Good enough, isn't
· 8 min readHere's something I keep noticing in conversations with founders, product leaders, and hiring managers. There's this growing confidence that AI has solved design, that the tools have gotten good enough that you can prompt your way to a solid product, ship it, and move on to the next thing. And honestly, I get why people feel that way, because the output really is impressive, the layouts are clean, the spacing is considered, the type choices are reasonable, and everything works on mobile without anyone having to think too hard about breakpoints or grid math. But something is being lost in that confidence, and I think it matters more than most people realize.
Technology has always raised the floor
If you zoom out far enough, the history of web design is really the history of technology making it progressively harder to build something truly terrible. In the mid-90s, we were all constrained by the same primitive tools, building with table layouts and spacer GIFs and whatever system fonts the browser happened to support, and the result was a web where almost everything looked roughly the same kind of bad. The floor was low, the ceiling was low, and the distance between the worst site and the best site was surprisingly narrow, because the technology itself was the bottleneck, not anyone's skill or taste or ambition.
Then CSS and web standards came along around 2000, and suddenly you could separate structure from presentation, you could use real typography, you could actually express design intent in ways that weren't just hacks on top of hacks. The floor started to rise, not dramatically but noticeably, because now the constraint wasn't whether you could make it look good but whether you knew the new tools well enough to take advantage of them.
Quality scores reflect visual polish, usability, responsiveness, and content hierarchy.
Values are illustrative, the shape of the distribution is the argument.
- 1993HTML
The web's earliest era. Everyone was constrained by the same primitive tools: table layouts, spacer GIFs, system fonts. The floor and ceiling were both low, and the distance between the worst and best sites was surprisingly narrow.
- 2000CSS
CSS separated structure from presentation. Real typography arrived. You could express design intent without hacks, but the constraint shifted to whether you knew the new tools well enough to use them.
- 2008Frameworks
Good patterns became copy-pasteable. You didn't need to understand grid math to get a decent responsive layout. The middle of the distribution compressed hard as the gap between experts and tutorial-followers narrowed.
- 2016Systems
A competent team following established patterns could ship something that would've looked exceptional a decade earlier. The bar kept rising, the gap between teams kept shrinking, and good enough became genuinely good.
- 2023AI
AI didn't just make good design easier, it made bad design almost impossible. The floor rocketed upward, but the outliers moved slightly. What separates them now isn't production quality, it's imagination.
The next wave hit around 2008 with responsive frameworks, like jQuery and Bootstrap, this is where things got interesting because good patterns became copy-pasteable for the first time. You didn't need to understand grid math to get a decent responsive layout, you just needed to know which classes to apply. The middle of the distribution compressed hard during this period, because the gap between someone who knew what they were doing and someone who was following a tutorial narrowed significantly.
By the time we got to the component era, React and Figma and design systems and accessibility tooling, the floor had risen to the point where a competent team following established patterns could ship something that would have been considered exceptional a decade earlier. The mean kept climbing, the variance kept shrinking, and good enough became genuinely good.
Then AI arrived, and the floor didn't just rise, it rocketed upward.
The floor has never been higher, and that's actually the problem
Here's what makes this moment different from every previous technology wave. AI didn't just make good design easier to produce, it made it almost impossible to produce bad design. The floor jumped more in this single era than in all the previous eras combined, because AI autocorrects your worst instincts, applies sensible defaults, generates responsive layouts with proper hierarchy, and produces output that sits comfortably in the upper half of what humans were building five years ago.
And I want to be clear, I think this is genuinely wonderful. More people building better digital experiences is a net positive for the world, full stop. The democratization of design quality is something worth celebrating. But it creates a strategic problem that almost nobody is talking about.
When every product looks competent, competence stops being a differentiator. The companies that used to stand out because their interface was clean and their interactions were thoughtful now blend into a sea of clean and thoughtful, because everyone has access to the same models producing the same patterns drawing from the same training data. The bar for good enough rose, which means good enough no longer gets you noticed, it just gets you into the pack.
The thing that jumps out isn't the rising floor, it's what's happening at the top. The outliers, the products that stop you mid-scroll and make you rethink what's possible, barely moved. And that's the point. When the floor rockets up but the ceiling stays put, reaching the top doesn't get easier, it gets harder. Not because the bar raised, but because production is now table stakes. What separates the outliers was never how well the thing was made, it was whether anyone had the imagination to conceive it in the first place. That gap, between what AI can produce and what exceptional design leadership can envision, is the one worth paying attention to.
The data says design is flat, the data is looking backward
If you follow Lenny Rachitsky's hiring data, you've seen the chart, PM and engineering headcount growing while design holds flat or declines as a share of product teams. And the natural reading of that data is that the market has spoken, that companies don't think they need more design. Which is true, as a description of what companies are doing right now. But it's a snapshot of where hiring has been, not where value is going.
What that data actually tells us is that companies are responding to AI by doubling down on an old mental model of what design does. They're pricing design as production, and since AI makes production cheaper, the math says you need less of it. That's a perfectly rational conclusion if you believe design is how things look. It's a catastrophically wrong conclusion if you believe design is how things work.
Right now, companies are implicitly betting that they can hold design headcount flat while PM and engineering grow, that they don't need more design leverage. And maybe they can get away with it, for now, because when the floor is rising for everyone, nobody notices the ceiling isn't moving.
But Jeff Bezos has a line I keep coming back to, "your margin is my opportunity," and what he means is that the slack you're comfortable with is exactly where someone hungrier is going to build an advantage. Today's product orgs are running lean on design and telling themselves it's fine. That complacency is the margin.
Whether the next hiring report shows a different story depends entirely on whether designers behave like production artists or like systems-level product shapers that AI amplifies. And for ambitious designers, the gap between what companies think they need and what they actually need is where the opportunity lives, the chance to make it painfully obvious that under-investing in design is now the expensive choice, that the teams who treat design as leverage instead of decoration will outperform everyone who treated it as a line item to optimize.
The misconception that's costing companies real money
Here's where most organizations go wrong, and it's a mistake I see repeated in almost every conversation about AI and design, they conflate production with imagination. They see AI generating polished screens and reasonable layouts and conclude that the design problem has been solved, that what remains is just execution, and that execution is now cheap. But design was never really about production. Not the kind that matters, anyway.
Design is the decision about what to build, not just how to render it. It's the choice architecture that shapes behavior, the information hierarchy that surfaces the right thing at the right moment, the interaction model that takes a complex workflow and makes it feel like the obvious thing. It's the strategic judgment about what to leave out, which is almost always harder than deciding what to include.
AI is extraordinary at production. It can generate screens, layouts, components, and variations faster than any human team could hope to match. But it's optimizing within known patterns, recombining what already exists, filling in the space between solutions it's seen before. It doesn't look at a product and recognize that the entire paradigm is wrong, that the team should be thinking about it differently. That requires something AI doesn't have and isn't close to having, the ability to imagine what doesn't exist yet and articulate why it should.
The real opportunity everyone is missing
Every major technology shift has created interaction paradigms that nobody predicted from the technology itself. The mouse didn't just make existing terminal commands easier to access, it enabled direct manipulation, which fundamentally changed how people conceptualize digital objects. The smartphone didn't shrink the desktop, it created an entirely new computing context, location-aware, always-on, gesture-driven, that had almost nothing in common with what came before. Touch screens didn't replace buttons, they made interfaces feel physical in a way that opened up entirely new categories of application.
AI will do the same thing, and we're already seeing the early signals. Conversational interfaces are replacing form-based workflows in ways that aren't just chatbots dropped into sidebars but genuine rethinking of how humans express intent to systems. Agent-to-agent communication is creating orchestration and visibility problems that look nothing like traditional navigation, where the design challenge shifts from whether someone can complete the task to whether they can understand what just happened. Predictive personalization is making static, one-size-fits-all layouts feel increasingly like printed brochures in a world that expects the page to know who you are.
But here's the thing, we're still in the horseless carriage phase. We're bolting AI capabilities onto existing interaction patterns, adding summarization buttons to dashboards and autocomplete to text fields and chatbots to help centers, and while all of that is useful, none of it is the paradigm shift. The paradigm shift will come from someone who looks at what AI makes possible and imagines an entirely different way for humans and systems to work together, not a better version of the current thing, but a different thing entirely.
That takes imagination. Specifically, it takes design imagination, the ability to synthesize technology capability, human behavior, and business context into something that didn't exist before. And that is exactly the capability that AI cannot provide, because you can't remix your way to a paradigm that doesn't exist in the training data.
This isn't a production gap, it's a leadership gap
AI can produce and iterate and optimize, but it can't set the bar. It can't look at a product and articulate why it's not good enough, not because of the pixels but because of the underlying model of how people should interact with it. It can't run a listening tour across an organization and diagnose that the real problem isn't the interface but the decision-making structure behind it. It can't sit in a room with a product lead and an engineering lead and navigate the tension between shipping this quarter and building the right foundation for the next three years.
These are design leadership problems, and they require taste, judgment, strategic thinking, and the ability to hold a vision for what the product should become while managing the messy reality of what it is today. They require someone who can look at a sea of competent, well-produced, perfectly adequate AI-generated interfaces and say it isn't good enough, not because it's broken, but because it isn't imagining hard enough.
The companies that will build exceptional products in this era won't be the ones generating the most screens or shipping the most AI-assisted prototypes. They'll be the ones with leaders who can imagine what those screens should be in the first place, who can see around the corner to interaction paradigms that don't exist yet, and who can set a bar that AI alone will never reach.
The floor has risen, and that's worth celebrating. AI will keep raising it, keep compressing the distance between bad and good enough, keep making production quality cheaper and faster and more accessible to everyone. The organizations that treat design as a production function will get exactly what AI gives them, work that is competent, interchangeable, and forgettable. The ones that treat it as a leadership function, as the discipline responsible for thinking in systems and discovering what comes next, will be the ones building things the rest of the market spends the next decade copying. Because the designers who win this era won't be the ones making the prettiest screens, they'll be the ones who can hold the whole product in their head at once, who can see how decisions in one corner of the experience ripple through everything else, and who can shape the connective tissue between agents, interfaces, data, and human judgment that AI alone can't reason about.
But if you think that design is how it looks instead of how it works, then this is where you are truly missing the real opportunity.