I’ve spent the last year rebuilding my own creative process around AI, while also helping change the way my product team and the companies I advise, adapt to this industry shift.
What’s become clear is that AI doesn’t just make the existing design process faster. Instead its true value is in fundamentally changing what a design team believes it’s capable of.
Suddenly we can explore more futures, turn rough ideas into working experiences, remove the distance between design and engineering, take bigger bets, understand with more depth… the list goes on.
No doubt, AI is magical.
The leverage is enormous… but spending time in this space has taught me that using more AI doesn’t automatically produce better design. Instead, AI is most powerful when designers treat it as a tool to bend to their will — extending their capability without outsourcing their judgment.
But that balance is hard to strike, especially when we’re being asked to move at lightspeed. These are my principles for expanding what a design team is capable of, while keeping human judgment firmly in control:
Protect your unique way of thinking, and build on the practices that strengthen your conviction. AI should challenge and strengthen that reasoning, but never replace it. Protect your ability to explain why something exists, what you believe and what evidence changed your mind. Uncertainty may slow you down, but a lack of conviction compounds against quality. Ultimately, building a defensible point of view matters more than the tools used to reach it.
AI wants to resolve ambiguity quickly — overlooking the fact that time spent understanding a problem can be just as valuable as executing on it. Design often requires us to remain uncomfortable long enough to understand why the obvious answer isn’t right. Be suspicious when an uncertain problem suddenly feels certain, and cautious when the answer is difficult to reverse. An even greater gift than instant answers is knowing when uncertainty is worth dwelling in — or when it’s time to say “fuck it” and move on.
Speed is an outcome, not a method. AI weakens the old fast/good/cheap trade-off — we can have all three — but speed should be used to deliberately to preserve the organizational room to slow down when it matters. Adapt your design process so speed becomes the default when the cost of failure is low: jump to code, own quality-of-life improvements, enhancements and nice-to-haves. But speed in one area creates the capacity to spend disproportionate time on the opportunities that deserve it. The point isn’t maximum velocity everywhere, it’s preserving the ability to work at different speeds.
Feeling the emotional tone of a design is just as important as its logical rationale. Use AI to recreate full experiences so you can feel the anxiety, anticipation, responsiveness, trust and delight of a real journey in ways static mocks never could. Letting AI carry more of the predictable structure and flow creates more time for the illogical magic that connects humans. As AI makes functionally adequate software commonplace, the weird, warm and disproportionately considered details will become evidence that a real person donated part of their soul to the process. Us humans can feel the difference.
AI is a chaos machine. It dramatically reduces the cost of creating complexity without reducing the consequences of maintaining it. Be alert to unnecessary tools, workflows, abstractions, code and process — we’re drawn to systems because they feel like progress, without always appreciating the cost of ownership. So unless something unlocks a meaningful capability that benefits your practice, or removes workload that doesn’t serve you, it probably doesn’t need to exist. AI should remove work, not create supervision.
AI unlocks an exciting future for designers. It expands capability and magnifies judgment. It lets us do things we couldn’t do before, faster and with greater ambition — but greater capability doesn’t automatically produce better work. The quality of what we make still depends on knowing what matters, what doesn’t and when to stop.
Bolting AI onto an existing design process can increase speed and output, but expecting it to improve quality by default is missing the point. The real opportunity is to rethink the entire process around what’s now possible — and lean into what we can achieve now that so many old constraints have disappeared.