Jagged Edges

AI doesn't fail loudly. It fails confidently.

It counts two Rs in strawberry while simultaneously outperforming the average lawyer. The output looks right. The reasoning sounds right. The answer is wrong in a way that compounds What four years of building AI systems that didn't work—in production, at scale, expensively—teaches you is something no success story can: the specific texture of where AI fails. Where the boundary is. What the wrong side looks like just before you cross it. That map is what this book is.

Jagged Edges is for leaders, operators, and builders already inside the work. The ones who can't wait for the tools to stabilize, the regulations to clarify, or the organization to align—because it doesn't stabilize, and the work still has to happen.

Drawn from real deployments across global supply chains, retail operations, and organizational leadership.

What you'll find inside:

1. Why the model that passed every test still got your customer wrong—and why the output that sounds most certain is sometimes the most dangerous

2. What "the faster I go, the faster it comes" actually costs, and how to break the cycle before it runs you

3. How to build gates that catch failure before it compounds—because prompting is generation, and gates are trust

4. The water cooler wisdom everyone in your organization already knows—and why it never reaches the person who can change it

5. Why the gap between AI leverage and AI noise isn't technical—and what actually determines which side you're on

This isn't a productivity book. It's a field guide for navigating terrain that keeps changing shape—written by someone who paid for the map at full price.