I always have a slight problem with books such as this. It is only a personal opinion but if you want to construct intelligent agents, you need to know the principles that allow you to do this, rather than mixing the ideas with a particular language.
This book takes a sensible approach. In the first part of the book, after discussing the Java language, it looks at the AI (Artificial Intelligence) aspects of the problem. This includes how to conduct searches, how to represent knowledge and how a program can reason. Having said that, these sections are still scattered with Java code.
Part two of the book looks at building applications. It starts by looking at a framework for intelligent agent applications and then moves onto to developing specific applications such as news filters and agents negotiating with one another in a virtual market place.
If anything this book tries to cover too much material. For example, it covers search strategies, predicate logic and neural networks; to name just a few. Some of you will appreciate that each of these subjects have entire books devoted to them.
I applaud the authors in trying to cover so many techniques but if your goal is in trying to develop a fully fledged intelligent agent you might need more information than is contained in this book.
If you need to use Java and you need an introduction to intelligent agents then this book is worth having, but if you need more in depth coverage of either subject then there are more complete books.