For two years I've used AI in real consulting work. Contracts, client briefs, research, strategic analysis in fields where I have thirty years of professional standards to measure the output against. That held up.

Then I tried it for social content. What came out was technically correct and sounded like it could have come from anybody. I spent more time editing it back to something real than I spent on the original idea.

turning point].

I was making the exact mistake I'd spent three decades helping organizations avoid. Asking for professional work without providing a professional brief first.

So I built one. A single prompt that turns Claude into the interviewer. It asks you eight questions, one at a time, builds your voice skill file from your real answers, and installs it automatically. After that, you type /my-voice and Claude loads your full profile before writing a single word.

The first session I used it, I didn't touch the tone once. I refined the content. The voice was already right.

If your AI content sounds like everyone else's, you're probably doing what I did. Asking Claude to write without ever telling it who you are. Giving it that brief once is the only thing that needs to change.

AI produces generic content because it has no brief. This gives it one, permanently.