The end of “design”

Lots of takes like this lately, surely not unrelated to the past year’s layoffs in tech and design. But I think there’s more to it than that. Traditional software product design is being disrupted from two complimentary directions: hybrids and AI.

Hybrids

My background is in design, but not pure design. My first real websites were ones I built myself, coded in PHP, HTML, CSS, JS. And these sites were built not for just my own enjoyment but for real communities and users.

In these ways I was part engineer, and part what would later be considered “product”.

To me there’s a spectrum that looks a lot like this, and everyone that makes software falls somewhere within by way of their interest, perspective, and aptitude.

In my later startups, and my latest role at a larger tech company, I skew towards design and product. And my days are as much filled with user research, prototypes, and design critique as they are with scoping, roadmapping, customer demos, and go-to-market.

Some would argue this makes me less of a designer, but I think in the modern sense it makes me even more so. And it’s a trait I see among some of the most experienced and capable designers I know. It’s inevitable. The best designers understand product and engineering. Same for the best engineers. And there are more of these senior-level practitioners than ever.

“Design is how it works” after all. Which means not just how it works in code, but how it works for the user and for the business.

This combination of disciplines might be frowned upon in some ZIRP-era large VC-backed tech companies, preferring specialization and the hierarchies that follow. But while there is some virtue to that structure, and will continue to be for certain businesses for awhile, it does not represent the future.

Further reading:

Source

AI

Like many over the past couple years I’ve learned to use AI in my daily practice.

One of its most empowering angles is how it makes software softer. You can fill in gaps in an app’s logic and ability with fuzzy LLM queries. Like “list the key names and dates mentioned in this document”. Or having LLMs write code directly which I’ve found surprisingly effective.

In the hands of a design/product hybrid with enough technical knowledge to be dangerous, it’s remarkable the kind of zero-to-one work that can be accomplished without much traditional development.

Now one to ten, ten to a hundred? That’s a different story. But given this reality, what is a prototype? Is it a slideshow of rectangles and animations, or does it become something more? Something that starts to actually work. Something that can be validated with real users and real data.

Add to this the advancement of design systems, of component-driven design workflows and tooling, and most recently the ability for AI to assemble UI and whole screens much like it does paragraphs of text.

The future of crafting software is undergoing a radical shift, putting broader power in the hands of the individual, and reshaping how teams of these individuals are composed, and what roles lie within.

The future

Some designers might see these trends and be concerned. But you shouldn’t be! If you love creating software to benefit real people, there’s never been a better time. And five years from now that will seem like an understatement.

Tools are evolving not just to help improve the quality and speed of design, but to empower designers and our unique perspective to close the gap between plan and reality. To remove layers of abstraction and get closer to manipulating The Thing Itself.

And that’s what we’ve been working toward this whole time, isn’t it?

——

I wrote this on my own blog last week, sharing here as I work on figuring out what the heck this place is!