Keyboard shortcuts

Press or to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Preface

🤖 "The future belongs not to those who write code, but to those who can direct AI to write code. And Agent is the bridge connecting human intent with AI capability."

Why This Book?

Since 2023, Large Language Models (LLMs) have sparked a technological revolution. But many people have found that simple "conversation" is far from enough — we need AI that can make autonomous decisions, use tools, and complete complex tasks. This is the purpose of Agents (Intelligent Agents).

However, learning resources for Agent development are scattered across various papers, blogs, and framework documentation, lacking a systematic, beginner-friendly, and practice-oriented learning path. This book is written to fill that gap.

Who Is This Book For?

  • 🐍 Developers with Python basics who want to enter the AI Agent field
  • 🎓 Students or researchers interested in LLMs who want to understand the principles and implementation of Agents
  • 💼 Product managers or tech leads who want to understand the capability boundaries and application scenarios of Agents
  • 🔧 People already using ChatGPT and similar tools who want to understand and customize AI capabilities more deeply

Features of This Book

FeatureDescription
📖 Progressive LearningStarting from "What is an Agent?", gradually deepening into multi-agent systems, reinforcement learning training, and production deployment
🐍 Python in PracticeEvery chapter includes runnable Python code examples — learn by doing
🎨 Rich VisualsExtensive architecture diagrams, flowcharts, and sequence diagrams to help understand concepts intuitively
🔨 Project-Driven3 complete comprehensive projects, covering the full workflow from requirements to deployment
🌐 Cutting-EdgeCovers LangChain, LangGraph, MCP/A2A/ANP protocols, Context Engineering, Agentic-RL, and other latest technologies
📚 Academic TracingEach chapter includes authoritative paper citations and references, covering the complete academic spectrum from symbolic AI to LLM-driven approaches
📊 Evaluation SystemIn-depth breakdown of mainstream benchmarks including BFCL, GAIA, AgentBench, and SWE-bench

How to Use This Book?

Recommended Learning Path

Chapter Structure:

  1. Concept Explanation — Core concepts explained in plain language with analogies
  2. Architecture Diagrams — Clear diagrams showing system design
  3. Code Practice — Complete, runnable Python code
  4. Exercise Challenges — Hands-on exercises to reinforce learning
  5. Further Reading — Recommended papers and resources

Technology Stack Overview

This book primarily uses the following technology stack:

  • Programming Language: Python 3.11+
  • Core Frameworks: LangChain, LangGraph, OpenAI Agents SDK
  • LLM Services: OpenAI API (GPT-4o/GPT-5), Anthropic (Claude 4), Open-source models (Llama 4, Qwen 3)
  • Agent Protocols: MCP (Model Context Protocol), A2A (Agent-to-Agent), ANP (Agent Network Protocol)
  • Vector Databases: ChromaDB, FAISS
  • Web Framework: FastAPI
  • Other Tools: Docker, Pydantic, asyncio, uv

Ready? Let's embark on the Agent development journey together! 🚀