How AI is changing web design agencies: experience just got more valuable
This morning I had a call with Dan Suhr, a Wix developer in Australia and a fellow Wix Leader. It was nine in the morning here in Hampshire and just after six in the evening for him. My coffee, his dinner.
Between us, we've been building websites for close to 30 years. Dan's been on Wix since day one and has spent the last couple of years helping people on the Wix Studio forum. I came over from marketing a little under 10 years ago.
We kept circling back to one question: how is AI changing web design agencies?
Where we landed surprised me a bit. AI has made the generic part of the job almost free. At the same time, it's made the experienced part matter more. The agencies doing well use AI on purpose, because they already know what good looks like. Everyone else is drifting.
The noise got louder this year
Dan reckons the industry's changed a lot this year alone. I'd agree.
Anyone can get a decent-looking site out of an AI tool in a few minutes now. Dan thinks you can usually spot them. The layouts and the styling start to blur into one.
I'm less sure that part is new. AI learned from the most popular designs out there, and designers have always borrowed from the same places. AI just made it obvious.
What I do know is that none of my projects has ever finished looking like it started. I'll sometimes start from a bought template. By launch you wouldn't recognise it, because no client wants what they see first. It's always "can you move this, can you change that, can you do it the other way?"
AI output works the same way. It gets you somewhere. It still needs a final push from someone who knows the client.
Dan put it well: "My main thing is trying to figure out exactly what their business is and what they want to achieve out of it. AI can't do that as well at the moment."
We went into the design side of this in AI websites look great, but who are they actually built for?
Is the Wix Marketplace dead?
That's the word in the Wix Facebook groups.
Our last four major projects all came from the Wix Marketplace. Each client found us there and got in touch directly.
I'll be straight about the other side of it. Those four came from about seven leads over three or four months, and earlier this year was quiet. If you judge the Marketplace on volume, it probably does look dead.
But every one of those leads was a good fit, and the conversations were easy from the start. They didn't come for a facelift. They came with real business problems that an AI build won't solve on its own.
I asked Dan if I'd just been lucky. He had a story that says otherwise. A business owner rang him out of the blue recently. He'd tried to build his own site with AI, wasn't happy with the result, and wanted help doing it properly.
My theory is that AI is sorting the work. If you want something quick and simple, you can now try AI first, and for plenty of businesses that's a sensible place to start. Dan tells smaller enquiries exactly that. The businesses that still come to an agency usually want help with something bigger than the website itself.
The businesses that still come to an agency want help with something bigger than the website.
Which got us onto the bigger question: what are clients actually coming to us for now?
What does value mean to us now?
When we started, value meant a good-looking site, delivered on time. Now it means a site that does a job in the business. Getting there takes a lot more thinking than building.
My background is in marketing, so I always wanted the sites I built to do something as well as look good. What's changed is how much of the job is that thinking.
Dan's path was similar. He started out building sites for family and friends and taking whatever small jobs came along. Over time he realised the knowledge he brought was worth more than a quick build. He moved towards businesses and agencies that know what they want. The work got more interesting too: custom code, integrations and problems the editor can't solve on its own. The part he enjoys most is figuring out a problem and getting it to work.
For me, the shift came through research. Years ago, agency founders I learned from told me a big pitch used to take them a serious amount of time and money in research alone.
I can do most of that over a coffee now. Before I write a proposal, I use AI to understand the business properly.
What the CEO has said in public about where they're heading. How long they've been trading and how big the team is. Every discovery call gets transcribed and goes into the project's knowledge base. By the time I write the proposal, it's built around their plans. Clients tell me it feels like we really heard what they wanted.
That research shapes the project. A business with a big team and plans to grow needs a different site from one that's just starting out. The proposal should show we know which one we're talking to. Bigger organisations notice when you've done your homework. It tells them you'll think about where the business is going as well as how the site looks.
AI gathered the facts. Ten years of projects told me what to do with them. Without that experience, the research is just a pile of numbers.
AI gathered the facts. Ten years of projects told me what the business actually needed.
When does AI save time, and when does it eat it?
Using AI on purpose sounds simple. In practice, this was the most honest part of the call.
Recently a campervan dealer we work with texted me. They wanted a finance calculator on their site. I was meant to be doing something else, but I wondered if I could build it with Claude. I spent the whole day on it.
It turned out really well, and it showed me how many small custom jobs clients could never afford before.
I still knew I shouldn't have spent that day on it.
I read a post that asked a simple question. Do you get AI to do the work and then do more work? Or do you get AI to do the work and go and enjoy your afternoon? I know which one I am. With AI I end up working more hours, and my focus wanders.
"You can definitely get lost in that," Dan said. His rule is simpler. He only uses AI when it saves him real time. Big changes, yes. Research before a build, yes. Quick minor changes, no.
Before he starts a new site, he gets AI to look at every competitor in the area. That way he doesn't copy what anyone else is doing. That used to take him two hours. Now it takes two minutes.
What separates his approach from my day on the calculator is knowing the job well enough to see where the tool fits.
The same goes for the editor itself. On the call, we swapped two habits that beat the AI shortcuts. I let Claude update a live site through Wix Studio's Content Mode, which keeps the design out of its reach. And for mobile layouts, I reach for Remove Overrides long before the AI optimiser, which Dan finds less predictable than making targeted adjustments manually. Neither trick is clever. You only learn them by building a lot of sites, and I'll walk through both properly in separate posts.
A new editor won't fix a broken business
Wix Studio 2 is on the way, and we're both looking forward to it. Dan's had a look and says it has upgrades people have been asking for for a long time. I wrote about why we're paying attention to Studio 2.
But I said something on the call that I'll stand by. A lot of people are betting on the next shiny tool when it's the business that's broken. A new editor won't tell you what a client's business needs. It won't stop you saying yes to every rush job, or spending a day on something nobody asked for.
AI makes it easier to say yes to everything. Experience tells you when a yes would short-change the client.
The things that made us good before AI are what make AI useful now. You know what a client means when they say they just need something simple. You know which button in the editor fixes the problem, and you know what a project needs before anyone asks. Ten years of getting those things wrong and then right is what lets me use AI on purpose.
A new editor won't tell you what a client's business needs.
If you run a Wix agency and you're working through the same questions, I'd like to hear how you're handling them. And if you're a business owner who tried the AI route and wants a second opinion, drop us a line.
Frequently asked questions
Is the Wix Marketplace still worth being on in 2026?
In our experience, yes, if you care about fit over volume. Our last four major projects all came from the Wix Marketplace, even though we had fewer leads than in past years. The clients who found us there came with real business problems to solve.
Will AI replace web design agencies?
AI is taking over the simplest end of the work. Business owners who want something quick can now get a basic site from an AI tool. Projects that need someone to understand the business and the platform still come to agencies, and that's where our work is.
What does an agency add now AI can build a website?
An understanding of the business. AI can produce a decent-looking site, but it can't sit down with a client, learn how the business works and decide what the site needs to do. An experienced agency also knows the platform well enough to use AI where it helps and leave it out where it doesn't.
What does Remove Overrides do in Wix Studio?
Remove Overrides resets an element on a smaller breakpoint back to its design and layout on the largest breakpoint. Select the element, click the More Actions icon, then click Remove Overrides. It's the quickest way to rescue a mobile layout that's drifted too far from the desktop.
Can AI safely edit a live Wix Studio site?
It can if you limit what it can touch. Wix Studio's Content Mode only allows changes to text and media, so the layout and design stay out of reach. Check every page yourself before you publish.
.png)