Dylan Field, the CEO of Figma, said something at a recent talk that has stayed with me: the first thing AI gives you is generic by definition — because it's the average of everything it's seen. Your job, as the person with something to say, is to push past it.
He wasn't arguing against AI. He was arguing for something harder to automate: taste, craft, and point of view. And if you run a business and you're thinking about who should build your website — a tool or a person — those three words are the clearest answer you'll find.
The average problem
AI website builders work by drawing on everything that already exists. They've seen millions of layouts, thousands of colour palettes, every conversion pattern ever studied. They produce something that looks reasonable — competent, even — in minutes. And that is genuinely useful. It is also, by definition, a composite. A weighted average of what has worked before, for someone else, in someone else's context.
The problem isn't that the result is bad. The problem is that it is familiar. It reads like a website. It feels like a template. And in a market where your potential clients are scanning dozens of options in a sitting,"looks like a website" is not a differentiator. It is camouflage.
Taste is not a style preference
Field makes a distinction that is easy to miss: taste and craft are different skills. Taste is knowing what's good — and being able to articulate why. Not just "I like this" but "this works because it creates the right feeling at the right moment for the right person." It's a judgment call with a reason behind it.
An AI can match a mood board. It can replicate a visual direction you've described. What it cannot do is hold your brand in one hand and your client's unspoken expectations in the other, and find the specific thing that satisfies both. That requires someone who has spent years developing opinions — not prompts.
When a designer chooses to break a layout rule deliberately, or pulls a colour slightly warmer than the brief suggested, or strips a homepage down to almost nothing because the product earns that confidence — that's taste. It's the kind of decision that doesn't survive a committee and doesn't emerge from a generator. It comes from someone who has looked at enough work, built enough sites, and developed enough of a perspective to know when to follow convention and when to break it.
Craft is what happens after the first draft
The other half of the equation is craft — which Field defines as pushing past where others stop, at every level, from the big structure down to the smallest detail. Spacing. Line height. The exact phrasing of a button. The way a section breathes. These are not finishing touches. They are the work.
AI is very good at first drafts. It generates fast, it iterates when prompted, and it will produce something usable in the time it takes to make a coffee. What it won't do is stay uncomfortable. It won't stare at a headline for twenty minutes because something is slightly off. It won't rebuild a section from scratch because the hierarchy is wrong even though it looks fine. It won't care that the mobile version of your footer is technically working but feels rushed.
A human who takes craft seriously does all of those things. Not because they are slower — because they are paying attention to the gap between done and right.
Point of view is what your clients actually feel
Here is the thing most AI-generated websites are missing that you might not be able to name immediately: they don't seem to believe in anything. They present information. They tick boxes. They guide you through a sequence of sections that follow the logic of a website without ever committing to a position.
Your business, presumably, does believe in something. You have a way of working, a kind of client you do your best work for, a reason why you charge what you charge and not less. That is your point of view. And a website that doesn't express it — that could belong to any of your competitors with a name swap and a colour change — is not helping you. It's just taking up space on a domain you're paying for.
A good web designer does not just execute a brief. They listen for what is underneath it. They ask why this business exists, what it's actually trying to say, who it's trying to reach — and they build something that expresses that with enough specificity that the right person reads it and thinks: yes, this is for me. That is what a point of view does. And it does not come from a prompt.
So what is AI actually good for?
It would be dishonest to frame this as AI versus humans with a clear winner. AI is genuinely useful in the process of building a website. It accelerates research, helps generate copy options, tests layouts, and removes a lot of the friction from moving between idea and execution. Figma's own tools — including their MCP, which lets designers round-trip between canvas and code — are built on this reality. The tools are not the problem.
The problem is the assumption that access to the tools is the same as the judgment to use them well. Field put it directly: 60 percent of Figma designs are now made by non-designers. That is a remarkable shift in access. It is not a guarantee of quality.
When everyone can generate something that looks like a website, the bar does not disappear. It moves. The question is no longer whether you can have a website. It's whether your website has anything worth saying — and whether it says it well enough to earn trust, attention, and eventually, business.
This is still a human argument
I am not writing this to be defensive about my craft. I use tools. I use AI in my process. What I am saying is that the tools work in service of something a tool cannot supply on its own: a developed perspective, real judgment about what is working and what is not, and the discipline to push past the first acceptable answer until you find the right one.
Your website is not a document. It is the first conversation your business has with someone who doesn't know you yet. It should feel like it was made by someone who cares what they think.