AI didn't just change how we build. It changed who gets to build.
For the past two decades, the product design process looked like this:
Research users. Define their problems. Design a solution. Ship it to everyone.
That last part? That's the part nobody questioned.
We spent months interviewing users, synthesizing insights, mapping journeys. All to arrive at one solution that was supposed to work for millions of people with slightly different versions of the same problem.
It was the best we could do. Building software was expensive. Design was expensive. Development was expensive. So we had to generalize. We had to find the common denominator and ship that.
That constraint is gone now.
The old model had a dirty secret
Here's what nobody in product teams liked to admit. That "one solution" we shipped? It was a compromise.
User A needed the feature to work one way. User B needed it to work another. User C had a completely different workflow. So we averaged it out. We picked the version that annoyed the fewest people. We called it "solving the user problem."
It wasn't solving. It was approximating.
The entire UX process was built around this approximation. Personas were approximations of real people. User journeys were approximations of real behavior. Design systems were approximations of what every screen needed to look like.
We got really good at approximating. We built entire careers around it. We wrote frameworks and books about how to approximate better.
But approximation was never the goal. It was the limitation.
What AI actually changed
AI didn't just give designers better tools. It didn't just speed up wireframing or generate UI components faster.
It broke the economic constraint that forced us to build one solution for everyone.
Think about what it costs to build a custom solution today versus three years ago. A person with no coding background can describe their specific problem to an AI and get a working tool built in an afternoon. Not a prototype. Not a mockup. A working tool.
I've tested this repeatedly. Claude Code, Gemini, Cursor. These aren't toys anymore. A designer who understands the problem can now also build the solution. A marketing manager who needs a specific dashboard can describe it and have it running by lunch.
The cost of building tailored solutions just collapsed. And when costs collapse, the old model breaks.
From "define THE problem" to "define YOUR problem"
This is the shift that most product people haven't internalized yet.
The traditional process: We define the problem FOR users. We decide what the solution looks like. We ship it. Users adapt to our design.
The emerging reality: Users define their OWN problems. They describe exactly what they need. AI helps them build exactly that. No adaptation required.
This doesn't mean professional designers and developers disappear. It means the nature of what they do changes fundamentally.
Instead of designing the solution, you might be designing the system that helps users create their own solutions. Instead of building the product, you might be building the framework that lets anyone build their specific version of the product.
The value moves upstream. Way upstream.
I've seen this happen in real time
Let me give you a concrete example.
A few weeks ago, I needed a specific tool to batch-process some content files. Very specific requirements. The kind of thing where no existing SaaS product does exactly what I needed.
Old world: I'd spend hours searching for the closest tool, pay for a subscription, then work around its limitations. Or I'd file a feature request and wait six months.
New world: I described my exact workflow to an AI coding tool. Had a working script in 20 minutes. It did precisely what I needed. Nothing more, nothing less.
No compromise. No approximation. No "close enough."
Multiply this by millions of people doing the same thing every day. That's not a trend. That's a structural shift in how solutions get created.
The "Duck Era" implications
I've been talking about the "duck era" for a while now. The idea that generalists who know how to use AI tools can compete with specialists in their domains.
This article is about the product-level version of that same idea.
If a generalist can build their own tailored solution, why would they settle for a generalized product that only partially solves their problem?
They won't. Not for long.
The products that survive this shift will be the ones that embrace it. The ones that say "here's a flexible system, shape it to your needs" instead of "here's our solution, take it or leave it."
Notion understood this early. So did Airtable. But AI takes it further. You don't even need a platform anymore. You can build from scratch, for yourself, in hours.
What this means for designers and product people
If you're a UX designer reading this, I'm not telling you to panic. I'm telling you to reposition.
Stop optimizing for the average user. That user doesn't exist. They never did. We just pretended they did because we couldn't afford to build for individuals.
Start thinking in systems, not screens. The value isn't in designing the perfect dashboard layout. It's in understanding the underlying problem space deeply enough to create frameworks that flex to individual needs.
Learn to build. Not because "designers should code." Because the gap between understanding a problem and shipping a solution has never been smaller. And if you can close that gap yourself, you're exponentially more valuable.
Get comfortable with users who build. Your future users might not use your product the way you designed it. They might use AI to modify it, extend it, or replace parts of it entirely. Design for that reality.
The uncomfortable question
Here's what keeps traditional product teams up at night, even if they won't say it out loud.
If anyone can build a solution tailored to their exact problem, what's the value of spending six months on user research to design one solution for everyone?
The answer isn't "zero." But it's a lot less than it used to be.
The value now is in the problems that are too complex for individuals to solve alone. The ones that require infrastructure, data, network effects, trust, or scale that a single person with an AI tool can't replicate.
Everything else? Fair game for the person who just describes what they need and lets AI build it.
The new floor
I've said before that AI hasn't just raised the ceiling. It created a new floor.
The same applies here. The floor for "acceptable solution" used to be a mass-market product that kind of worked for your use case. Now the floor is a custom-built tool that works exactly for your use case.
When the floor rises, everything above it has to rise too. Or it becomes irrelevant.
Products that offer generic solutions to problems that individuals can now solve themselves with AI? They're standing on a floor that's rising to meet them.
The ones that will thrive are the ones building above where any individual with AI can reach. Complex systems. Deep integrations. Things that require expertise, infrastructure, and scale.
Everything else is getting democratized. Fast.
What to do about it
If you build products: Start thinking about how your users might solve their own problems with AI. Then ask yourself what value you provide beyond that.
If you design experiences: Shift from designing fixed solutions to designing adaptable systems. Your expertise in understanding human behavior is more valuable than ever. But the output of that expertise changes.
If you're an individual with a problem: Stop settling for "close enough" tools. Describe your exact problem to an AI. You might be surprised how quickly you get a perfect solution.
The era of one-size-fits-all is ending. Not because it was wrong. But because the constraint that made it necessary no longer exists.
Build for one. Build for yourself. Build exactly what you need.
The tools are here. The only question is whether you'll use them.