The flagship

AI from Beginner to Advanced

What these models are, how to direct them, and what to build

The book people keep asking for: not a list of tools, and not a math textbook. It explains what a language model actually does, why that explains almost all of its failures, how to direct one reliably, and how retrieval, agents and fine-tuning fit together — in the order that makes each one make sense.

  • Starts at zero — no coding background assumed
  • Explains the failures, not just the features
  • The longest book in the library, and the one to read first
Pages
312
Chapters
12
Read time
~8h
Language
English

$5.99$29.99

Founding price · instant download on launch

Join the founding list

Nothing is charged today. AI, Start to Advanced is being written — join and you keep this price and get it the day it is published.

  • PDF and EPUB, both DRM-free
  • Every prompt in the book, as a copyable file
  • A glossary of every term, in plain English
  • Free updates as the field moves — this is the title that will change most

Tax included, so the price never changes at checkout. Instant download, no DRM, readable on this site or on any device, yours to keep.

E-book rises to $11.99 on 1 October 2026 · 4 days away

Read this if

  • You have used ChatGPT and want to understand what is actually happening.
  • You keep being told to "learn AI" and have never found a starting point.
  • You want to judge a new model announcement for yourself.

Do not read this if

  • You are already fine-tuning models and reading papers.
  • You want a directory of 500 AI tools.
  • You want to skip the chapter on how it fails. That chapter is the reason for the book.

After reading

What you will be able to do

1

Explain in plain language how a language model produces its next word, and why that explains most of its failures.

2

Write prompts that hold up repeatedly — with structure, examples, constraints and a fallback.

3

Understand retrieval well enough to know why it, not fine-tuning, is the answer to "it does not know my data".

4

Recognize the three ways agents usually break before building one.

5

Read a new model announcement and work out in ten minutes whether it changes anything for you.

The full contents

12 chapters, 312 pages

Every chapter, and what you have at the end of it. Nothing here is behind the payment.

There is a course too

The book is the map. Learn AI is the guided build.

This book gives you the whole picture and the templates — most people only ever need that. The course is 12 weeks of building it with the work checked, for 6 shipped projects. Read the book first; it costs a fraction and tells you whether you want the rest.

See the course

What this book does not promise

This block is on every book and course page and is never removed to make a page read better. It is what makes everything above it believable.

  • This book will not make you an AI researcher. It makes you literate in AI, which is a different and more immediately useful thing.
  • It contains very little code. The course is where you build; this is where you understand.
  • Parts of it will go out of date. That is why the e-book updates are free and why the last chapter teaches you to check things yourself.
  • It does not promise you a job or an income.

Questions about this book