Traditional publishing is great. Really. The prestige, the distribution, the little publisher logo on the spine. Michelle and I talked about it for exactly one evening.
Then we did the math. Best case scenario? You'd be reading this book in late 2026. Maybe early 2027 if everything went perfectly.
By then, the agentic AI landscape will have transformed three times over. The examples in Chapter 1 would be kaput. The security challenges we're solving today would be replaced by problems we can't even imagine yet.
That math made the decision easy. Self-publish. Get it out now.
I'm watching companies make AI decisions TODAY that they'll be living with for years. Some are racing ahead without security. Others are paralyzed, waiting for "best practices" to emerge. Both approaches scare me.
The good news? The foundations don't change. Zero Trust worked before AI. Ask John Kindervag--he talks about it in our foreword. It works with today's AI agents. It can work with whatever wild AI capabilities emerge next year. Taking time now to build security-first into your AI approach isn't just about protecting today's deployments. It's about being ready for whatever comes next.
Could we have made this book prettier with a traditional publisher? Probably. Could we have gotten it into more airport bookstores? Maybe. But you know what we couldn't have done? Gotten it into your hands while it's still relevant.
The companies deploying AI agents aren't waiting for 2027. Their competitors aren't waiting. The security threats definitely aren't waiting.
So we're not waiting either. The book is done. It's practical. It's based on what's actually working right now, not what might work someday.
Because "someday" is the enemy of "secure today."