Designers have had the strangest two years of any role in this series. The tools that matter to their craft went from cool novelty to core workflow in a shorter time than almost any technology shift I have watched. New capabilities arrive every month from Midjourney, DALL-E, Stable Diffusion, Firefly, Figma AI, Canva, Runway, and v0. Each of these quietly kills a category of work that used to take hours and now takes seconds.
If you’re a designer, the surface experience of the last two years has probably been: “Wait, AI can do that now? And now that too?“
Underneath the surface, though, something more interesting has happened. AI has turned out to be weirdly good at making individual images and weirdly bad at making coherent brands. It can produce a beautiful hero illustration in thirty seconds and still, somehow, produce one that doesn’t look like yours. It can generate a thousand variants and somehow none of them feel like a system. The gap between “an image” and “a brand” has turned out to be exactly where the human designer’s job now lives.
This piece is about what that means if you design for a living.
The old designer’s job
For most of my career, the designer on a small marketing team had an exhausting but well-defined role. You made everything.
Designers handled landing page layouts and ad creative, often producing banners, stories, and carousels in dozens of variants. They created email templates, social graphics, and decks for sales or investor meetings. The work extended to icon sets, simple illustrations, and event materials. When a company was young, they even handled brand identity and full web design work. Sometimes, they produced product marketing visuals that sat at the line between marketing and product design.
The work was a constant tension between two things: craft, meaning making things look and feel right, and throughput, meaning producing enough output to keep the rest of the team unblocked. Most designers I’ve worked with leaned toward craft and felt perpetually behind on throughput. The role was chronically understaffed because throughput demand was infinite.
Underneath all of it, the actual skills that mattered were fluency in tools like Figma and the Adobe suite, along with typographic sense, layout, color, and visual hierarchy. You also needed enough brand thinking to keep everything from looking like it was made by different people on different days. Senior designers added art direction and systems thinking. If the company was lucky, they also provided taste.
The job worked the way it did because production was expensive. Every image was a decision multiplied by an hour in Figma.
What AI actually changed
The economics of image production have collapsed. This did not happen gradually; it happened suddenly.
The tools that matter here aren’t exotic. Midjourney and its peers for imagery. Figma AI, Canva AI, Adobe Firefly for in-tool assistance. Runway and Pika for video. v0 and similar for code-driven interfaces. Each tool, individually, automates some piece of production. Together, they’ve pulled the floor out from under “producing an image” as a chargeable unit of work.
What used to take an afternoon now takes fifteen minutes. What used to take a week now takes an afternoon. Variant production, which represents the tedious middle of most design work, is now trivial.
But here’s the part that’s taken longer for the industry to notice: none of this has made AI good at brand.
AI produces beautiful outputs that do not belong to your brand. It produces thirty variants that technically meet the brief but collectively look like thirty different companies. It generates illustrations in whatever style the latest model is biased toward, rather than yours. It defaults to generic visual conventions, which have become the stock photography of the AI era. Left alone, it pulls every brand toward the same beige, competent, and slightly uncanny average.
This gap is the designer’s job now. AI has made production almost free. It has made coherence rare. The work has moved from “make things” to “make sure everything that’s made fits together and looks like us.”
What’s dying
The parts of the designer role that are visibly fading:
Pure production work. “I make the images” is no longer a full job at most companies. If your value to a team is throughput, you are essentially trying to make things faster than they can. AI just made that value less scarce. Production speed is not the moat it used to be.
Stock photo selection as a discipline. A lot of design work used to be sourcing and combining stock photography tastefully. That work is going away. AI-generated imagery, if well-directed, is usually better and always more flexible.
Variant work as a skilled task. Producing 30 ad variants used to be a real chunk of a designer’s week. Now it’s a prompt and an hour of curation. Designers who built their identity around being fast and disciplined at variant production need to move their value somewhere else.
Designers who can’t direct AI tools. I know this is controversial in design circles — there’s real discomfort with AI in the profession, some of it ethical, some of it aesthetic. I respect that discomfort. But practically, a designer in 2026 who doesn’t have a working AI toolkit is losing ground every month. The people being hired right now, at every level, are the ones who’ve adopted these tools aggressively.
“Pretty things” without systems thinking. A designer who can make a single beautiful artifact but cannot build a system is in a role that is shrinking. Without a system, the team cannot produce 100 beautiful artifacts without that designer. The systems question, which asks how everything can cohere without a designer personally touching it, has become central to the job.
What’s becoming more valuable
The parts of the role that have grown in value — often dramatically:
Taste. I’m going to use that word a lot in this section because there’s no good substitute. When AI can produce a thousand outputs, the person who can look at them and say “this one, not these 999” is the person running the design operation. Taste is the ability to know, on sight, what’s working and what isn’t. It’s built through thousands of hours of studying good and bad work, and it’s almost impossible to shortcut. It’s also the single hardest thing to automate.
Brand systems thinking. The job is not about making one artifact. Instead, it is about making a system that produces consistent artifacts through design tokens, component libraries, typographic scales, and color systems. When AI handles much of the production, the system is what keeps everything from drifting. Designers who can build and maintain these systems are building the substrate that the rest of the team and the AI stack operate on.
Art direction. Describing what you want clearly enough that another human or an AI can produce it is no longer just for creative directors at big agencies. It is now a core skill for any designer directing AI workflows. If you can describe a visual concept precisely through references, mood, composition, and constraints, you can produce a thousand assets in the time it used to take to produce one. Without that precision, you get the AI default output, which is everyone’s default output.
Typography and layout craft. AI is still weirdly bad at typography. It struggles with nuanced type pairing, sophisticated layout, and the micro-decisions that separate competent design from genuinely great design. This is an area where human skill still compounds. Designers who double down on these skills will maintain a durable advantage.
Custom visual language development. Every brand using default Midjourney outputs looks like every other brand using default Midjourney outputs. Designers must develop custom styles through trained references, signature treatments, and proprietary visual vocabulary. These elements give a brand something AI alone cannot provide. This is the defensible brand skill.
Cross-platform and cross-format fluency. The modern designer has to think about how a brand shows up on Instagram Stories, LinkedIn, a landing page, a Google ad, a podcast cover, an email, and an in-product moment. The craft has broadened. Generalists who design across the whole brand surface are outperforming specialists who only design for one format.
Design-and-code fluency. Tools like v0 and Figma’s dev handoff have blurred the line between design and frontend implementation. Designers who can take concepts to working code (or very close to it) are doing work that used to require two or three handoffs. This isn’t a requirement yet, but it’s quickly becoming an advantage.
How to become this
If you’re a designer, or someone who does design as part of a bigger marketing role, here’s what I’d concretely focus on.
1. Adopt AI tools aggressively and build your own workflow.
Midjourney for imagery, Figma AI for in-flow assistance, Firefly or its peers for variations, v0 or similar for quick prototypes. Build your personal stack. Learn which tool is best for which job. Develop prompt templates for recurring tasks. This is table stakes now, not an advanced skill.
2. Invest in taste, deliberately.
Study design you admire and design you dislike. Articulate why things work. Keep a collection. Follow designers with strong points of view. Read design history. Write about why a specific design choice succeeds or fails. Taste sounds ineffable; it’s actually trainable if you pay attention.
3. Build brand systems, not just artifacts.
Every project is an opportunity to extract a system. How do we handle type? What’s the illustration style? What’s the photography treatment? What are the components? Document these. A designer with a portfolio of systems they’ve built is a much stronger candidate than one with a portfolio of one-off pieces.
4. Develop art direction muscle.
Practice describing what you want. Take a design you admire and write a 200-word art direction brief for it. Learn to prompt AI precisely and judge the results critically. Art direction is the skill that scales your taste across any tool or team.
5. Go deeper on typography.
In an AI-flooded visual world, typographic craft is one of the last reliable markers of serious design work. Learn it seriously. Read the classics. Practice pairings. Develop opinions. Typography is where most AI-generated design reveals itself as AI-generated; it’s where human designers can still show clear superiority.
6. Build a custom visual style.
Rather than relying on default AI aesthetics, develop your own references, your own trained looks, your own treatments. Brands that look unique now are brands with a human designer who’s spent time building something specific. That designer is increasingly hard to replace.
7. Get comfortable working alongside the rest of the team.
The designer who sits in a silo and hands off finished files is fading. Instead, the modern designer collaborates with content, performance, and product teams. By showing up early in decisions and staying involved late into shipping, they remain central to how modern growth teams operate.
Signals you’re ready
The designers I’d hire now pass most of these:
- Can you show me an AI-integrated design workflow you actually use?
- Can you look at an AI-generated image and tell me what’s wrong with it, specifically?
- Does your portfolio show brand systems, not just individual artifacts?
- Do you have a point of view on typography?
- Have you developed a custom visual style or direction that doesn’t look like default AI?
- Can you take a project from concept to something close to shipped without being a bottleneck?
- Do you work well alongside non-designers, taking input without being defensive?
If most of those are yes, you are operating at the level the role needs. If most of them are no, there is a real opportunity in front of you. This exists because the floor has risen, but the ceiling has risen further.
The thing about taste
Every designer I have talked to about this shift eventually lands on the same word: taste. The production work is gone. The directive work is now the whole job. This involves knowing what good looks like and knowing how to get it.
Taste is interesting because it’s the one thing in the design profession that’s always been undervalued by the market. Companies paid for production because production was billable; they got taste for free from the same designers as a bonus. AI has flipped that. Production is now cheap; taste is scarce. The designers who’ve spent years quietly developing taste without getting particularly paid for it are about to find out that the market has finally caught up to them.
If that’s you, this is a good moment.
