Who’s shaping design?
Charles and Ray Eames Kazam Machine, for bending plywood into formed chairs
"We shape our tools, and thereafter our tools shape us."
— John M. Culkin
For most of human history, makers built the tools they needed to make things. Charles and Ray Eames needed to bend plywood into compound curves no existing equipment could achieve — so they built the Ka-Zam machine in their apartment, a steam-and-pressure mold that let them shape the shells for what would become some of the most iconic furniture of the twentieth century. The tool didn't exist, so they invented it. The invention made the work possible.
That tradition runs deep in design. Designers have always understood that the tool is never neutral: it determines what can be made, how it can be made, and ultimately what gets made at all. Which makes the last two generations of designers an anomaly. For the first time in the history of the profession, designers did not shape their own tools. Their tools shaped them.
The sandbox and what it cost us
Adobe, Figma, Macromedia, Sketch, and more recently Lovable and Framer have done more than build software. Through close partnerships with universities and education programs, they have defined how design is taught, how it is practiced, and what counts as design in the first place. The boundaries of their software became the boundaries of the discipline. Designers became fluent in tools they had no hand in creating, optimized for workflows they had no say in defining, working inside a sandbox whose walls they rarely questioned.
The promise was efficiency: freed from the labor of making, designers could focus on thinking. Strategy. The coveted seat at the table. What the promise didn't mention was what would be lost in the trade.
Grid systems are a useful illustration. A grid was never a stylistic preference; it was a method based on Gestalt psychology for bringing order, hierarchy, and coherence to visual communication. Built around typographic scales, line lengths, and character proportions, a well-constructed grid was a reduction to essentials: a limited set of decisions made in advance so that every subsequent decision served the whole. It treated design as a system, not a series of screens. Today's grids have devolved into meshes of dozens or hundreds of columns, with no typographic foundation and no coherent system beneath them. The software will handle it.
The same erosion happened at the level of letterform. When I was in school, expressive typography began with drawing: constructing letterforms by hand in different weights to understand their proportions, the relationship between stroke and counter, the geometry that gives a typeface its character. One student in that class produced something I still think about. The assignment was expressive typography. Their solution was a single word: ERECTION — with the leg of the R rotated to point straight outward, horizontal, mimicking exactly what the word describes. One letter. One small adjustment. The joke is instant and the execution is perfect. But it only works because the student understood the letterform well enough to know precisely which stroke to move, and how far. You cannot make that decision from a dropdown menu.
I have yet to see a designer whose only experience with typography is on a screen demonstrate that level of understanding. Not because they lack talent, but because the software never required them to develop it. Software has all but eliminated the need for designers learn real craft—the hand skills to sketch, draw and shape their designs.
Designers have become the client,
and the software is now the designer.
The undo button removed something else: the cost of a decision. When resources don't need to be conserved and mistakes can be reversed in an instant, the discipline of thinking before making disappears. The ability to collapse entire timelines by generating final artifacts while still ideating removes the time that once existed for reflection, iteration, and refinement. Risk and reward no longer need to be weighed in advance.
Most consequentially, the tools eliminated the sketch. A designer's sketches were not just drawings; they were the visible record of thinking in progress. Like a mathematician's chalkboard, they walked the problem through its iterations until reaching a solution. They sharpened the designer's eye, forced attention to detail and nuance, created a tangible artifact that could persist as a point of reference, and compounded the designer's skills over time. A career's worth of sketchbooks made a designer better in ways that a career's worth of screen time simply does not.
In their place: proficiency with software. Which is, at best, one level of abstraction removed from the work itself. Fingers no longer hold tools; they input on keyboards. Increasingly, designers simply describe what they want and the software produces it. The designer has become the client. The software is now the designer.
And because that software is built on large language models trained on existing images, it has no understanding of the underlying structure that produced those images. It can reproduce the surface. It cannot understand what's beneath it. Every AI-generated logo comes with its required sparkles, set against a pink-orange, blue-purple gradient. The model has learned what design looks like. It has no idea what design is.
The return of the tool-maker
There is an obvious irony here. The technology that has most aggressively automated the surface of design is also the technology that could return designers to something they lost: the ability to build their own tools.
Today most designers are learning to write structured prompts, curating outputs, requesting refinements. They have assumed the role of critic and client. If that is the full extent of the ambition, design will eat its own tail.
But there is another path. AI development tools now make it possible for designers, without deep engineering backgrounds, to build the tools they actually need: tools shaped around their own thinking, their own process, their own definition of what good design requires. Not the tools a platform company decided to build. Not a sandbox with invisible walls. Tools designed by designers, for the specific work designers want to do.
The Eames didn't wait for someone to build the Ka-Zam machine. They needed it, so they built it. The result wasn't just a tool; it was the condition of possibility for everything that followed.
A new wave of design tools is coming regardless. Some will be genuinely useful. Most will consolidate into the hands of the same handful of platform companies, promising efficiency and delivering templates. One size fits all, with embedded workflows and automated craft.
Designers who choose the other path; who build tools that embrace both rigor and curiosity, that balance human-centered practice with scalability, that treat craft not as nostalgia but as foundation; will find themselves doing something the profession hasn't done in twenty years. Shaping the tools. And being shaped, in return, by something they actually made.
We need to define those things.
We need to own those things.
We need to return to shaping our tools.
Not because it's tradition.
Because it's how the best work has always gotten made.
*Often attributed to Marshall McLuhan, this formulation was written by John M. Culkin in his 1967 essay "A Schoolman's Guide to Marshall McLuhan." Culkin was one of McLuhan's closest interpreters, and the line is widely considered a distillation of McLuhan's own thinking.
