tools

🧰 Tech Stack

🐍 Python 3.10+
🧠 LangGraph (Agent orchestration)
🔗 MCP (Model Context Protocol)
⚡ FastAPI (Backend API)
📧 Yagmail (Email automation)
🌐 SERPAPI (Search API)
🔐 Python-dotenv (Environment management)

💡 What I Learned

This project helped me understand:

How real AI agents are structured in production
How tools are integrated into LLM-based systems
How MCP standardizes tool communication
How LangGraph manages decision-making workflows
How backend APIs expose AI capabilities

🚀 Future Improvements

🧠 Add long-term memory to the agent
💬 Add chat history persistence
📱 Integrate WhatsApp / Telegram bots
📄 Add PDF / file reading tools
☁️ Deploy on cloud (Render / AWS)
🖥 Build frontend UI chatbot

📌 Why This Project Matters

This project represents a shift from:

💬 “Chatbots that talk” to 🤖 “AI Agents that act”

It is a step toward autonomous AI systems capable of real-world actions.

🏷️ Tags

#AI #AgenticAI #LangGraph #MCP #FastAPI #Python #Automation #MachineLearning

⭐ If you like this project

Give it a ⭐ on GitHub and feel free to fork and improve it!

MCP Servers