What I Saw and Why I Left
Seven years ago I had a clear picture of something that didn't exist yet. Not a product. A capability. The friction was the problem, and it was solvable.
Seven years ago, before anyone had heard of ChatGPT, I had a clear picture of something that didn't exist yet.
It wasn´t a product, it was a capability. The ability to turn an idea into something real without the brutal, mechanical friction that sits between the thought and the output. I didn't have words for it then. I just knew the friction was the problem, and that it will become solvable.
Speed
Velocity, not shortcuts
People who know me know my reputation as a designer was built on one thing: speed. Not shortcuts. Velocity. My previous life as a visual designer was a marathon of long hours and deliberate practice. Years at a screen, building the kind of muscle memory where the tool disappears and the thinking goes straight to the canvas. The pain of that mechanical work, combined with the hard limits of the technology, led me to a conclusion most designers never reach. I had to stop designing. Not because I'd fallen out of love with the craft. Because I'd hit the ceiling of what a human being with a mouse and a keyboard could produce. And that ceiling was nowhere near what I could see in my head.
The obsession didn't disappear. It changed shape.
Building
I started building tools
First for myself: small automations, templates that thought for themselves, systems that could carry more of the mechanical weight. Then for others. Then for the problem itself. Along the way, a friend pointed me toward something I'd been circling without naming: the validation loop. Build a small thing. Measure what happens. Learn from the gap between what you expected and what you got. Then build again. So that's what I did. Prototypes for on-brand video automation. Motion systems that generate themselves. Layouts that compose from rules instead of from scratch. Each prototype was a test. Each test produced principles. Each principle fed the next build.
Through that divergence: the mistakes, the dead ends, the rebuilds. I extracted something reliable. A set of architectural principles for turning brand intelligence into automated output. Not prompting an AI and hoping. Programming the logic of how a brand works into something a machine can execute with precision.
Years of that loop, and something real emerged.
Syvon
An AI ecosystem for creative production
Syvon is an AI ecosystem for creative production. It takes a brand system (the components, the rules, the logic) and produces finished work across formats, at a scale no team can match manually.
It's still early.
The system is working. The product is in beta. But it marks a line I can see clearly now: the end of one career and the start of something I've been building toward for a very long time.