Code-Based Design

For thirty years, product design has had an expensive limitation: we couldn't fully experience what we were designing until somebody built it.

So we worked through representations: wireframes, comps, prototypes. Then engineering turned them into the coded product, where missing states, awkward interactions, inaccurate assumptions, and other gaps became obvious. Using the product often sparked new ideas too.

Then we went around again.

Design. Dev. QA. Repeat.

It worked. It also made change expensive.

It was the best design method our tools and workflows allowed.

Today, the tools let us work differently.

Enter code-based design.

Today, I prototype in hours, not days. I can use the design, catch broken states and awkward interactions, tune motion in context, and explore another direction with stakeholders before engineering has committed significant time.

More importantly, that working experience can get into customers' hands sooner. Stakeholders can react to behavior instead of interpreting a comp. Customers can use something real enough to expose whether the idea works. We learn sooner, while changing direction is still inexpensive.

That is useful for designers. It is more consequential for the business. Code-based design shortens the distance between an idea and evidence that customers actually want it.

That can radically accelerate the path to market fit.

In a way, AI is finally letting design work more like Agile was intended to work: smaller bets, working software earlier, tighter feedback loops, and learning through use.

The economics changed

Same two weeks. More directions explored. Faster decisions. Less code wasted.

Code-based design lowers the cost of iteration. More shots, faster signal, less wasted build. That's how you accelerate the path to market fit.

Wireframes aren't going away. Neither are sketches, whiteboards or traditional prototypes. They still earn their keep when they're the fastest way to answer the question in front of us.

The difference is that static artifacts no longer have to carry the full burden of product design. Working code can move earlier in the process, giving teams something usable to share sooner and radically reducing the cost of iteration.

Cheap iteration is a market advantage

Most products don't fail because nobody had ideas. They fail because finding the right combination of problem, product, experience, positioning, and timing is expensive. Companies make bets, put them into the market, and learn which assumptions survive contact with reality.

The faster a team can make those adjustments, the more shots it gets. If one team can explore five credible product directions for roughly the effort another spends getting one through the traditional loop, the advantage isn't simply productivity. It's optionality.

Engineering embraced this logic more than two decades ago. The Agile Manifesto put working software over comprehensive documentation, and engineering spent the next twenty-five years building around smaller releases, tighter loops, and continuous delivery.

Design has often remained one step removed from that loop. Code-based design closes some of that distance.

The talent market is already moving

Design hiring is down, while expectations for the role are expanding.

Move early. The advantage compounds.

McKinsey described AI as a competitive reset, not a lasting productivity advantage. The reason matters: once everyone has access to similar tools, speed alone stops differentiating you.

What compounds is what you do with that speed. Teams that use it to explore more directions, put working software in front of customers sooner, and kill weak ideas before engineering invests heavily start learning earlier. Those learning cycles build on one another.

That is the opportunity with code-based design. The advantage isn't faster design. It's learning what works sooner.

Run the experiment

There will be a learning curve,¹ and changing how work happens carries a cost. The question is whether the return justifies it.

In my experience, it does: faster iteration, faster shipping, and more shots at finding what the market wants.

The grass is greener.