Creating an AI Trained on $500M+ of Copy [Not ChatGPT]
Over a decade ago, I got my start in marketing as a graphic designer building websites. Every project hit the same wall. I'd show the client a beautiful layout: great fonts, clean structure, compelling visuals... and they'd point at the placeholder text and ask: "So… what should we write here?" I didn't know. They're the business owner. Surely they know how to talk about their own business? But here's what got me confused (and frankly very frustrated): One website would convert and the other wouldn't. Both looked great. Same quality of products and design. The difference? The words on the page. That changed everything for me. So as a dyslexic graphic designer, I decided to learn "words" Fast forward and I'm still obsessed with the same problem, just at a different scale. Because when AI writing tools exploded, I thought we'd finally solved it. Finally a business owner could write the words to sell. The output was fine. Forgettable. We tried better prompts. More context. Longer briefs. The output got marginally better, but it still didn't sound like us. And more importantly...it didn't sell. And that's because of how AI is trained: the average of every website that exists (even those really bad ones from the 90s) So I asked: what if the AI wasn't trained on the average website? What if it was trained on a master? Ray Edwards has generated over $500 million in sales writing for Tony Robbins, Michael Hyatt, Jeff Walker, and Amy Porterfield. What he has isn't just skill, it's a philosophy of persuasion baked into every sentence. We didn't just feed his million dollar copy into a model (which we did...my wife @Noelle Switalski and I manually entered every line of copy...it took several months) but we also built his thinking directly into the AI's logic. The result? Copy that sounds like us. Emails, sales pages, scripts, all written by something that actually knows the difference between a feature and a feeling.