You know what’s ridiculous? We have one of the most powerful technologies software has ever seen, and we’re building chatbots and command line interfaces. People talk about the “iceberg model” of AI capability, but this is like showing users an ice cube when an iceberg the size of the one that killed the Titanic is lurking underneath. You can’t see the potential in a tiny text box.
I think I know why we’re seeing this failure of imagination. We’ve left design out of the discourse.
I read a dozen AI newsletters and listen to podcasts constantly—this is the most exciting moment in software since the web, and I’m trying to keep up. But I keep noticing a pattern in what people talk about: product managers vibe-coding prototypes, software engineers worried about their jobs, infrastructure scaling challenges. Design, the third leg of the product trio, is just… missing from the conversation. We’re in the middle of the biggest shift in software creation since the internet, and we’ve somehow decided that the discipline responsible for understanding humans doesn’t matter anymore. When I say design, I don’t mean making things pretty—Steve Jobs said it best: design isn’t just how something looks, it’s how it works. Design is the discipline of learning from humans, understanding what they actually need rather than what they say they want, and translating that understanding into something valuable that a business can capture.
Now we can make things incredibly fast. An engineer with Claude or Cursor can spin up a working app in an afternoon. That’s amazing, but it’s also revealed the real problem: we can build anything, and we have no idea what to build.
The Taste Problem
People talk about “taste” as if it’s some mystical quality certain founders possess, something you either have or you don’t.
Taste is not innate. Taste comes from experience—from interacting with hundreds of products critically, from years of observing what works and what fails and developing the pattern recognition to know the difference. When a founder has “product sense,” what they have is design training they picked up somewhere, consciously or not. They learned to see.
We could teach this. We could hire for it. Instead we’re acting like it’s a personality trait.
AI Is a Mediocrity Generator
AI is trained on everything—the good, the bad, and the profoundly mediocre—so when you ask it to design something, it gives you an average: the thing that looks like everything else.
Vibe-code something right now. It’s fine. Functional. Probably uses Tailwind, has some nice drop shadows. But does it demand your attention? Does it communicate what makes your product different? Does it help users understand the power of what you’ve built? Or does it look like a template—bland, competent, forgettable?
When everyone has access to the same AI tools generating the same average outputs, the companies that win will be the ones that can direct those tools toward something actually distinctive, something that connects with humans. Design matters more now, not less.
What Design Actually Does
Let me break this down, because I think people genuinely don’t know what UX design contains.
Interaction design figures out how things behave—what happens when you click, what’s the flow, what mental model does a user build about how your software works. This matters enormously right now because the chat interface most AI products use tells users almost nothing about actual capabilities; there’s a reason people use AI for simple questions when it can do enormously complex work.
Information architecture organizes information so humans can learn and comprehend, helping users navigate thousands of features without drowning in feature soup.
Interface design makes the possible visible, creating affordances that show users what they can do. The current state of AI interfaces is primitive—a text box and some suggestions—and we’re barely scratching the surface of what these interactions could be.
Graphic design communicates vibes and manifests your brand—the difference between “this company gets me” and “this is generic garbage.”
This can be one person with deep training and broad experience, or a team of specialists, but the knowledge has to exist somewhere in your organization. Otherwise you’re just regenerating the same mediocre patterns the AI learned from.
And new competencies will emerge. I can see a future where designers take their insights and experience and directly create working prototypes that tell a better story of what the product should be than any Figma wireframe ever could.
You Need Humans Who Understand Humans
AI can generate a hundred ideas, a thousand, combining and recombining everything it’s learned into new configurations. What it can’t do is know which idea is right.
That takes someone who understands humans—who has watched real people struggle with real interfaces, who knows that what people say they want and what they actually need are often different things. AI doesn’t know your users are exhausted mothers checking their phones while making dinner, or overwhelmed executives with fourteen tabs open, or anxious college students at 2am.
A good UX designer translates between the human world and the technical world—they’re the reason your product might solve a real problem instead of an imagined one. And when building becomes trivially easy, when any kid with a laptop can ship an app that does what yours does, that understanding becomes your competitive moat.
What’s left when engineering is no longer your advantage? The ability to see what others miss and build what others can’t imagine. That’s design.
What To Do About It
Learn how to hire a real UX designer—not someone who can “make it pretty” but someone who can tell you what to build and why. Explore interaction patterns beyond the chatbot, because a text box is not the final form of human-AI interaction. Include designers in the conversations about product direction, not just execution.
Here’s what’s thrilling: nobody’s doing this well yet. The new paradigms haven’t been defined. We’re at the moment before the patterns crystallize, when everything is still possible. We need design’s knowledge to figure out what these new interactions will look like and how they’ll behave.
This is an extraordinary time for product, for engineering, and for design. But only if product remembers the power of their partner, design. The AI revolution isn’t about making things faster. It’s about making better choices about what to make.
That’s a design problem.