🤖 Agentic AI Tutorial: A Comprehensive Guide

Welcome to the Agentic AI Tutorial! This repository is your ultimate, hands-on guide to mastering the world of Autonomous Agents. We go beyond simple chat interfaces to build systems that can reason, plan, and execute actions using state-of-the-art Large Language Models (LLMs).

🌟 Why Agentic AI?

Traditional AI responds to prompts. Agentic AI takes it a step further:

Autonomy: It decides which tools to use and how to solve a problem.
Reasoning: It breaks down complex tasks into manageable steps.
Persistence: It maintains state and memory over long interactions.
Action: It interacts with the real world (APIs, databases, files).

🗺️ Learning Roadmap

Chapter	Level	Focus Area	Status
Chapter 1	🟢 Beginner	LLM Fundamentals, Providers (Ollama/OpenAI/Gemini)	✅ Complete
Chapter 2	🔵 Intermediate	LangChain Orchestration, LCEL, Chains & Tools	✅ Complete
Chapter 3	🔵 Intermediate	Memory Systems, Entity Tracking & RAG	✅ Complete
Chapter 4	🟠 Advanced	Autonomous Agents & LangGraph Patterns	✅ Complete
Chapter 5	🔴 Expert	Multi-Node Agents & MCP Server Integration	✅ Complete

🛠️ Core Tech Stack

Frameworks: LangChain, LangGraph
Protocols: Model Context Protocol (MCP) by Anthropic
Models: OpenAI (GPT-4o, GPT-3.5), Google Gemini (2.0 Flash), Ollama (Local)
Vector DB: Chroma, FAISS
Embeddings: Sentence Transformers (HuggingFace)

🚀 Quick Start

1. Prerequisites

Python 3.8 or higher.
API Keys for OpenAI/Google (optional if using Ollama exclusively).

2. Installation

Choose your preferred method:

SSH

git clone [email protected]:zkzkGamal/Agentic-AI-Tutorial.git
cd Agentic-AI-Tutorial

HTTPS

git clone https://github.com/zkzkGamal/Agentic-AI-Tutorial.git
cd Agentic-AI-Tutorial

3. Environment Setup

We recommend using a virtual environment for each chapter or a global one for the project.

# Create & Activate
python3 -m venv venv
source venv/bin/activate  # Linux/macOS
# OR: venv\Scripts\activate  # Windows

# Install Base Dependencies
pip install -r requirements.txt

4. Configuration

Each chapter contains its own .env.example. Copy it to .env and fill in your keys.

# Example for Chapter 1
cp Chapter1/.env.example Chapter1/.env

📚 Deep Dives

Chapter 1: LLM Fundamentals

Direct API calls to OpenAI, Gemini, and Ollama.
Streaming techniques.
System prompt engineering (Personas).

Chapter 2: LangChain Orchestration

Mastering LCEL (LangChain Expression Language).
Building sequential and router chains.
Binding and calling external tools.

Chapter 3: Memory & Context

ConversationBufferMemory for full history.
ConversationEntityMemory for fact extraction.
RAG (Retrieval-Augmented Generation) with local vector stores.

Chapter 4: Autonomous Agents

LangGraph StateGraph fundamentals.
ReAct, Router, and Sequential Pipeline patterns.
Multi-Agent Collaboration and Self-Refine loops.
Human-in-the-Loop for production safety.

Chapter 5: Multi-Node LangGraph & MCP

Building a decoupled architecture using the Model Context Protocol (MCP).
Deploying a local FastMCP Server with Mail and Math tools.
Routing requests intelligently across multiple highly-specialized LangGraph nodes.
Handling multi-turn state cleanly between Router, Execution, and Summary nodes.
Real-Time Automated Testing & GitHub Actions CI: Features unbuffered real-time test execution and continuous integration via GitHub Actions (.github/workflows/chapter5-ci.yml). Automated testing is essential in Agentic AI to verify non-deterministic LLM intent routing, validate precise MCP tool schemas/contracts, and guarantee pipeline resilience when upgrading underlying foundation models.

Chapter 5 Demo: See Agentic Workflow in Action

This chapter includes a live demo of the multi-node assistant handling different intents:

Conversation routed to a chitchat node
Math requests executed by the MCP tool server
Email composition and sending via a secure tool layer

Sample interaction:

User: "Please add 42 and 17, then send the result to my email."
Router: detects math + tool request
Execute: calls MCP Math tool, then MCP Email tool
Summarize: returns a human-friendly response with results and confirmation

Read more in the Chapter 5 guide and view the demo flow: Chapter 5 Demo

🔗 Related Repositories

Explore more tutorials and tools by the same author:

Repository	Description
🤖 Hands On AI Tutorials	A comprehensive, open-source AI tutorial covering ML Fundamentals (Regression, Classification, Clustering) and Deep Learning (NLP, CV) with practical code and mathematical intuition.
⚡ Concurrent LLM Serving	Hands-on guide to serving Large Language Models efficiently at scale with concurrency

These repositories complement this tutorial — once you've mastered the fundamentals here, explore agents and production LLM deployment next!

🤝 Contributing

We love contributions! Whether you're fixing a typo or adding a new agent pattern:

Fork the project.
Create your Feature Branch (git checkout -b feature/AmazingFeature).
Commit your changes (git commit -m 'Add some AmazingFeature').
Push to the branch (git push origin feature/AmazingFeature).
Open a Pull Request.

👤 Author

Zkzk - AI Engineer & Educator

GitHub: @zkzkGamal

📖 Featured articles

Disclaimer: This tutorial is for educational purposes. Costs may apply for cloud LLM usage.