AI Hasn't Killed Design's Core Rules, Naya Studio CEO Says

James Dyson built 5,127 vacuum prototypes before one worked — and Saad Rajan, CEO and cofounder of Naya Studio, says that number explains everything about surviving as a designer in the AI age.

Rajan lays out his case in a new essay built around three principles he says have held true through every technological shift. His core definition hasn't moved either: good design is "refined, thought through, and connects emotionally and practically with its intended user." It is a simple output synthesized from complex inputs. AI changes the speed of that synthesis, not the need for it.

The timing matters for practitioners. Tools that appear to replace designers now ship weekly, and Rajan says the questions — what is good design, and how do you still produce it — haunt growing numbers of people across design, engineering, research and product.

What does iteration look like now?

Rajan's first principle: great work must pass through numerous iterations, marked by trial, error and occasional bursts of creativity. He quotes Dyson directly: "I made 5,127 prototypes of my vacuum before it right. There were 5,126 failures. But I learned from each one."

Per Rajan's own math, it takes roughly 100 iterations before something feels right — a pitch deck, a feature, a physical product. That's the practical frame he works with.

AI makes each pass faster and cheaper, but it hasn't removed the drafting stage. Rajan is blunt on this point: it is rare that AI's first output works perfectly, and when it does, designers should be wary that the result may be too generic.

He pairs this with Antoine de Saint-Exupéry's standard — "Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away" — arguing faster cycles demand sharper judgment about what matters and what doesn't.

Who should designers listen to?

Principle two is the power of feedback. Co-creation deepens work, and Rajan counts a wide circle of useful voices: end-users, fabricators, engineers, other designers, marketing teams, plus watching people struggle through existing products in real life.

AI is now another feedback mechanism — one that draws on a vast store of knowledge. Rajan calls that "incredible" but draws a hard line: don't make AI feedback the only feedback. The skill is learning which feedback to prioritize, and when.

He anchors the point in a Gujarati saying: "The wise person listens to all but makes their own decision." Designers, he argues, are the sum of complex experiences and stories. It's easy to assume AI understands that and trust it blindly. It doesn't, and personal voice still has to come through.

Why stepping away still works

Principle three: connecting dots. Rajan compares design thinking to marinating food overnight or letting a curry simmer — chemical reactions and neuron conversations need time in the dark. Stepping back lets designers see the bigger picture, spot drift from the original goal, and separate the important from the disposable.

Here AI genuinely adds something new. Remembering the "in-between" knowledge — the steps and design assets amassed during a project — has always been hard. Capturing it creates a blueprint of how a design took shape, highlights decision points, and provides a framework for future evolutions. It also shows which iterations mattered most and where to draw inspiration from other work.

But Rajan warns that AI's speed can outrun human processing. His prescription is analog: step away from screens, go for walks, see friends and family, visit cafes, then return.

What does this change for practitioners?

Very little in the fundamentals, which is Rajan's whole argument. Every generation has had its technological revolution — pencils to printing presses, steam engines to electricity, computers to automation — and every generation has faced the same fears about adaptation and relevance.

His close is a wager on the profession: designers will survive this era and help shape it, and their adaptability may be the most important quality of all. The working formula stays constant — iterate relentlessly, gather diverse feedback, zoom out before zooming back in.