akitaonrails / ai-book-2026

Português · English 
 Competente o Suficiente Competent Enough 
 Um ano colocando a inteligência artificial à prova, sem medo e sem propaganda 
 A year putting artificial intelligence to the test, without fear and without hype 
Fabio Akita 
---
A book for people who are anxious or confused about AI: the professional
afraid of being replaced, the parent who read that AGI is two years away, the
manager being sold an "AI transformation". It follows one veteran programmer
through 2026 as he builds real software with AI agents, measures dozens of
models in his own benchmark, gets things wrong, corrects them in public, and
reads the industry's announcements against the evidence.
The book grew out of the year's articles on
akitaonrails.com, reorganised and rewritten
as one narrative, fact-checked again, and illustrated.
## Read it
Download the latest edition from
Releases:
| File | For |
|---|---|
| `Competente-o-Suficiente.epub` | Kindle, Kobo, Apple Books, any e-reader |
| `Competente-o-Suficiente-ebook.pdf` | Reading on a screen |
| `Competente-o-Suficiente.pdf` | Hardcover interior (6 × 9 in) |
| `Competente-o-Suficiente-paperback.pdf` | Paperback interior (6 × 9 in) |
| `Competent-Enough.epub`, `Competent-Enough-ebook.pdf`, `Competent-Enough.pdf` | The English edition, same formats |
**Status:** version 1.0, published in Brazilian Portuguese and English.
Future corrections and updates will be versioned separately.
## What's inside
| Part | The question it answers |
|---|---|
| I. O que é essa coisa, afinal | Is it intelligent? Is AGI coming? |
| II. A maratona | Does it actually work, or is it all demo? |
| III. Números e propaganda | Which AI is best? Should I believe the announcements? |
| IV. Quem fica com as chaves | Is it dangerous to let it touch my things? |
| V. E o meu emprego? | Will it take my job? What comes next? |
Plus two interludes, an optional no-math appendix on how the machine works
inside, a timeline, practical checklists, a glossary of 144 terms and 292
references.
## Em português
Um livro pra quem está ansioso ou confuso com a inteligência artificial. Um
programador veterano passa 2026 construindo software de verdade com agentes de
IA, mede dezenas de modelos no próprio benchmark, erra, corrige em público e
confronta os anúncios da indústria com as evidências. Baixe a edição mais
recente em Releases.
## How it's made
This repository is the book's source and the machinery that keeps it in sync
with the blog. Every paragraph records which article it came from; when an
article changes, `rake drift` shows which chapters are affected and the diff.
Diagrams are drawn as code, AI-generated illustrations are disclosed in the
image credits, and photographs have source and rights-status records (including
explicitly unverified permissions). The details are in
docs/DESIGN.md.
 [code] 
Requires Ruby, Java (for epubcheck) and poppler-utils.
The PT-BR build emits hardcover and paperback interiors with separate ISBNs;
English emits a hardcov