Design teams will become smaller. That part is fairly evident. When one designer with AI can do work that used to require a designer plus dedicated engineering support, headcounts compress. The math is simple and one that I suspect we'll start to see play out across the industry in the next year or so.
The more interesting question is what happens to the people trying to enter the field.
The traditional path into product design was built on production work. You started by doing the work senior designers didn't want to do. Redlining specs. Building out component variants. Preparing assets. Cleaning up design files. Anyone who's been through it knows the feeling. The work wasn't glamorous, but it served a purpose beyond getting the job done. It's how you developed an eye. You learned the system by working inside it. You built taste by doing a thousand small things and slowly understanding which ones were better than others.
AI does most of that work now, and does it well. A junior designer in 2023 might have spent days building out all the responsive variants of a card component. Today AI does that in seconds, and the output is often more consistent than what a junior would have produced. The redlines, the specs, the asset preparation, the tedious-but-educational busywork that formed the first two years of a design career is largely automated.
On the surface, this looks like it eliminates the on-ramp. The entry-level work that trained an entire generation of designers is disappearing. Where do the reps come from if the reps are automated?
I think the answer is that they come from building.
Here's the thing about the current disruption: it doesn't just affect juniors. It's disrupting the entire industry. Most senior designers built their careers in a specific paradigm. Figma mastery, handoff fluency, managing design systems inside design tools, navigating the spec-to-engineering pipeline. If the medium shifts to code and the primary collaborator becomes AI, a lot of that accumulated advantage becomes less relevant.
A junior entering the field right now has no muscle memory to unlearn. They're not fighting habits built over a decade of handoff-driven workflows. They can learn the new paradigm natively, without having to first dismantle the old one. In that sense, this technology almost resets the playing field. The gap between junior and senior narrows, not because juniors got better overnight, but because a significant portion of what made seniors "senior" was mastery of a system that's being replaced.
Not all of it, though. Seniority is partly tools and process, but it's also judgement about users, the ability to make decisions under ambiguity, understanding of business context, knowing which problem to solve before solving it. Years of experience builds design and product intuition. Those things don't reset. A junior who can build fast with AI but doesn't understand why a particular flow confuses users isn't more valuable than a senior who can identify the problem instantly.
So the playing field partially resets. The tool-and-process layer of seniority gets compressed. The taste-and-judgement layer remains. The juniors with the biggest opportunity aren't the ones who learn AI tools fastest. They're the ones who combine new-paradigm fluency with a genuine investment in developing the judgement that used to take years.
The good news is that I think judgement can develop faster now than it ever could before.
The old apprenticeship model was slow. You'd design a card component in Figma, submit it for review, wait for feedback, revise it, wait again. The cycle from decision to feedback might be days. You learned, but you learned at the pace of the review process.
Building with AI changes the feedback loop completely. When you build a component, you see it immediately with real data, at real screen sizes, in real interaction states. You find out in seconds that your card layout breaks when the title wraps. You discover that your animation stutters on mobile because you triggered a layout recalculation. You don't need a senior designer to tell you something is wrong. The browser tells you, and it tells you right now.
The reps are faster, which means taste develops faster. A junior who builds ten components a day with AI and evaluates each one critically is getting more design feedback in a week than a junior in 2023 got in a month of Figma-based production work. The medium is more honest and more immediate than a design tool ever was.
This doesn't mean the judgement develops automatically. You still need to look at what you've built with a critical eye. You still need to understand the fundamentals: how spacing creates rhythm, how type hierarchy guides attention, how colour carries meaning. AI won't teach you those things. But it will put you in a position to learn them faster, because you're working in the real medium from day one instead of working in an abstraction of it.
The catch is that this requires a different skill set at entry level. A junior designer in 2023 needed Figma proficiency, basic visual design principles, and the ability to follow a design system. A junior starting today needs all of that plus enough technical literacy to work with AI in code. They need to understand components, tokens, basic CSS, and how the browser renders things. Not deeply, but enough to direct AI and evaluate what it produces.
So the junior designer problem is real, but it's not the problem most people think it is. The problem isn't that there's no way in. It's that the way in has changed.
Fewer seats at the table, yes. But the people who get a seat can develop faster than any previous generation of designers. The feedback loops are tighter, the barriers to building are lower, and the playing field has been reset enough that being new is less of a disadvantage than it used to be.
If I were starting a design career today, I'd learn to build from day one. Not after mastering Figma, not as a side skill, but as the core of my practice. I'd invest in fundamentals over tools, because the box model and colour theory will outlast whatever software is popular next year. And I'd find people who are already working this way and learn from what they ship.
The path into design has changed. I think the new one is actually better, but only if you know it's there.