Wireframes Are Dead. Just Vibe Code It.
Wireframes are dead.
I know. That sounds like heresy if you went to design school. I graduated last year from the Stanford d.school, and for the past eight months I've been doing the opposite of everything I was taught.
No paper prototypes. No wireframes. No low-fi before high-fi.
Just prompts and builds and vibes.
It felt like cheating - until I realized the d.school taught me something deeper than the artifacts. But I'll get to that…
Wireframes were sacred - the low-fidelity rectangles that proved you thought before you built. They were how we showed we cared.
But here's the thing: I haven't made a wireframe in months. Neither have most designers I know who are actually shipping. Instead, we're typing prompts into v0 or Cursor and watching real UIs materialize in seconds. We're building the thing to figure out if we should build the thing.
And it works.
So if we can go from idea to working prototype faster than we can sketch boxes on a whiteboard - what are wireframes even for anymore?
What Wireframes Were Actually Doing
Before we bury them, let's give wireframes their due.
Wireframes were never really about the boxes.
They were a thinking tool disguised as a deliverable. They forced you to slow down when building was expensive. They externalized half-baked ideas so teams could argue about structure before anyone wrote a line of code. They were a communication layer - a way to say "here's what I'm imagining" without committing to pixels or logic.
The artifact wasn't the point. The conversation was.
When a full prototype took weeks and real code took months, wireframes made sense. They were cheap insurance against building the wrong thing. You'd spend two days sketching screens so you didn't waste two months engineering something nobody wanted.
That math made sense. Until it didn't.
Taste and judgment are the new bottleneck - not your ability to code
Building got cheap. Absurdly, almost-free cheap.
Cursor hit $500M ARR faster than any SaaS product in history. Lovable went from zero to $200M ARR in a year, with users creating 100,000 new products daily. Vercel's v0 has 3.5 million users generating full-stack apps from plain English. AI now writes 46% of all code among Copilot users.
The implications are wild.
Uber reportedly cut design concept testing from six weeks to five days. Agencies are building production websites in an afternoon. A developer on Reddit described delivering a client's entire site - something that "normally would take days or even weeks" - through a single prompt session.
When Andrej Karpathy tweeted about "vibe coding" in February 2025 - describing a workflow where you "fully give in to the vibes" and "forget that the code even exists" - it got 4.5 million views.
Not because it was new.
Because it named something people were already feeling.
The cost of making collapsed. The cost of deciding what to make didn't.
For a long time, designers felt blocked by one thing: not knowing how to code. That's no longer the bottleneck.
The bottleneck is judgment.
Think about what's actually scarce now:
- Knowing what to build first
- Deciding what's worth testing versus ignoring
- Spotting bad assumptions before you've invested weeks
- Recognizing when something "works" but shouldn't exist
- Stopping at the right time
AI removed the cost of building. It exposed the cost of deciding.
Here's the uncomfortable part: taste is now a real skill. Not "I have good taste" as a vague self-compliment. Taste as pattern recognition. Taste as knowing what users will accept. Taste as sensing when something is off, even when it technically works.
Taste is just judgment under uncertainty. And uncertainty is the only constant now.
The designers I see thriving aren't the ones who learned Cursor fastest. They're the ones who brought the strongest opinions about what those tools should make. They know when to stop generating and start shipping. They know when a prototype needs another hour and when it needs to be killed.
Dylan Field, Figma's CEO, put it directly: "As AI transforms tech, a designer's judgment, taste, and agency will matter more than ever."
He's right. The filter changed. It used to be: can you build this? Now it's: should you?
So what did I learn during my time at the d.school?
Here's the thing I finally understood about my time at the d.school: they weren't teaching me to make wireframes.
They were teaching me to think before I build. To articulate what I'm trying to make. To state intent clearly enough that anyone - a teammate, a user, now an AI - can understand what we're making and why.
The artifacts were just the vehicle. The skill was clarity. That skill didn't become obsolete. It became the whole job.