The Question Nobody Asks Before Starting an AI Project

Everyone starts with "what should I build?" It's the obvious question, and it's not the right one.

The right first question is: "What do I know that the AI doesn't?"

This sounds abstract. It's not. It's the most practical question you can ask, and the answer shapes every decision that follows — what to build, how to collaborate, and what quality standard to hold the output to.

What you know that the AI doesn't is your context. Your industry's unwritten rules. Your audience's real concerns, not the ones they state publicly. The constraints that don't show up in any documentation. The history that makes certain approaches non-starters. The taste that lets you distinguish between output that's technically correct and output that's actually good.

When you start with "what should I build," you get a project idea — which you then execute with AI as a general-purpose assistant, getting general-purpose results. When you start with "what do I know that the AI doesn't," you get a strategy for collaboration — a clear picture of what you bring and what you need the AI to do, which produces work that only someone with your specific expertise could have created.

This is the difference between AI that makes you faster and AI that makes you more powerful. Faster means you do the same work in less time. More powerful means you do work that wasn't possible before.

I asked myself this question at the beginning of every project in the book. Each time, the answer redirected my approach. For the publishing calculator, what I knew that the AI didn't was how authors actually think about printing costs — which questions they ask, which numbers they get wrong, which decisions keep them up at night. That knowledge shaped the tool's design in ways that generic AI output never would have.

For the book itself, what I knew was thirty years of watching organizations fail at cybersecurity and succeed at creation. The AI knew more facts than I did about almost everything. But it didn't know what those facts meant, in context, to the people I was writing for.

Start there. Before the project. Before the prompt. Before anything. Figure out what you bring that the AI can't, and then build from that foundation.

Everything I teach about AI creation starts with this question. It's the foundation the rest is built on.