Comments
![James Odene [User-Friendly]](https://res.cloudinary.com/cea/image/upload/c_crop,g_custom/c_fill,dpr_2,q_100,f_auto,g_auto:faces,w_36,h_36,b_white/Profile/lfbicsnwyhmah2ybuurk)
TLDR: Amazing websites don't need to be expensive anymore with AI vibe coding, stop wasting money on fancy pants agencies.
What are we doing?
Recently I heard of an organisation that was pleased the range of quotes they had received for building them a new website had all come in just under 100K. One. Hundred. Thousand. Pounds.
Honestly, this isn't out of the ordinary, I know many organisations that gravitate towards expensive web builds. I regularly quote for website builds at User-Friendly (my top price with all the bells and whistles is closer to 6K) and so often, the client opts for a web agency charging them +£30,000. I have this sense that it's an area most don't feel comfortable with, so just assume more money is best.
I typically build in Squarespace or a bespoke design in Wordpress, however recently I've been building more and more with AI 'vibe' coding. It's made me question whether I should still be building them the way I have been for years. (I just rebuilt the Userfriendly site and it's nicer, more responsive and endlessly more adaptable. Before it was a little awkward and stuck at the platform limitations).
I'm want to offer a small trial at the end of this post, but I wanted to explain the thinking first, because I think this matters beyond whether anyone wants to work with me.
The thing that's changed my mind isn't just how quickly and easily you can build a website now (it's taken weeks down to days, or hours). It's how quickly you can change anything...
Usually, someone in an org notices something isn't working, or you have a new programme that you want a new functionality fit for. Maybe people don't understand what you do, or a programme needs its own page or form, or you've got an idea for explaining something that your current website can't really accommodate.
Then you have to figure out what the solution might be, figure out who to hire, explain it to them, get a quote, wait for them to do it, give feedback. Hopefully you end up with what you wanted. And probably a few thousand pounds later...
yuck
AI coding
With AI coding, you can have a conversation about what you want to change, get a version in front of you, and keep working on it. Some of those smaller things could be done over a coffee break. Bigger things will take longer, obviously, but even being able to try an idea before deciding whether it's worth pursuing is a pretty big change.
You also have more freedom over what the website can actually be. I've spent years building within website platforms, and they're useful, but you eventually hit the edges of what they let you do. Then you're finding workarounds or changing the idea to fit the platform which is such a killer for good ideas. And some of the limitations are still infuriating, like why can't you have a dropdown menu in Squarespace where the parent page is also clickable?! Or create a table. These are the market leading platforms and after years of using them I can honestly say, they're slow to sort themselves out, and now AI is leaving them for dead.
I'm not suggesting everyone should immediately bin a perfectly good website. If yours does what you need, great. But if you're about to spend money on a new one or harbouring an ugly duckling, maybe it's time to jump ship.
The RFP
I've typically charged around £5K for a website through User-Friendly. For this trial, I'd like to work with three organisations at £2K per site to build or rebuild their website using AI, then hand over the keys and help them understand how to keep working on it. I already have one website confirmed, so there's two slots left currently.
We'd agree what's included, ongoing costs (not to me, but for the site hosting etc) and the handover before starting. I'm particularly interested in working with teams who have someone willing to have a go at making changes themselves, because I want to find out how well this works once I'm out of the picture.
If you're interested, email me [email protected]
FWIW, I think there are some comms risks here — I've noticed that Claude sites in particular start looking really same-y without a good design eye editing them. The new User-Friendly site I would have immediately clocked as Claude-made (and applied some automatic discount to it based on this had you not said it, based on things like):
Little glyphs and words above headings
Random italicized words for emphasis
Using 01, 02, 03 in things that don't need numbering, with the leading 0.
Along with the neutral color palette.
Here are some examples of sites that are clearly made by Claude, and have a lot of these features:
Claude Academy's example websites also look this way: https://academy.claude.com/use-cases/create-a-custom-webpage
I used Claude design with just this prompt: "Can you make me some slides? Don't use my design system - just use whatever you feel like. Fill them in with filler content."
Here are some screenshots of that:
I think that working with a good designer (like you!) seems great for folks doing this, but also, if we're already concerned about EA websites being really same-y, this poses the same kind of risk, or at least for me, as someone who has starting feeling negative about seeing endless amounts of Claude-designed things.
Hey, thanks for the comment. This actually wasn't made by Claude, and a few of the things you highlight are preexisting features of the UF brand.
I take your point on things looking similar, but we already have this issue with lots of platforms. I could likewise spot a Squarespace site a mile away. That said, at least this time we're tied to an engine that's got significantly less boundaries.
Also, in terms of my offer, I'll still be designing the website, not just saying to AI 'make this site for me'
Useful! And to be clear, I think the website looks great and much better than anything I linked - sorry if that didn't come through! I think my point was something like, there are these design trends here that might start to cause people to react negatively to AI-designed stuff, if it is too obvious, in the same way that many people react to obviously negatively to AI writing, so I am much more excited about everyone working with designers who vibe design still, over making their own websites (which is the same pitch you're making!) as I still think designers are adding a very helpful eye.
Hey, thanks for that clarification. I entirely agree - that's why there's still a cost here for my time. It's all too easy for a 'novice' to just clunk through with AI and create something bad. With all of AI use, human taste and experience remains entirely critical to success.
I don't mind this at all. My new site for First Domino is Claude made and would admit it looks similar to other sites you list, but doesn't bother me at all. In a previous life as a web developer this would have taken me a big chunk of time to get right and in the current world would have cost a lot of money to outsource, but instead I got this up and running very quickly and can make edits easily as well. I've received pretty positive feedback as well. I'd encourage other orgs to do this as well, at least out of the gate - I've seen people overthinking the website and other non-essentials early on with their startups as a pretty common failure mode.
If and when the vibe changes and e.g. Claude sites are seen as negative it won't be hard to hire an expert to give a different flair.
I agree with all these points! I think I under emphasized in my original that I think working with designers is still really valuable, and definitely agree that Claude will make a better website than me by default. I think I just see risks around design going the same way as AI-generated writing, where if it is too obvious, people will just write it off. E.g. when I see a service provider with an AI-generated website, my mind does a slight discounting on them being real/legitimate, because I have a bit less of a sense of a human having done the work to make the thing.
Hey Devon, thanks for adding your thoughts.
The money and time saving is a huge part. I'd favour a good answer quickly than a perfect answer slowly in most cases.
I think it's also clear to me that a lot of the things highlighted are actually closer to website design trends than 'Claude' design. I've designed websites for over 10 years and there is often a sweep of similar looking features that sites take on. When I first started, they all had a picture at the top that wasn't full screen, with a solid colour navigation bar. Then suddenly everyone went to a full screen image that was visible behind the navigation bar. Then it was large titles... So in a sense, you're picking up 'current web design trends' more than 'a tell' for AI Claude design.
I disagree. This feels close to letting the perfect be the enemy of the good. If you're running an org on a shoestring and your total marketing budget is less than 50k, a vibe coded website is probably much better than Wix/Squarespace/spending over 10k on a professionally built website.
I agree with your broader point, and am pretty confused as to why any small org would now spend serious money on a website (although I do like And-Now’s work, I figure it is firmly outside the price range of small orgs).
But I’m sympathetic to the top-level comment, and to the downvoters, because the vibecoded look does have a decent cost, and this cost is higher than the cost of usual faddish web design, because even faddish web design used to be gated by a decent price tag. The average obviously vibecoded website is a thrown together side project, and this fact reflects poorly on serious organizations that share the style.
And this could all be a moot point, since the cost savings are so large and maybe it’s hard to cheaply fix the vibecoded look. But in fact it’s very easy to make a vibecoded website look more normal! I’m not sure I can accurately describe the full process, but the first step is to remove the extraneous elements and switch fonts. After that, it only takes a couple of hours and a few honest conversations with your peers! Just do it! Please!
IMO it goes Wix/Squarespace site < vibecoded site < nicely vibecoded site.
Sure I discount a little when a site is vibecoded. But I discount even more when it's Wix/Squarespace - 'you're using Wix in 2026?? You could make a substantially nicer and easier to maintain website in an afternooon'.
And I think when I see a professionally made site I might do a little discount as well? Beyond a nicely vibecoded site, the marginal improvement probably isn’t worth the marginal cost, so I find it hard to believe a professional build was a good use of money.
Thanks for saying this, I was thinking it too. I designed a website recently and tried to make it not-AI... would it pass your filter? https://projectcoa.org/
I think it would have!