Thoughts
/Design & ProductDesign Systems: Now More Than Ever
AI changed who builds interfaces. Design systems changed whether those interfaces look like yours or everyone else's.
Abdulqahar
Product Designer & Design Engineer
You know that thing where every AI-generated landing page looks... the same? Purple gradients. Rounded fonts. Identical card layouts. You can spot them instantly. Not because they look bad, but because they look like everything else.
And that gap between "fast" and "consistent"? That is exactly why design systems have never been more important.
The Nomba Moment
A few weeks ago, the Nigerian tech space had a field day. Nomba, a well-known fintech company, launched a redesigned website, and people immediately started asking: was this vibe coded?
The conversation on X was telling. Some called it out. Some defended it. But the real takeaway was this: people could tell. Whether it was actually vibe coded or not, the fact that the question was even asked says everything about where we are right now.
P!
@iPariola
Was Nomba's new website vibe coded, or they're just bored? Wetin be this?!
3:14 PM · Mar 23, 2026
Replying to @iPariola
Fola Olatunji-David
@folasanwo
What's actually wrong with this site? Genuinely curious. It's a functional site that passes its message cleanly. I'm a (very happy) Nomba client, and as long as my money dey intact and the buttons are where they should be, it works!
3:39 PM · Mar 15, 2026
Replying to @folasanwo
P!
@iPariola
My point? Their old site was way better.
3:39 PM · Mar 15, 2026
This is the tension in a nutshell. On one side: "it works, the message is clear, what's the problem?" On the other: "their old site was way better." Both are valid. But they highlight a gap that a strong design system would close.
Whether Nomba's redesign was a marketing stunt, a vibe-coded experiment, or just a bold creative choice, the conversation it sparked proved something important: people notice when a product loses its visual identity. And in a world where AI can generate a full page in seconds, that is going to keep happening unless teams invest in the guardrails.
Design systems used to be instructions for people. Now they are also instructions for machines.
The Problem Nobody Saw Coming
The barrier to building interfaces basically disappeared overnight. Anyone can generate a full page from a text prompt. Founders are shipping landing pages before lunch. Teams are prototyping at speeds that would have seemed absurd two years ago.
But when everyone builds from the same defaults, everything looks the same. Without clear design rules, AI just gives you its best guess. And its best guess looks like everyone else's best guess.
Companies are catching on. Tools like Google Stitch, Lovable, and Figma Make now let users set up colors, fonts, and spacing rules before they start building. Mini style guides baked into the setup process. That is not just a feature. That is an admission that AI without design guardrails produces generic work.
There Is a New Player at the Table
Traditionally, design systems served two audiences: designers and developers. Now there is a third: the AI agent. It does not sit in your design reviews. It does not absorb your brand by osmosis. It has zero taste. It only knows what you explicitly tell it.
Designers
Define the vision, rules, and intent. They decide what the brand should look, feel, and behave like.
Developers
Translate design into code. They build the components and systems that bring the vision to life.
AI Agents
Generate code and layouts from prompts. They follow rules well, but only the rules you actually give them.
So if your design rules are vague, the AI fills in the blanks with defaults. If your component guidelines are loose, it improvises. And "AI improvising your brand" is how you end up with that purple gradient problem.
AI improvising your brand" is how you end up with that purple gradient problem.
How My Process Has Evolved
My traditional workflow was thorough, reliable, and 100% manual. It has not gone away, but I have added a new layer: a reference file that tells AI tools exactly how to work with the system. Think of it as an instruction manual written specifically for machines.
Before
Every step: fully manual
After
Human thinking + AI execution (This will continuw to evolve, but you get the idea)
The key shift is where the effort goes. Less time doing repetitive work. More time making decisions, reviewing what the AI produces, and refining the system itself. The thinking has not been automated. The busywork has.
AI Is Not Magic
It would be easy to make this sound seamless. It is not. Here is what actually happens in practice:
Things get lost in translation
AI tools have limited memory. Complex components cause the tool to lose track halfway through. For companies, every redo costs real money.
Rules get skipped
You can write the most detailed instructions imaginable. AI will still occasionally pick the wrong color or invent something that does not exist in your system.
Tool-switching adds friction
Different AI tools are good at different things. Moving between them introduces overhead, not unlike switching between design tools you already use.
Human review is non-negotiable
Every piece of AI-generated work needs a human eye. You move faster, but you cannot stop paying attention. Intentionality still requires judgment.
The honest math: AI makes you faster, but the human effort does not disappear. It shifts from "doing the work" to "directing and reviewing the work." From building to thinking. From writing code to writing better instructions.
Design and Development Are Blurring Together
When your design decisions live inside your Figma file as structured, reusable values... when your component guidelines are clear enough for both humans and machines to follow... the line between "designer" and "developer" starts to dissolve.
A designer with the right system can go from concept to working component without touching code manually. A developer with the same system can generate pixel-perfect UI without opening Figma. And an AI tool, properly guided by the system, can help with both.
The design system is the convergence point. Not the tools. Not the AI. The system itself.
The Takeaway
When AI is generating interfaces at scale, the design system is the only thing standing between a product that feels like yours and one that looks like it was assembled from a template.
If you are shipping products in 2026 without a design system, you are not just slowing yourself down. You are handing your brand identity over to whatever the AI thinks looks good.
What a time to be alive.