Agentic AI Explained

From Neural Networks to Autonomous Agents: How It Works, How to Build It, and What It Means

Introduction

1. What this book covers
2. How to read this book

Chapter 1: What Is Intelligence, Anyway?

1. An Agent at Work
2. Defining Intelligence
3. The Long History of AI
4. Narrow AI, General AI, and the Hype Gap
5. Why This Book Exists

Chapter 2: How Machines Learn

1. Patterns Everywhere
2. Training, Testing, and the Illusion of Understanding
3. The Magic of Gradient Descent
4. When Learning Fails
5. From Rules to Learning

Chapter 3: The Brain Inspired the Network

1. Biological Neurons vs. Artificial Neurons
2. Layers of Abstraction
3. Deep Learning and the Scaling Phenomenon
4. Specialized Architectures
5. What Networks Actually Represent: Embeddings and the Geometry of Meaning

Chapter 4: The Transformer Revolution

1. Why Previous Models Hit a Wall
2. Attention Explained
3. Self-Attention and Context
4. Encoder and Decoder
5. The Big Paper

Chapter 5: What Is a Large Language Model?

1. Language Modeling as a Prediction Game
2. The Data Diet
3. Size, Scale, and Emergent Abilities
4. Foundation Models and the Shift in AI
5. What LLMs Are Not

Chapter 6: Inside the LLM Brain

1. Tokenization
2. Context Windows and Working Memory
3. The Inference Dance
4. Temperature, Sampling, and Why LLMs Are Probabilistic
5. The Hidden Space

Chapter 7: Prompting and the Art of Talking to LLMs

1. The Prompt as Program
2. Zero-Shot, Few-Shot, and Example-Based Prompting
3. Chain of Thought and Structured Reasoning
4. System Prompts and Role-Playing
5. When Prompting Fails

Chapter 8: Reasoning, Hallucination, and the Illusion of Understanding

1. Statistical Reasoning vs. Human Reasoning
2. Chain of Thought as a Tool
3. The Hallucination Problem
4. Confidently Wrong
5. Tools for Checking and Grounding

Chapter 9: From Chatbots to Agents

1. The Chatbot Paradigm
2. The Agent Paradigm
3. What Makes an Agent “Agentic”
4. A Simple Example: Email Assistant vs. Email Agent
5. The Spectrum of Autonomy

Chapter 10: The Anatomy of an AI Agent

1. The Core Loop: Perceive, Think, Act
2. The Brain: The LLM as the Central Reasoning Component
3. The Toolbox: Tools, APIs, and How Agents Call External Functions
4. The Memory System: Short-Term Context, Long-Term Storage, and Knowledge Retrieval
5. The Orchestrator: What Coordinates the Agent’s Behavior and Enforces Constraints

Chapter 11: How Agents Use Tools

1. Function Calling Explained
2. Web Search and Browsing
3. Code as a Tool
4. Integrating with Business Systems
5. Tool Selection and Chaining

Chapter 12: Memory, Knowledge, and Retrieval

1. The Memory Problem
2. Short-Term Memory: Conversational Context and Working Memory
3. Long-Term Memory: Vector Databases, Knowledge Bases, and Persistent Storage
4. Retrieval-Augmented Generation
5. Embeddings and Similarity Search

Chapter 13: Planning and Decision Making

1. Goal Decomposition
2. Planning Algorithms
3. Reflection and Self-Correction
4. Uncertainty and Adaptation
5. The Limits of AI Planning

Chapter 14: Multi-Agent Systems

1. Why Multiple Agents?
2. Communication Protocols
3. Roles and Responsibilities
4. Multi-Agent Frameworks
5. When More Agents Help and When They Hurt

Chapter 15: Building an Agentic System, Step by Step

1. Problem Definition
2. Model Selection
3. Designing the Agent Architecture
4. Prompt Engineering for Agents
5. Permissions, Guardrails, and Human Oversight

Chapter 16: Guardrails, Safety, and Control

1. Why Agents Need Guardrails
2. Input and Output Filtering
3. Tool-Level Safety
4. Human-in-the-Loop Patterns
5. Monitoring and Kill Switches

Chapter 17: Security Risks and Attack Vectors

1. Prompt Injection Explained
2. Indirect Prompt Injection
3. Data Leakage and Privacy
4. Tool Abuse and Escalation
5. The Attacker’s View

Chapter 18: Reliability, Evaluation, and Monitoring

1. Defining Success
2. Evaluation Frameworks
3. Red Teaming
4. Observability
5. Continuous Improvement

Chapter 19: Real-World Applications

1. Software Development and Coding Assistants
2. Research and Analysis
3. Customer Service and Support
4. Business Operations
5. Healthcare, Finance, and Regulated Industries

Chapter 20: Limitations and the Realistic Horizon

1. The Hard Problems
2. Context and Attention Limits
3. The Reliability Gap
4. Cost and Latency
5. What’s Coming Next

Chapter 21: The Bigger Picture: Ethics, Society, and the Future

1. Jobs, Automation, and Work
2. Centralization and Power
3. The Environment
4. Alignment and Control
5. Navigating the Transition

Chapter 22: Conclusion: Agentic AI, Clearly

1. What We Have Learned
2. Agentic AI in Context
3. How to Think About Agentic AI Going Forward
4. The Human Element

References