Building Applications with AI Agents: Designing and Implementing Multiagent Systems

Very thoughtful, well-organized, comprehensive overview of agents. Used it as a reference when building an agentic system, and it provided a good initial framing to get me started.

amazing book

Super disappointed! Not sure what the author is trying to achieve. Verbose at the same time not technical enough. I will see if I can return it. The author did not make any attempt to help the reader to run the applications like author books where you go to the github and there the code is linked to say Google Colab.

I think this book is a pretty good high level overview of using agents. However, I think the name of the book is a bit misleading - the book gives no real insights into building actual applications with agents; rather, it's mostly just disparate code snippets. Moreover, the book was obviously written by a data scientist - the book's github repo is terrible, and is clearly missing a lot of the dependencies that would be needed to run the code. Also, the code itself is terribly broken, with references to variables that are never defined, among other things. I also had an issue where the author is clearly misdefining or conflating memory and knowledge at different points in one chapter. All that being said, I think that the high level info is pretty good, and gives a decent account of retrieval techniques and agents.

Excelente libro, recomendado

This book tries to cover a lot of ground on agentic system design, and to be fair, it does collect many of the right ideas in one place. If you’re new to the topic, you’ll probably recognize most of the concepts and buzzwords. The problem is that everything stays very high level, so you never really feel like you’re learning how to build or reason about real systems. It’s also extremely repetitive. The same points come up again and again, often in back-to-back chapters, with little added depth. Instead of reinforcing ideas, the repetition just becomes tiring. The code examples don’t help much either, they feel artificial and often unrelated to the actual discussion, as if they were added just to check a box. Overall, it’s not a great read. While it works as a loose collection of ideas, it lacks focus, depth, and practical relevance. I wouldn’t recommend it unless you just want a shallow overview and don’t mind wading through a lot of repetition.

It reads like AI-generated slop.

Entirely AI written, not in a good way. I was looking for an author's judgment, as a first book its disappointing Michael. I looked on your website and seems like you could know what you are talking about. Have more faith in your ability. If you do use AI, use it in a way that improves the work.

Lite för långrandig, men helt ok!

Ottimo testo, chiaro e con esempi dettagliati

Very insightful. A must read.

This book was fantastic! I bought it after building my first multi-agent system, and it gave me so many excellent ideas that addressed all of my common failure modes that I rewrote it as a single agent using only the tools code from the old one and it works so much better! This is not a book that gives you a ton of Langgraph code (though there is some!) but rather it gives you great ideas!

This book covers a lot of useful ground, and serves as a good anchor for building agentic systems. However, the code examples are riddled with errors that meant I spent more time debugging the examples to run them than actually reading some of the chapters.