Contents

Front matter

1. How to read this book1 page · 2 min

   The four parts of an AI system, and how to turn the pages.

2. About this book1 page · 2 min

   Where the facts come from, and how AI helped.

Part IThe basics

What these systems are, in plain English.

1. Chapter 1What is a language model?2 pages · 5 min

   Where an LLM fits inside "AI."

   1. p. 1 Autocomplete that read a library
   2. p. 2 Where "LLM" fits
2. Chapter 2How does it work?3 pages · 8 min

   Tokens, prediction, training and the desk.

   1. p. 1 Predict, pick, repeat
   2. p. 2 Tokens
   3. p. 3 The desk

Part IIThe roots

The ideas are older than you think.

1. Chapter 3Shannon's bits to Claude's tokens1 page · 3 min

   A 1948 idea about guessing the next letter. The full story in the Retro Lab →

2. Chapter 4Interlude: talk to ELIZA (1966)1 page · 2 min

   A chatbot from before machine learning. Play ELIZA in the Arcade →

3. Chapter 5From early models to today3 pages · 7 min

   Seventy-five years of predicting the next symbol.

   1. p. 1 From counting to transformers
   2. p. 2 Assistants for everyone
   3. p. 3 Thinking and doing
4. Chapter 6Why there are different sizes3 pages · 8 min

   Haiku, Sonnet, Opus, Fable and Mythos.

   1. p. 1 A family of sizes
   2. p. 2 The top rung
   3. p. 3 Pick the right model
5. Chapter 7Small and free models2 pages · 5 min

   What a model you download is good for.

   1. p. 1 What small models are good for
   2. p. 2 Where they stumble

Part IIIThe parts at work

The prompt, the harness and the tools around the model.

1. Chapter 8The prompt1 page · 4 min

   What the model actually reads.

2. Chapter 9The harness2 pages · 5 min

   Why the same model feels different in different apps.

   1. p. 1 The body around the brain
   2. p. 2 One brain, many bodies
3. Chapter 10Tools2 pages · 6 min

   How a model searches the web or opens a file.

   1. p. 1 The hands
   2. p. 2 The agent loop
4. Chapter 11How it all fits together2 pages · 5 min

   Step through the whole loop.

   1. p. 1 The whole loop
   2. p. 2 Build your own

Part IVUsing it wisely

Your data, and the questions people ask.

1. Chapter 12Where do my words go?3 pages · 10 min

   Personal and work accounts, privacy and HIPAA, as concepts.

   1. p. 1 Whose agreement is it?
   2. p. 2 Each hop, its own agreement
   3. p. 3 Health and student records
2. Chapter 13Questions people ask1 page · 4 min

   Memory, made-up answers, agents and more.

Appendices

1. Appendix AThe parts list2 pages · 4 min

   Every term in the book, in plain English.

   1. p. 1 The model and the big words
   2. p. 2 Prompt, harness and tools
2. Appendix BCase study: small models in a coding harness7 pages · 13 min

   29 models, one small coding job, one harness.

   1. p. 1 The model is only one part
   2. p. 2 The score, in three groups
   3. p. 3 The task
   4. p. 4 What went wrong
   5. p. 5 Does a big model's plan help a small one?
   6. p. 6 Cost and time
   7. p. 7 Where small models fit
3. Appendix CCheck yourself3 pages · 6 min

   Quick questions to see what stuck.

   1. p. 1 The parts
   2. p. 2 Your data
   3. p. 3 Where can this go?
4. Appendix DBibliography1 page · 2 min

   Every source cited in the book.

5. Appendix EHow this book was researched1 page · 2 min

   The research standard, the AI disclosure, and how to report an error.