If you ask a designer what they want, they will say faster horses.

I wrote about this in 2014. 12 years later, we are still shoveling. Put the shovel down.

Twelve years ago, I wrote an article called “No one said they wanted faster horses, they wanted less horseshit.”

The point was to show that had it really happened (it did not), people would not have asked for faster horses (as slow horses was not a problem), they would have asked for the problems they had: lack of automobile financing, or if you asked before Ford’s heyday, the 9 feet of manure burying London.

And yet… this week, I watched it happen again.

Someone posted a viral experiment on LinkedIn. They had connected Claude Code (a tool that builds working software from natural language) to Figma (a tool for drawing pictures of software) via a browser console hack. The result: Claude could now programmatically create UI components, variants, and states inside Figma. The design community lost its mind. “Huge.” “A new paradigm.” “This changes everything.”

And I sat there, looking at my screen, thinking: they strapped a jet engine to a horse. And then I saw another post, and another. I won’t link to them; this isn’t about dunking on someone; experiments are encouraged.

But… Faster horses. I’ll be darned; people WILL ask for faster horses 🤦‍♀️.

We shall continue to flagellate the deceased equine for eternity.

The Original Horseshit

For those who have not read the original (link here), the short version: the famous Henry Ford “faster horses” quote is misattributed, misunderstood, and misused. Ford probably never said it. And even if he had, the point was never about what people asked for. The point was about what they were drowning in.

• In 1894, The Times of London predicted that within fifty years, every street in the city would be buried under nine feet of horse manure.

• New York had 100,000 horses producing two and a half million pounds of the stuff every single day.

• Working horses lived about three years before collapsing in the streets, where their carcasses were left to rot until they could be sawed apart for removal.

• The 1898 international urban planning conference in New York broke up after three days of its scheduled ten. They could not find a solution.

Ford did not build a better horse. He did not invent a faster manure shovel. He made horses irrelevant. Parallel innovation. The problem was never that horses were too slow. The problem was that horses were horses, and everything that came with them was killing cities.

I wrote that twelve years ago. The thesis holds up better today than it did in 2014. The horses just look different now. Our destination may be the same, but our vehicle has changed. And we cannot keep investing in the horses of the past.

The 2026 Horseshit Problem

Figma, and other design tools of the pre-AI era, are the horse. Not because Figma is bad. Figma was genuinely revolutionary. It took the Sketch/InVision/Zeplin fragmentation nightmare and unified it into a single, collaborative, browser-based design tool. For years, it was the best transportation available. I am not here to trash Figma.

But Figma’s output is a picture of software. Not software. Figma is Illustrator++ that can also give you some CSS you can give to another person who can then put it in another system that then is consumed by another system… it’s a whole wagon train of horses all lined up like we’re crossing the plains.

And the manure is everything between that picture and the thing it depicts. The handoff document. The redline. The “can you make it match the mockup” meeting. The specification. The design token export. The developer squinting at a Figma frame trying to determine whether that gap is 8 pixels or 12. The filed ticket that reads “padding does not match design.” The six rounds of visual QA where a designer and a developer argue about the distance between what was drawn and what was built.

That is the horseshit. The translation layer between intent and artifact. We have been wading through it for so long that it smells normal. We budget for it. We schedule sprints around it. We have built entire job titles around managing it. And for years, that was fine. The horse was the best we had.

Remember: 9 feet deep. That is a conversation that occurred in real life. “In 1894, The Times of London predicted that within fifty years, every street in the city would be buried under nine feet of horse manure.”

It is not anymore.

The Jet Engine on a Horse

So here is what happened with that LinkedIn experiment. Claude Code can build a working dashboard. Real components. Real state management. Real interactions. Deployable to production. It can build the actual thing, not a drawing of the thing.

And someone took that capability and used it to draw rectangles inside a rectangle-drawing program. Through a browser console hack. Adding two extra abstraction layers to get further from the artifact, not closer.

The curiosity is admirable, and I do mean that genuinely. The technical ingenuity of getting Claude to talk to Figma’s plugin API through Chrome DevTools is clever work. But the direction is backwards. It is using a car engine to make a horse run faster. The destination has not changed. You are still riding a horse. The manure is still accumulating. You are just producing mockups at higher velocity, and those mockups still need to be translated into working software by someone else, who will still squint at your frames and file tickets about padding.

See, when you spend years mastering Figma (or Sketch, or whatever your canvas of choice is), the tool becomes part of your professional identity. “Figma expert” is a real credential. It is on resumes. It is in job descriptions. And when someone suggests the intermediary tool might be becoming optional, it feels like an attack on your expertise.

But your expertise was never in the tool. It was in the thinking. The tool was just the vehicle, and now if we try to apply the capabilities that a future of AI-enabled design and development offers, retrofitting into the outdated processes is adding a jet engine to a horse.

We see our destination, it is the same as it was. But we have new vehicles, entire modes of transportation.

The Comfort Trap

I want to be honest about something, because I think the “just learn to code” crowd is not being fair. This transition is genuinely hard. Figma is not just a tool. It is where designers think. The canvas is spatial, visual, immediate. You can see the thing taking shape under your hands. The feedback loop is instant. It feels like creativity, because for many designers, it is how creativity works. It is the medium they trained in, the surface they are fluent on, the place where intuition shows up.

Code does not feel like that. Not yet. Maybe not ever in exactly the same way. And asking someone to give up the canvas where they do their best thinking is not a small ask. It is not trivial. Ten thousand hours of mastery. Genuine expertise. Real craft.

I get that. We have all been there. We have all held onto a workflow because it was familiar, because we were good at it, because starting over felt like admitting the previous decade was wasted. It was not wasted. The design thinking you developed during those years is exactly what matters now. The vehicle is changing. The driver is not.

But (and this is the part we need to sit with) staying comfortable is how you end up neck-deep in manure, wondering why nothing smells different even though your shovel is really, really fast now. Your intention as a designer is the same, but the tools have changed. And just like you probably haven’t opened up CorelDRAW, QuarkXPress, or PageMaker Pro in a while, you’re not going to really open up Figma in a while either.

What Less Horseshit Looks Like

A designer sits down with Claude Code. She is building a dashboard for monitoring API health. Three panels. A sidebar with navigation. A notification system that surfaces the right alerts at the right moment.

But she does not describe it all at once and wait for a prototype to appear. That is vibe coding. That is talking to a microphone and hoping. Instead, she works methodically. Component by component. Decision by decision.

First, she builds the layout scaffold. Just the bones. She reviews it, understands what was generated, asks why the grid was structured that way. She learns something about CSS Grid she did not know before. She adjusts the approach based on what she now understands.

Then she builds the sidebar. Not “add a sidebar.” She thinks through the navigation hierarchy first, decides what belongs at the top level versus nested, considers the user’s mental model. Then she builds it, reviews it, iterates. The first version is too heavy. She does not say “make it lighter” and wait for magic. She looks at the code, sees that the padding and font weights are creating visual bulk, and makes specific requests. She understands the change she is asking for.

Then the notification system. She does not say “add notifications.” She thinks through the alert taxonomy first. What is critical? What is informational? What deserves a badge versus a toast versus a quiet log entry? She builds the state management with intention. She asks Claude to explain the pattern it chose. She agrees or redirects.

Forty-five minutes later, she has a working application. Real components, real layout, real interactions. She can click through it. She can deploy it.

But the difference between this and vibe coding is not the output. The difference is that she understands every piece of what she built. She made deliberate choices at every step. She learned while she worked. The AI was a collaborator, not an oracle.

No handoff. No redline. No specification document. No “is this 8px or 12px” conversation. No developer interpretation. No six rounds of visual QA. No filed tickets about padding.

The thing is the thing. Intent moved directly to artifact, through a process that was rigorous, not magical. I call this Zero-Vector Design, and it is not a shortcut. It is a discipline.

The manure disappears. Not because we found a better shovel. Not because we asked nicely and something appeared. Because we stopped needing horses, and we learned to build the car ourselves, piece by piece, understanding every component we installed.

Let Go of the Reins

In 2014, I ended the original article with this observation:

“People may never know what they want, and focusing on the problems they are specifically having might take us down paths towards iterating on ‘good enough’ instead of understanding the base emotional problem.”

Twelve years later, the base emotional problem is identical. We are still iterating on good enough. We are still focused on the specific problem we are having (how do I draw mockups faster?) instead of asking whether we need to be drawing mockups at all.

The question is not whether the jet engine exists; it does. Now, it is whether you are going to strap it to a horse, or let go of the reins. And look, I know what letting go feels like. It feels like vertigo. It feels like starting over. It feels like your expertise is evaporating under your feet. I have done it 3 times: first as a web designer who was late to shift to UX design, then as a service designer who tried and failed to actually get that approach to take hold at work, and now again as a Zero Vector designer trying to practice what I preach.

But it is not real vertigo, and you’re not falling. Your design thinking, your systems thinking, your instincts, your judgment, your ability to see what users need before they can articulate it: none of that requires Figma. None of it requires any specific tool. All of it transfers. Every hour you invested in understanding how people interact with interfaces, what makes a flow intuitive, when to add friction and when to remove it: that knowledge is more valuable now than it has ever been, because you can finally apply it directly instead of losing half your intent to translation.

You are not losing your craft. You are finally getting to use all of it at once. The horseshit was never inevitable. It was just the cost of the best transportation available at the time. The transportation has changed. The cost is no longer required.

So, stop shoveling and start driving.

---

Source: me, twelve years ago: “No one said they wanted faster horses, they wanted less horseshit“