Librería Samer Atenea
Kálamo Books
Librería Elías (Asturias)
Librería Kolima (Madrid)
Librería Proteo (Málaga)
The industry is racing to make AI agents smarter. This book is about making them trustworthy.Trustworthy Agentic AI teaches experienced software engineers how to build reliable, production-ready multi-agent systems using Java and the Akka SDK. You will learn the engineering discipline behind agents that are predictable, auditable, recoverable, and observable. You will learn about the Four Pillars of Trust that separate systems you can explain and defend from systems you merely hope will work.Through Sentinel, a fully worked incident triage system that evolves across every chapter, you will build:- Agents with durable memory, tool integration, and human-in-the-loop approval- Orchestrated multi-agent workflows with supervision and failure recovery- Guardrails and evaluation pipelines that catch drift before users do- Security boundaries, observability, and testing strategies designed for non-deterministic systemsBuilt on the Actor Model, the same concurrency foundation behind some of the world’s most demanding distributed systems, every pattern in this book is backed by runnable Java code, not pseudocode or hand-waving.If you already know how to build production software and want to bring that same engineering rigor to agentic AI, this is your book.
