The next five years of product design

We’re in a weird spot. AI tools can now generate full interfaces from a text prompt. Figma just partnered with Anthropic to let teams bring AI-generated code back into the design canvas. The SaaSpocalypse wiped a trillion dollars from software stocks in a month because investors realized AI agents might not need software seats at all.

Meanwhile, 39,000 tech workers have been laid off in 2026 and we’re not even through February. Amazon cut entire design teams. Companies are restructuring around AI and quietly not replacing the roles that used to exist.

And yet, somewhere in all of this, products still need to make sense to the people using them. Someone still has to figure out what to build, who it’s for, and whether it actually works. That job hasn’t gone away. But what that job looks like is changing fast.

Here are five questions I keep thinking about, with two possible outcomes for each. Not predictions. Just two versions of what could happen, and what you can do either way.

Will designers still make the thing, or will they decide what the thing should be?

Right now, a big chunk of design work is production. Laying out screens, building components, wiring prototypes, documenting specs. This is the work that AI is already good at and getting better at fast. Figma Make generates multi-screen flows from a prompt. Tools like v0 and Lovable turn descriptions into working front-ends. Claude writes copy that’s better than most placeholder text.

Scenario A: Production work shrinks dramatically. The screens, the components, the specs, all of it gets generated or semi-automated. Design teams get smaller. The designers who remain spend most of their time on research, strategy, and decision-making. Less time in Figma, more time in documents and conversations.

Scenario B: Production work doesn’t disappear, it just speeds up. AI handles first drafts and repetitive layout work, but designers still refine, polish, and make judgment calls on every screen. Teams stay roughly the same size but ship two or three times more. The bar for quality goes up because everyone can produce more.

What to do about it: Either way, the designer who only knows how to push pixels is in trouble. If you spend 80% of your time on production and 20% on thinking, start flipping that ratio now. Get comfortable with research, with writing, with presenting your reasoning. Practice articulating why a design works, not just showing that it looks right. The thinking was always the real job. AI is just making that obvious.

Will entry-level design jobs survive, or will AI replace the tasks that juniors used to learn on?

Junior designers historically learned by doing the work nobody else wanted to do. Cleaning up Figma files. Building out component variants. Documenting spacing and color tokens. Filling in the screens between the ones the senior designer already figured out.

AI is now capable of doing most of that. And companies know it.

Scenario A: Junior hiring slows down. Companies figure they can give a senior designer AI tools and get the same output that used to require a senior plus two juniors. Entry-level roles get scarce. The people who do get hired are expected to be AI-proficient from day one and operate more like mid-level designers immediately.

Scenario B: Junior roles evolve instead of disappearing. As AI handles the production grunt work, juniors are freed up to do more interesting things earlier in their careers. More research exposure, more stakeholder interaction, more strategic work. The learning curve is steeper but the ceiling is higher. Companies that invest in junior talent early end up with stronger senior designers five years from now.

What to do about it: If you’re early in your career, don’t just learn the tools. Learn the thinking. Be the junior who can run a usability test, synthesize interview notes, and present findings to a PM, not just the one who builds clean components. Use AI to skip the blank canvas, but make sure you understand why the AI’s suggestion works or doesn’t. Build things outside of work. Show judgment, not just output. If you’re senior, think about what happens when the pipeline of junior designers dries up. Today’s juniors are tomorrow’s senior designers and design leaders. Cutting them off creates a leadership gap that’s hard to fill later.

Will “designer” stay a focused role, or will it absorb responsibilities from five other jobs?

This is already happening. Companies are expanding what they expect one designer to handle. Research, visual design, prototyping, copywriting, light front-end, data analysis, presenting to stakeholders. Roles that used to be split across a team are consolidating because AI makes one person faster at all of them.

Scenario A: The designer role expands into something closer to a product generalist. You’re expected to write your own copy, analyze your own data, prototype in code, and present to leadership. Specialists still exist in large organizations, but most companies want one person who can do all of it. The title stays “product designer” but the job description doubles.

Scenario B: The opposite happens. As AI handles surface-level work across all those domains, companies realize they need depth over breadth. Specialists in research, interaction design, content design, and design systems become more valuable because AI-generated work is decent but not deep. Generalists hit a ceiling where they’re okay at everything and great at nothing.

What to do about it: I think reality will land somewhere in the middle, which means the safest position is the T-shape. Go deep on one or two things (research and interaction design, or systems thinking and visual craft) and be competent across the rest. AI makes it easier to be competent at adjacent skills. It doesn’t make you an expert. Use AI to cover your weak spots, but invest real time in your strengths. That depth is what will be hard to replace.

Will designers be measured by what they make, or by the decisions they influence?

For years, design portfolios have been full of beautiful screens. Case studies follow the same arc: problem, research, wireframes, final UI, results. The implicit message is “look what I made.”

But if AI can generate a decent-looking interface in minutes, that message loses weight.

I’ve been thinking about this with my own work. When I look at the projects that had the most impact, the screens were almost never the reason. It was the research that uncovered the real problem. The decision to cut a feature instead of adding one. The conversation where I convinced a stakeholder to change direction. The screens were just the artifact of those decisions.

At the same time, I don’t buy the idea that craft is dead. AI-generated interfaces all start to look the same because they’re trained on the same patterns. Products that feel generic will lose to products that feel intentional. Someone has to bring taste, emotional detail, the “something’s off and I can’t explain why yet” instinct. That’s still human.

So I think the answer is both. Invest in your ability to think clearly and communicate your reasoning. But don’t abandon craft. The combination of strong thinking and strong execution is rare and will stay rare. When you present your work, lead with the problem and the tradeoffs, then show the design. When you build your portfolio, explain what you considered and rejected, not just what you shipped. That’s the part AI can’t generate.

Will our tools stay design-first, or will the canvas become code?

Figma has been the center of the design workflow for years. But something is shifting, and I feel it in my own work.

Figma’s partnership with Anthropic lets teams bring AI-generated code into the design canvas. Tools like Cursor and Claude Code let designers (and anyone else) build working interfaces by describing what they want. I’ve been using v0 to build front-end pieces of Brisk, and the speed is wild. Rough, but wild.

The workflow used to be clear: design in Figma, hand off to developers. Increasingly I find myself doing the opposite. Prompt an AI, get working code, then open Figma to refine the details and make decisions about the interaction. The starting point moved and I didn’t really notice until I looked at how I spend my days now compared to a year ago.

I don’t think Figma is going anywhere. But I do think designers who can move between visual tools and code (even AI-assisted code) will have a real edge over those who can only work on one side. You don’t need to become a developer. But understanding enough HTML, CSS, and basic React to evaluate whether AI output is good or garbage makes a difference. It means you can prompt an AI to build a working prototype and actually know if what came back is right.

The real advantage in 2027 won’t be mastering one tool. It’ll be knowing which approach is faster for the problem in front of you and switching without friction.

None of these scenarios are mutually exclusive. Some companies will shrink design teams while others expand them into new domains. AI will automate production work while raising the bar for everything humans touch. Designers might spend mornings reviewing AI-generated layouts and afternoons in user interviews.

The consistent thread across all five questions: the value of a designer is shifting from what you produce to what you decide. The artifacts (screens, prototypes, specs) are becoming easier to generate. The judgment behind them is not.

If I had to bet on one piece of advice for the next five years, it would be this: get really good at the parts of design that happen before and after the screen. The research that tells you what to build. The strategy that tells you why. The communication that gets your team aligned. The evaluation that tells you whether it worked.

AI is making the middle part, the actual production of interfaces, faster every month. Everything around it is still deeply, stubbornly human.

That’s where you want to be.

Thanks for reading :) If something in this issue resonated, please reply and tell me.

— Balint

What else I’m working on?