There was a time, not so long ago, when UI design happened without Figma. I spent a good chunk of it in the bullpen at Sandwich, making other people's unfinished products look real on camera.
It is genuinely quaint now to remember that there was a time before Figma. Not before design, before Figma specifically, before the browser tab that every product team now lives inside all day. When I started doing interface work, the tools were Photoshop and Illustrator, bent into shapes they were never meant to hold. Then Sketch arrived and felt like a revelation. Then Adobe rolled out XD and we all gave it a real shot. I was bullish enough that I branded my own outfit XDHD, a little nod to the tool I assumed was going to win. XD came and went. Figma ate the world. And somewhere in that churn, a strange and specific job existed that I happened to be very good at.
Here's the setup. Sandwich, the production studio out of LA, built its reputation making launch films for technology companies. If you've watched a startup's product video in the last fifteen years and actually laughed, actually understood what the thing did, and actually wanted it by the end, there's a decent chance Sandwich made it. They more or less invented a genre: the smart, funny, self-aware tech explainer that treats the audience like adults.
But there was a problem hiding inside every one of those shoots, and it was always the same problem. The product wasn't done.
Think about the timing. A startup raises a round, books Sandwich to make the film that will announce them to the world, and schedules the shoot for, necessarily, before launch. Which means before the product is finished. Sometimes well before. The app on the phone in the hero shot might be a clickable prototype held together with tape. Sometimes it was pre-alpha. Sometimes the interface we needed to show gracefully solving the user's problem did not, in any working sense, exist yet. The company knew what it wanted to be. It did not yet have the screens to prove it.
So somebody had to make those screens. Somebody had to take a half-built, half-imagined product and produce the polished, legible, beautiful UI that the camera needed, the version of the product the company was promising it would become. That somebody, for a good stretch of years, was me. Me, or me and one other guy, rotating through the bullpen at Sandwich, picking up these asks as they came.
The craft of it was particular and I loved it. You'd get whatever existed: a Figma file that was 40% real, a founder's voice note describing the flow, a screenshot of a competitor, sometimes just a conversation: and you'd have to invent the rest in a way that was believable, clear enough to read in a two-second cut, and consistent with the brand and the product's actual logic. Then you had to make it live in the film. Sometimes that meant compositing the UI literally onto the iPhone in the actor's hand, tracked and lit so it looked like the phone was really doing that. Sometimes it meant lifting the interface off the device entirely: abstracting it, exploding it, floating the screens and components around the actor and through the scene so the audience could actually see the idea instead of squinting at a one-inch rectangle.
That second move, pulling the UI off the glass and into the world of the scene, became part of what people now recognize as the Sandwich style. I want to be careful and honest here: I did not invent that, and Sandwich is full of brilliant people who shaped it. But I had a tiny, real hand in crafting how it worked, over enough films that the patterns became muscle memory. How do you make floating UI feel like it belongs to a person and not a tech demo? How much do you abstract before it stops reading as the actual product? How do you fix a flow that doesn't make sense yet without lying about what the product does? That was the job.
And here's the part that makes it a story worth telling now, rather than a nostalgia trip about old tooling. For years, this wasn't a role. It wasn't a department. It didn't have a name. It was just "can the motion guy figure out the screens." It existed in the cracks between the design world (which made the real product) and the film world (which shot the commercial), and nobody owned the cracks. You either had someone who could stand in both rooms at once, or you had a beautiful film with an unconvincing product at the center of it.
It wasn't until around 2020 or 2021 that I watched this congeal into something Sandwich and others started treating as a proper discipline: a named GFX ask, a thing you could request, scope, and staff, rather than a favour you hoped the motion person could pull off. The work had been real for a decade. It just finally got a label.
That's the seed of a course I'm building now, called GFXR, "The GFX Fixer." Because that's what the job actually is. You are the person who gets handed the unfinished, the half-baked, the pre-alpha, the "we'll have it by the shoot, promise," and you make it whole on screen. You fix the GFX. And it turns out almost nobody teaches this, because for most of its existence it wasn't a thing you could point at. It lived in the heads of a handful of people who happened to be fluent in interface design and motion and compositing and the specific grammar of how a product reads on camera.
The deeper reason I think it's worth teaching: this skill is getting more valuable, not less. Every company is a software company now, every launch needs a film, and the gap between what the product is on shoot day and what the product promises to be is permanent. It's structural, it's not going away. Someone always has to stand in that gap and make the promise legible. The tools changed, Photoshop to Sketch to XD to Figma to whatever's next, but the gap didn't. The fixer's job didn't.
I think about it sometimes as the inverse of VFX. A visual-effects artist makes the impossible look real. The GFX fixer makes the not-yet-real look inevitable. You're not faking a dragon; you're faking a product that will, in six months, actually exist: and your job is to make it look so right that when it ships, it matches the film, not the other way around. The best compliment on this work is that no one ever notices it. They just believe the product.
So: there was a time before Figma. I spent a good chunk of it in a bullpen, fixing other people's screens so a camera could believe them. It was one of the best apprenticeships in interface thinking I could have asked for, precisely because the stakes were so concrete: it either read on camera or it didn't, in two seconds, no second chances. GFXR is my attempt to finally write it all down.
[DRAFT NOTE: be careful naming specific clients alongside "half-baked / pre-alpha", describe the phenomenon, anonymize the embarrassing specifics. Naming Sandwich and the general client type is fine; pairing a named startup with "unfinished product" is not.]