Book Review: The New Rules

Note: I was given a free copy of the book for review. Although you its digital version appears to be free (registration required).

Review

Title: The New Rules - Developer's Guide to AI Era

Author(s): Andrzej Tuchołka

There is a disclaimer stating that the book was created with AI assistance, which in itself is not an issue. The main problem is the complete absence of references, combined with a hyperbolic, alarmist, and highly prescriptive tone. The book makes sweeping claims, sometimes contradicting earlier chapters, and frequently cites very specific metrics (for example, "this made their product 34.5% faster") without any sourcing. After checking several of these claims and numbers, I either found no supporting evidence, or evidence that directly contradicted them. Because of this, after about 20% of the book, I stopped treating individual affirmations, anecdotes, and quotes as trustworthy, and instead focused only on the high-level ideas and intended takeaways of each chapter.

Despite these issues, the book does contain genuinely valuable advice, particularly in two areas: working in or running a startup, and navigating the current wave of Generative AI adoption. I ended up taking dozens of notes on practical ideas that are worth considering. To give some examples, these include a strong push toward being documentation-driven, with concrete suggestions on how to structure, maintain, and evolve documentation; guidance on improving AI performance by relying on predictable patterns and clear conventions; thoughtful discussion around when AI-assisted automation makes sense, and when it is more likely to create friction or hidden costs. Stripped of the exaggerated claims, many of these ideas are pragmatic, actionable, and can even benefit in human/non-AI scenarios.

A revised edition with proper references and corrected claims would significantly improve the book's credibility and reach. The core approaches and ideas are already there and are often compelling; they mainly need grounding and evidence.