If you're designing with AI, you're already behind.
Stop building with AI and start building for AI. Craft is no longer the moat — tribal knowledge you've acquired, style, systems, skills and speed are.
AI enables us to prototype and build faster than ever before, but much of that output is noise, not signal. Iteration after iteration can look like progress. So can the sudden ability to make everything animate. It isn’t progress unless the product gets meaningfully better.
Let me explain, all of this amazing craft and interaction you’re doing, it’s wonderful now but to think that AI can’t do it in the near future is ridiculous and you’re fooling yourself. You’re setting yourself up for failure.
Switch from building with AI to building for AI. You have a window in time right now to build the systems and processes ready for that future.
Optimise for how the depth of the input you put in creates an output that is inherently world class.
You have to understand that the SaaS you’re building, the app you’re building, the website you’re building, soon, can be built by anyone in an afternoon, but what people can not emulate or fake is the tribal knowledge you’ve acquired about what you’re building. This is the secret, unsaid sauce when someone uses your product that I believe AI will struggle to do for some time yet.
This is one of your key inputs. The other inputs are your style and your interactions.
Look, there’s still a place for truly bespoke software, software that has a single wow moment that has been painstakingly crafted, but the reality is that’s not what gets traction or scale outside of the tech bubble, and even within the tech bubble our industry leaders don’t do it. Think Shopify, Stripe, Slack, Greptile, Graphite, Notion, Claude, ChatGPT. Can you say any of these huge titans of the work we do everyday have any elements that AI couldn’t whip together right now?
The best of us right now are releasing features at 10x, 50x, 100x the speed they once were. Time spent on interaction is still valid, but there is a point where it stops adding enough value to justify the time. Past that point you’re polishing something that may change within days once users get their hands on it. The skill is knowing when the interaction is good enough to ship, learn from, and improve. It’s a balancing act.
We’re in our own Industrial Revolution. Steam engines, power looms and railways didn’t arrive fully formed. They were built, put to work, tested, broken and improved. That’s us now, except the feedback loop moves at software speed.
If you’re spending time “crafting”, you’re too slow. If you’re spending time as a committee deciding if a sentence could be worded better or if the information architecture is exactly right you’re too slow. You’re designing for a world that may as well be 100 years in the past because whilst you perfect those pixels the world is moving on and you can’t stop it. Keep up any way you can, don’t put a single road block, gate check, or process in the way, work out a way to go as fast as the world is.
This is where the craft goes. Not feature by feature, polishing every interaction in isolation, but into the system you’re designing for AI. Spend the time making that system look great and feel great. Give it the right components, rules, constraints, examples and taste, so quality is present in every output. The craft hasn’t disappeared; it has moved up a level, where it compounds with every feature you release and grows faster than ever before.
All designers are poised to make more impact than they ever have. We have the eye and knowledge to make AI products world class output, we can make the things that are in our heads instead of just drawing them, but we have to stop thinking only like designers and have to borrow the mindset of a developer, the mindset of a PM, and we have to combine that with what we know, and build.