Skip to content
Muhammet Şafak
tr

A series in 3 volumes

The Unreliable Systems Trilogy

Reliable systems are not built from reliable parts. They are built around the parts that fail: access to production data, messages that arrive twice, and the authority handed to agents.

Three free books on one idea: reliable systems are not built from reliable parts but around the parts that fail — access, events and AI agents.

Start with Book I Free to read · EPUB & PDF
  1. I Access: production data
  2. II Events: failure semantics
  3. III Agents: bounded authority

The argument

Each book's thesis: what the problem is not, and what it is
Vol. Book Not But
I Access: production data The Production Database Access Guide a credential-distribution problem an authorization, execution and accountability problem
II Events: failure semantics Designing Reliable Event-Driven Systems message brokers failure semantics
III Agents: bounded authority Engineering Reliable AI Agents trusting the model to be right making its mistakes survivable

About the series

A shared admin password, an instance that shut down before sending its event, a lookup called 40 times in a row. These are the opening scenes of the three books; all three are set in fictional companies, and in none of them is there a single big mistake.

In the first book, the credential did exactly what it was issued to do. In the second, every line of code did what it said. What failed was the space around the parts: a door nobody designed, the gap between a commit and a network call.

The first book, The Production Database Access Guide, looks for the problem not in credential distribution but in authorization, execution and accountability. In the second, Designing Reliable Event-Driven Systems, the problem is not message brokers but failure semantics: you have to decide in advance what happens when a message is late, lost, duplicated or out of order.

The third book, Engineering Reliable AI Agents, applies the same reasoning to the decision-maker itself. If the model will sometimes be wrong, what must the surrounding control system guarantee so that being wrong is survivable?

In all three books, every control comes with its cost: what it asks of people, which mistakes are common and when not to use it.

Which book to start with

The series is numbered I Access, II Events, III Agents, and the books appear on this page in that order. The books do not argue for this order, and their prefaces ask for general engineering experience, not for having read another volume. If one of the three matches the problem in front of you, you can start there.

Read in numbered order, the series moves from who may touch production data, to what happens to messages between services, to software that decides for itself. The first book’s least privilege and the second book’s idempotency reappear in the third, applied to agents. The books do not say that this is why they are numbered this way; it is what the order gives a reader who goes through all three.

What the books share

The first book covers who is allowed to touch production data: identities, roles, credentials and the record of what each principal did. The third book returns to similar material for software that decides for itself. It applies least authority per tool, per resource and per run, and it keeps the decision on every call in a policy engine outside the model.

The second book gives a full chapter to idempotency, the property that makes a repeated message safe. The third book uses the same idea: a write proposed by an agent carries an idempotency key, so a repeat returns the first result and not a second effect.

How to read them

Each book is free to download, in PDF and EPUB. Every book closes its chapters with key takeaways and ends with a glossary, so you can check a book against your own system before committing to a full read. The book pages carry chapter summaries, a glossary and a verbatim excerpt from one chapter for the same purpose.

Editions

Editions by book and language
Book English Türkçe Español 简体中文 Português Deutsch Français
I The Production Database Access Guide Available Available Available Available Available Available Available
II Designing Reliable Event-Driven Systems Available Available Available Available Available Available Available
III Engineering Reliable AI Agents Available Available Available Available Available Available Available

Search the site

Start typing to search posts, projects and pages.

Esc to close Powered by Pagefind