Every tool promises to change everything. Here's how SeaLab thinks about what actually matters when AI enters the design workflow.
The Noise Is Real
It's hard to open a design newsletter, attend a conference, or sit through a team meeting right now without AI coming up. New tools drop weekly. Figma is building AI in. Every SaaS product has an "AI-powered" badge on it. And the discourse swings wildly between "AI will replace designers" and "AI is just autocomplete and you're all overreacting."
The truth, as usual, lives somewhere in the middle — and it's more nuanced than either camp tends to acknowledge.
At SeaLab, we've been doing AI UX work for long enough to have real opinions. We've designed AI features for products, built conversational AI experiences, applied our C.L.E.A.R. framework to AI-powered interfaces, and yes — experimented with AI design tools ourselves. Here's what we've learned.
What AI Actually Does Well in Design Workflows
AI is genuinely good at a specific category of design work: the high-volume, lower-stakes, generative stuff. First-pass explorations. Copy variations. Rough component generation. Research synthesis. Documentation drafts.
These are tasks where speed matters more than nuance, where the output is a starting point rather than a final deliverable. AI excels here. It's fast, it's tireless, and it doesn't need coffee breaks.
What AI is not good at — yet, and possibly for a long time — is the judgment work. Knowing which exploration to keep and which to kill. Understanding why a particular interaction pattern will confuse a specific user demographic. Navigating the political reality of a client organization to get good work through. Sensing when a design is technically correct but emotionally wrong.
Those things require experience, context, and taste. They're not automatable. They're what senior designers do.
The Risk Nobody's Talking About
The thing that concerns SeaLab most about AI in design isn't job displacement — it's quality displacement.
When AI makes it cheap and easy to produce lots of designs quickly, the temptation is to skip the hard thinking. To generate options instead of defining problems. To show clients volume instead of clarity. To ship something that looks like a design system without doing the systems thinking that makes a design system actually work.
AI amplifies what's already there. If your foundations are strong — sharp problem definition, solid user research, clear design principles — AI makes your work faster. If your foundations are weak, AI makes your weak foundations faster. The output scales. The quality doesn't.
This is exactly the argument we made in our work on design systems: LLMs and AI tools make a strong design foundation more important, not less.
How SeaLab Is Using AI
We're using AI tools in our workflow. We're not shy about that. But we're deliberate about where.
We use AI for first-pass copy and content generation — it's a good starting point that we edit heavily. We use it for research synthesis when we have large volumes of user feedback to process. We've experimented with AI-assisted component generation in Figma for low-fidelity explorations. We use it for documentation drafts.
We don't use AI to make design decisions. We don't use it as a substitute for user research. We don't let it shortcut the strategic thinking that's at the center of good UX work.
And we apply the same standard to our clients' products that we apply to our own work: AI in a product should make users more capable, not more confused. It should give users control, not take it away. It should be explainable, accountable, and honest about its limitations.
The Clarity Part
The haystack is real. There are a lot of AI tools right now, a lot of AI features being shipped, and a lot of noise about what it all means for design.
Here's our clarity: design is still about people. The tools change. The fundamental goal — creating experiences that work for real humans, in real contexts, with real needs — doesn't.
AI is a tool. A powerful one. But the designers who use it well are the ones who already knew how to do good design work before it showed up.
Want to talk through how AI fits into your product or workflow? Let's talk.