Chapters

Chapters (0–10)

11 chapters

1. 

   Chapter 0 — Introduction V 1.0

   Vision This chapter sets the stage: what “agent systems” are, why the paradigm exists, and what you will build as you go. What you will do (preview) Understand how components interact (model, loop, tools, memory) Build locally using open-source components (no paid agent APIs required)

2. 

   Chapter 1 — Foundations V 1.0

   What this chapter will cover Basic concepts of AI, why LLMs became the dominant approach, and the mental model you will use throughout the book. Planned outcomes Build intuition for tokens/probability without heavy math Understand limitations: hallucinations, context windows, and evaluation

3. 

   Chapter 2 — How Models Work V 1.0

   First-principles explanation This chapter explains how LLMs transform text into predictions: tokens, probabilities, and the practical implications for reliability. Why this matters for agents Agents depend on the model’s output distribution. Understanding “what the model is doing” is the foundation for tool-use, planning, and evaluation.

4. 

   Chapter 3 — System Architecture V 1.0

   How components fit together Model, loop, tools, memory, planning, and evaluation are not separate “modules” in practice—they interact continuously. This chapter will provide a clear reference architecture and show how to wire it together locally.

5. 

   Chapter 4 — Environment & Setup V 1.0

   Practical local setup You will set up the environment to run models locally (or on your private VM), configure dependencies, and keep everything reproducible. What you will be able to do Install the core Python stack Run an agent loop locally end-to-end

6. 

   Chapter 5 — The Agent Loop V 1.0

   How iteration creates intelligence over time This chapter focuses on the “loop” that turns single-shot model outputs into multi-step behaviors: think, act, observe, reflect. Key idea Agents get better through repeated cycles, guided by constraints and evaluation—not just by larger prompts.

7. 

   Chapter 6 — Tool Systems V 1.0

   Current state Tool-use is powerful, but it is also where agents can fail in subtle ways (inconsistent depth, wrong assumptions, brittle tool calls). This chapter is partially written and will be rewritten for clarity and robustness.

8. 

   Chapter 7 — Memory & Retrieval V 1.0

   Scaling beyond context When the conversation grows, agents need selection and compression strategies to stay useful. This chapter explains how to retrieve the right information and how to represent memory so it works reliably.

9. 

   Chapter 8 — Planning & Decomposition V 1.0

   Direction and structure over time This chapter explains how agents use planning and decomposition to turn ambiguous goals into executable steps. What you will learn When to plan How to decompose How to evaluate plans as the loop progresses

10. 

    Chapter 9 — Multi-Agent Systems V 1.0

    Collaboration patterns Multi-agent systems can outperform single agents when you have clear roles, interfaces, and evaluation methods. This chapter will map the main patterns and show how to reason about the trade-offs.

11. 

    Chapter 10 — Advanced Systems & Production V 1.0

    From prototype to production Production systems require orchestration, evaluation, safety controls, and robust deployment practices. This chapter will connect architecture decisions to real-world constraints.