Designers are best positioned to build with AI
For years we were told to "learn to code," and in a humorous plot twist the hardest parts of spinning up code have gotten abstracted away. We sit at the center of product creation: we pull requirements from PMs and execs, turn ambiguity into specs and prototypes, build alignment, then stay close through implementation > QA > shipping. Now we can do that end-to-end. There's a broader question of whether we should do this, but let's tackle that in another think piece.
We're faced with a massive rethinking of what our role is in building and what the longevity of our careers looks like. What I'm hearing and seeing is fear. So let's talk about it.
The fear is real, but it's pointed at the wrong thing
There's a lot of anxiety in design circles about AI right now. Some of it is about jobs, some of it is about craft, the worry that what we've spent years learning is about to be devalued. Some of it is just the discomfort of feeling like a beginner again. I understand this fear, but I think it's misplaced.
The threat isn't that AI replaces designers. It's that designers who understand AI will move faster than those who don't, and the gap will show. I've seen something similar happen before: the emergence of UI/UX design then its morphing into Product Design where visual skills became the gap that widened across the industry.
What AI actually changes (and what it doesn't)
AI makes generating things cheap and fast. Layouts, code, copy, whole design systems, variations, components. All the stuff that used to take hours or weeks now takes us minutes. And if you're Ralph Wiggum-ing it, you might have an agent power through an overnight session and finish a whole multi faceted product in one go.
What AI doesn't make cheap: knowing what to build, knowing who it's for, and knowing when something is good enough to ship versus when it's missing the point entirely.
That judgment is the whole job and it's always has been. I think about the quote from The Lion King: "You have forgotten who you are, and so forgotten me. Look inside yourself, Simba. You are more than what you have become." Sorry I couldn't help fitting this in. Anyways, I'm telling you: Remember who you are.
You've been training for this
The skills that make someone good at working with AI are the same skills designers have been building for years. Here's an unorganized list:
- Talking people off the ledge when they say "make it pop"
- Turning ambiguous goals into concrete requirements
- Describing what you want with enough precision that someone else can build it
- Iterating as you learn what actually works through UXR
- Catching the details that break an experience
If you've ever taken a mangled idea from a stakeholder's head and turned it into something that shipped, you already know how to work this way. The collaborator(s) just changed. And hopefully they stop asking you to "make it pop." You get to ask it to do that now.
The process question is our conundrum
The long-winded, artifact-heavy design process that many of us were trained on for the last decade isn't valid today. When building was expensive, we needed weeks of research, specs, and handoffs to earn the right to try something. I remember working on Interested Candidates while at LinkedIn (which eventually became the Open to Work profile ring) and it took me and Matthew Rendely, my PM at the time, about 4 months just to get approval to run a beta. The risk was high, so the long process of convincing people made sense back then.
When building is cheap, you can start with a prototype and learn what the problem actually is by putting something in someone's hands. A working deliverable isn't the end goal anymore. It's the starting point. This isn't about abandoning craft. It's about moving your craft closer to users.
A low-stakes way to start
You don't need to overhaul how you work right now. Just try one small thing today. Claude Cowork just launched, so go try it out. Ask it to help you organize your desktop. Pay attention to the UI and how it's trying to help you achieve your goal. Or open up Cursor and tell it: "Make me a website so I can track days until summer. Make no mistakes."
Take a small idea you've been sitting on and try building a rough version with an AI tool. Don't worry about doing it "right." Just see what happens. You might hate it. You might find it useful. Either way, you'll have an opinion based on experience rather than fear.
It's a lot tbh
I want to end with saying, there's a lot of stuff happening in our country and world. All this dreadful news compounds what we feel with the swift movement of AI in our industry. You're not alone in feeling misplaced from various angles. Hang in there and I believe in you.
Visit CarmenOS.AI — it's my AI playground.
Find the icons I used in the header here.