✦ The book

The Builder’s Playbook

Your AI writes the code. You make the calls.

System design for people who build with AI. 36 chapters in six parts that teach you to direct the AI that writes your product, and to choose designs that still hold after your first thousand users.

Chapter 12, its companion skill and a practice lab, by email. Nothing to buy, and no list sold on.

System design for people who build with AI The Builder’s Playbook Architect, not just prompter. Alireza Chamanrouy

01 Sound familiar?

  • “My app works, but I don’t know what I don’t know.”
  • “My AI always says it’s done. How do I know it’s right?”
  • “What if there’s a data leak, a surprise bill, or everything falls over at launch?”

These are the questions this book answers. Not with code, but with the judgement to know what to ask.

02 How it works

Written for the way
you build now.

For AI builders

Every chapter talks to your AI.

Each one has What your AI will probably suggest, Questions to ask your AI, and Red flags in AI answers. You read, then you open your agent and use them.

Here’s a quick admin panel. I hid the button for non-admins. Who can still call the endpoint directly?
Stage-honest

Design for the stage you’re in, not for Google.

Every idea is shown at four sizes, so you learn what you need now, and what you don’t need yet.

  1. Smallup to ~1k users
  2. Medium~1k to 100k users
  3. Large~100k to tens of millions
  4. Gigantichundreds of millions
One running story

From 60 customers to millions, one decision at a time.

In the book, StudioBook starts as a booking app for one small yoga studio and grows chapter by chapter into a global platform. Every concept is shown at each step of that growth.

Real incidents

Built around what actually breaks.

The book is built around real, public incidents, with sources cited: the leaked key, the surprise cloud bill, the open door into other people’s data, the app that died on launch day.

03 Inside

36 chapters.
Six parts. Seven appendices.

Part I

Thinking like an architect

  • Why AI-built products break
  • The journey of a click
  • Stages of a product
  • Requirements before code
  • Trade-offs: the core skill of design
  • The design doc

Part II

The layers of a digital product

  • Frontend
  • Backend and API design
  • Modeling your data
  • Choosing and running a database
  • Authentication
  • Authorization and RBACChapter 12 · Free
  • Multi-tenancy
  • Third-party APIs and integrations
  • Background jobs, queues and events
  • Files and media
  • Real-time features

Part III

Performance and scale

  • Latency: where the time goes
  • Caching layers
  • Scaling out
  • Reliability and failure

Part IV

Environments, infrastructure and operations

  • From laptop to production
  • Shipping safely (CI/CD)
  • Cloud services and infrastructure
  • Observability
  • Security essentials
  • Cost
  • Data, privacy and compliance

Part V

Real-world architectures

  • Introduction: how to read a case study
  • Small: one of each, big results
  • Medium: the majestic monolith
  • Large: scaling teams and systems together
  • Gigantic: custom infrastructure
  • StudioBook grows up

Part VI

Playbooks for working with your AI

  • The architecture session
  • Reviewing what your AI built
  • When your AI is wrong

Appendices: a master question bank, a red-flags index, decision trees, stage checklists, a glossary, a companion skills guide, and practice labs.

04 The free sample

Hiding a button
isn’t a lock.

Read the chapter, then check your own app with the skill, then practise on a broken one. Three things, sent by email:

  1. Chapter 12: Authorization and RBAC. Roles, permissions, ownership checks, admin panels and row-level security. The questions to ask before your AI ships an admin panel, and the red flags in its answers.
  2. The authz-review skill for Claude Code. It reviews the authorization in your own app. It is written to change nothing without asking, to cite a file and line for every finding, and to say what it did not check.
  3. The StudioBook practice lab. A broken copy of the book’s example app with planted authorization holes, and an answer key, so you can find them yourself.

Tested on a practice app with 12 planted security holes, our authorization skill found all 12 in every run with a strong model.

Why it matters: broken object level authorization, loading a record without checking it is yours, is #1 on the OWASP API Security Top 10 (2023).

Chapter 12 of 36 Authorization and RBAC What are you allowed to do? Rule: check ownership on the server, every time. Red flag: “I hid it in the UI for non-admins.”

05 The author

Alireza Chamanrouy

Product designer turned founder. Years of designing B2B platforms, ERP and telecom tools, then product, then ShapeShifters, a consultancy that rebuilds how businesses run. This book is the thinking behind that work, written down for people who build with AI.

Read my story

06 Questions

Before you ask.

Do I need to know how to code?

No. The book teaches you to direct the AI that writes the code: which questions to ask, when to push back on the obvious answer, and how to choose designs that fit your stage.

Who is it for?

Founders, indie makers, designers, product managers and anyone building real products with AI assistants like Claude, ChatGPT, Cursor, Lovable, Bolt or v0. It is not interview prep, and it is not written for senior engineers.

What is in the free sample?

Three things, sent by email: Chapter 12, Authorization and RBAC; the authz-review skill for Claude Code, which reviews the authorization in your own app; and the StudioBook practice lab, a broken copy of the book’s example app with planted holes and an answer key.

Which AI tools does the skill work with?

The authz-review skill is a skill for Claude Code. The book’s questions and red flags work with any AI assistant.

Can I buy the book?

Nothing is for sale right now, and nothing can be ordered in advance. The free sample is the only thing on offer.

Architect,
not just prompter.

Get the free sample