For centuries, astronomers used Ptolemy’s geocentric model, with the Earth at the center of the universe. When astronomical observations didn’t match the theory, scientists didn’t question the theory. They added epicycles: circles within circles that explained why planets sometimes appeared to move backward.
As astrolabes and early telescopes improved, astronomers began to notice inconsistencies. But instead of questioning the premise, what did they do? They added more epicycles. The mathematics became incredibly polished, complex, and aesthetically satisfying to the scholars of the time. The mathematical tool distorted the astronomers’ thinking so thoroughly that they preferred to adjust reality to fit the formula rather than accept the obvious. They were doing the pixel-pushing of the cosmos. For the academics of the era, it was progress and it justified the observed world.
It wasn’t. It was a tool distorting the thinking of an entire generation of scientists.
Figma and the Epicycles of Design
Over the past few years, I’ve watched a similar pattern unfold among designers.
The rise of Figma as the central tool and the maturity of design systems brought something extraordinary: ready-made components, shared tokens, consistency at scale. It was possible to build sophisticated interfaces without ever leaving the tool.
But something more subtle happened. When the component is already built, the temptation is to start assembling the interface before understanding the problem. The double diamond shrank. Low-fidelity sketches disappeared. Design became assembly.
And Figma’s technical limitations became invisible in a dangerous way.
Figma has poor support for responsiveness. It doesn’t natively support Flexbox or Grid. Animation is limited. At a time when the web was exploring parallax, advanced filters, and CSS-rich transitions, Figma couldn’t represent those solutions. And what the tool can’t represent stops being considered as an option.
Designers added epicycles: workarounds, nested instances, locked components to maintain consistency, UI templates crystallized in the name of design system fidelity. The mathematics got polished. The process drifted away from what the code and the product actually required.
Here’s an example that illustrates this precisely: multi-brand, multi-platform component libraries. The ideal solution would be a single component library where tokens apply and transform the component to serve multiple brands and multiple platforms. Figma simply doesn’t support the number of tokens needed at the final layer of component tokens. So designers built workarounds. We added more epicycles. The architecture got more complex to compensate for what the tool couldn’t do.
We’re fitting our needs into the tool’s limitations.
AI Is Going to Replace Us All. Or Is It?
Vibe coding arrived and forced something unexpected.
When a designer uses Claude, Cursor, or Lovable to build an app, the tool asks for different things than Figma ever did. It asks for rules. States. Logic.
What happens when the user clicks here and the API fails? How does this data behave on mobile? What’s the error flow if the input is invalid?
The designer is forced to think about information architecture, data models, and interaction flows before thinking about the color of a button. Function returns to the center. The UI becomes the last layer, not the first.
And with that, the constraint changed.
With Figma, the question was always: can the tool do this? With vibe coding, that question disappeared. The limitation stopped being technical and became creative and systemic. What the designer can think, the tool can build.
It’s the equivalent of replacing the geocentric model with the heliocentric one. It’s not that the Sun is better than the Earth. It’s that changing the reference point simplifies everything else.
Nicolaus Copernicus, and later Johannes Kepler, changed the game with a different tool: the heliocentric hypothesis and elliptical orbits. By shifting the focal point, putting the Sun at the center, hundreds of redundant equations and epicycles disappeared in the blink of an eye. Physics returned to its essential function: explaining the movement of the stars in the simplest and most elegant way possible.
But Copernicus Didn’t Solve Everything Right Away
A shift in models is neither automatic nor risk-free.
Building fast isn’t the same as building the right thing. When the friction to see something functional dropped to zero, the trigger to start building became too quick. The most real danger of vibe coding isn’t technical: it’s designers skipping the research and validation phase because building is more gratifying than asking questions. You get to the right solution for the wrong problem much faster.
The happy path isn’t the product. LLMs are excellent at generating the main flow. But a real product lives in the edge cases, error states, and accessibility. A designer without a strong technical background may not know what questions to ask the AI to cover those scenarios. The result looks functional on the surface and is fragile underneath.
UI isn’t cosmetic. Deferring visual polish to the end assumes any interface will do for validating logic. It won’t. The affordance of a button, contrast, typographic hierarchy, and animation feedback are cognitive cues about how the function works. Bad UI can invalidate a UX test not because the business logic is wrong, but because the interface failed to communicate that logic.
The Tool Is Not the Profession
Vibe coding doesn’t automatically return designers to the essence of the profession.
It reveals that the essence was always there, waiting. What changes is what the tool stops prohibiting.
Ptolemy wasn’t wrong because he was unintelligent. He was wrong because his model made it impossible to see what was right in front of him. When Copernicus shifted the focal point, hundreds of redundant equations disappeared. Physics returned to its function: explaining the movement of the stars in the simplest and most elegant way possible.
Design was always about solving problems. Not about aligning vectors.
What vibe coding is doing is removing Figma from the center. And when the tool stops being the center, the problem becomes the center again.
That’s not a new trend. It’s a return.