From Pixels to Prompts

As a designer, I never thought I'd get away from the pixel by pixel craft. I remember the days of MS Paint, creating art dot by dot. I grew up in the days of Geocities, MySpace, and connecting to AOL through a dial-up connection. The edge of analog and digital blending together. I had a taste of growing up in the 1900s.

Being connected to the internet at a young age, I grew up morphing my physical world into a digital one. I coded HTML into my MySpace page. I used Dreamweaver to build websites. I graduated to Photoshop, Illustrator and InDesign. Tools that let me manipulate the sky so it popped more, clone a blemish-free spot over a blemish, vectorize a photo into a logo. The same things that Facetune does in a couple of taps today required a paid license, hours of practice, and actual technique back then. Those were the tools of my trade for over a decade as a graphic designer. Every one of them put me closer to the pixels. Direct contact with the thing I was making.

A lot of these tools did the same things. We just kept repackaging them. Dreamweaver became Webflow. Photoshop became Figma. "Web design" became "product design" when most designers are really doing visual UI and interaction design, which is a skill wholly itself. The titles change, the tools change, but the core work has been remarkably consistent.

What keeps shifting is the distance between you and the thing you're making. In the early days, you wrote raw HTML. You had direct contact with the code. The web looked like fever dreams because people had full control and wild imaginations. Brutalist aesthetics with fifteen fonts on one page. Then WYSIWYG editors came along. Dreamweaver, WordPress. Suddenly you could drag and drop, click a button to move an object. You didn't need to know what was underneath. The visual layer became the interface and it felt like magic. You were still touching the pixels, just through a different layer.

And every time a tool got good enough, it got turned into a product.

As I moved into UX, I saw this happen in real time. InVision, Adobe XD, Sketch, and the sweetheart of the group: Figma. When I landed my first UX role at an agency, I didn't even know how to use it, but that's the bet the agency placed on tooling. After juggling so many tools over the years, it was a relief to settle into one that met most of my needs with a gentler learning curve. Its usability proved itself and it became another tool of my trade.

I could gush about Figma. It helped me move quicker. Design systems, components, global assets. Breaking content and patterns into reusable pieces that helped maintain standards and quality. It was a warm and fuzzy time. Things were simple and usable.

Then it started morphing into a product. The same pattern again.

The UI started shifting. This panel moved here, that got hidden there. Relearning where things live is normal with software. Things get updated. With Adobe, I remember new features that enhanced my existing workflows. Even when they started drizzling in AI, it was pretty neat. With Figma, I felt something different. A shift towards monetization. Changes in pricing. Slides. Make. AI this and that. It felt bloated, confusing, and I hate that a once darling of a tool became another victim of product led growth. There's a constant push to be "productized" and "growth-oriented" and all these buzz words that keep getting invented. There's always something new you need to learn before you get automated by AI. You can hear about it on Lenny's podcast or sign up for a class on Maven. If you couldn't tell, there's a lot of noise in my feed, my algorithm, my notifications, my wall, my inbox. The list goes on.

And now with LLMs, the cycle has come full circle in a strange way. The pixels are gone. Instead of touching the thing directly, I'm describing what I want in a text box. Writing prompts instead of pushing pixels. The GUI has been stripped back down to a chat interface. We went from the wacky handwritten world of HTML to templatized drag-and-drop, and now we're back to describing things from scratch in plain language. The output is cookie cutter. The control feels different. From pixels to buttons to prompts.

Coming from an era of simple software that was designed to help users do something, watching each generation of tools follow the same arc is frustrating as both a user and a designer. A tool starts simple. It gets good. It becomes a product. It gets bloated. Something new replaces it. And we start over.

I will eventually talk about my own experience with vibe coding, but as an observer and relic of the analog days, the vibe is shifting. It's important to acknowledge that shift to understand its impact and what it might mean. As a designer, I think about systems and how things ripple out. So I want to take this opportunity to not only catalog my experience with vibe coding but learn about it from all angles.

In this age of high frequency everything, do the tools make the maker or does the maker make the tools? How does vibe coding change the way we learn, think, communicate? How might we benefit from it and at what cost? Who am I becoming from using this tool?