The tools are getting smarter. The work isn't getting easier.

May 6, 2026- by Jonas Verheijden

On April 17, Anthropic launched Claude Design. Figma's stock dropped 7% within hours. The internet declared another designer-killer had arrived. We've seen this before.

A few weeks earlier, Google had quietly launched Stitch, their own AI-powered design tool. Before that, it was Claude Code rewriting assumptions about what developers actually do all day. Each launch arrives with the same energy: this changes everything, your role is obsolete, adapt or disappear.

 We do adapt. That's not the question.

 The question is whether these tools actually change what good design requires. And I don't think they do.

What Claude Design actually is

It's worth being precise about what launched. At its core, it's a workspace where you describe what you need and Claude builds a first version. Prototypes, wireframes, slide decks, one-pagers. You refine, export, hand off. 

Anthropic is explicit that it's aimed at people who need to get from an idea to something visual before they open a design tool. Founders, product managers, marketers without a designer available. The Canva handoff, not a Figma handoff, tells you something about the intended audience.

 The 7% stock drop is a market signal, not a product review.

The exploration problem

Here's where I actually disagree with the "it's just a complement" framing, at least partially.
Claude Design does something real for exploration speed. Anthropic's own launch post makes the point that experienced designers have to ration their exploration because there's rarely time to prototype a dozen directions. That's true. And if a tool changes that calculus, it matters. 

But there's a difference between exploration volume and exploration quality. 

In the early phase of a project, the most valuable thing an experienced designer does is not generate options. It's read a brand's visual potential and push against its defaults. It's understanding what a company could look like digitally, not just what it currently is. That judgment is built from years of pattern recognition across work that didn't exist until someone made it.
 
AI tools, including Claude Design, are extraordinarily good at recombining what already exists. They optimise toward the recognisable. That's useful. It's also, by definition, not challenging the status quo.

Why style tiles still matter

My preferred method for helping clients find a visual direction is style tiles: presenting vastly different variations on typographic patterns, motion principles, and layout tension. Not polished mockups. Not final answers. Provocations. 

The goal is to surface what a client responds to before they know how to articulate it. To show them the distance between where they are and where they could go. That requires genuine creative range, and the discipline to hold back the "safe" option until the conversation has earned it.  

I have not been able to recreate that output with any AI tool. 

Not because the tools can't generate style tiles. They can produce something that looks like one. But the underlying logic is missing. The choice of what to make tense, what to leave quiet, what typographic contrast says about a brand's confidence, what motion communicates about trust: these are decisions that require a point of view. One that's developed over time and can be argued for in a client meeting.
 
That's not a gap that closes with a better model or a higher resolution.

AI tools are extraordinarily good at recombining what already exists. They optimise toward the recognisable. That's useful. It's also, by definition, not challenging the status quo.

What actually changes

Claude Code has changed how we think about the design-to-development handoff. In meaningful ways. Prototypes that were previously low-fidelity conversations are becoming higher-fidelity starting points for real builds. That's a workflow shift worth taking seriously.

 Claude Design will change where non-designers start. It lowers the barrier to producing something visual, which reduces the back-and-forth in early project phases and gets everyone to a more concrete conversation faster. For our work with clients, that's a net positive.

 But neither of these tools changes what we're actually hired to do: take a brand into a digital context in a way that's coherent, distinctive, and built to last. That work requires taste, experience, and the willingness to push a client toward something they didn't know they wanted.
 
The tools are faster. The judgment still has to come from somewhere.

Staying current is not optional

At Born Digital, we make a point of being on top of releases, tools, and shifts in the landscape. That's not optional when your clients are making decisions based on what they read on a Friday afternoon. 

But staying current is different from being swept along. We use AI tools where they make the work better or faster. We don't use them where they flatten the work into something generic.
 
The distinction matters more now, not less.