[00:00:20] Nathan Wrigley: Hello there and welcome once again to the WP Builds podcast. You've reached episode number 457, which has a very short title. Its title is building a human centered web by saying no to AI, Andy Bell on ethics agency life CSS, and the open web. See extremely short. It was published on Thursday, the 19th of February, 2026.
My name's Nathan Wrigley, and before we begin, just one very short bit of housekeeping. If you like what we do over at WP Builds, and I know many of you do, because of our analytics and our subscriber list and things like that. Well, if you've got a product or service in the WordPress space and you would like to get it mentioned in our podcast around about here in the episode, but also towards the end of the episode. Well, I've got some good news for you. You can do that. Head to wpbuilds.com/advertise to find out more. It will give you some information about the different things that we do to help products and service companies in the WordPress space get their bits and pieces noticed. If that's not your bag and you don't really like reading things online, well just send me an email [email protected], and I'm sure we can sort something out.
Just also to mention, if you don't know, we have a Monday show, we do it live, it's called This Week in WordPress. I repackage it as a podcast episode the following day, so every Tuesday. But if you want to join us live for that, you can do that. wpbuilds.com/live. It's at 2:00 PM UK time every Monday. And then, like I said, we repackage it into a podcast episode on a Tuesday, and we'll also be sending you another podcast episode every single Thursday. If you wanna keep updated, wpbuilds.com/subscribe to find out more.
Okey dokie, what have we got for you today? Well, today it's a chat, not necessarily strictly focused on WordPress, because I am chatting to Andy Bell. Now, Andy Bell has all the chops, all the credibility you could possibly muster in the agency, and particularly, CSS and open web space.
I actually saw him at a conference in London in 2025 and thought it would be good to get him on the podcast, and so here he is. And we talk about a whole range of things. A whole piece about his agency's decision to really push against AI and to not use AI in any way, shape or form. Not to produce content with AI, not to work for companies that are involved in the AI space. So that's a really interesting topic.
But then we also spend a lot of time talking about the updates to CSS, and how it's now becoming a real proper language, which is gonna help you develop on the web. The open web, open standards, and a whole lot more.
It really was delightful connecting with Andy and hopefully you'll be able to go and find the different bits and pieces that he's doing. He's got CSS courses, he's got an agency, and his work is truly incredible. I hope that you enjoy it.
I'm joined on the podcast by Andy Bell. Hello.
[00:03:29] Andy Bell: Hello.
[00:03:30] Nathan Wrigley: You have, without a doubt, the easiest to pronounce name that I've ever had on an interview. Not quite often. It's an unpronounceable name to me at least.
But, Andy Bell, nice and straightforward. Here comes a dad joke for you. Well done for getting out of erasure, duh. Okay. And with that out of thought.
[00:03:51] Andy Bell: I can never compete with the Andy Bell. Never. Yeah.
[00:03:54] Nathan Wrigley: Andy be,
[00:03:55] Andy Bell: Yeah.
[00:03:56] Nathan Wrigley: those of you who weren't born, I guess anytime in the last 30 years will not understand how low that joke was. Anyway, moving on. Andy's joining me today. I've just given him the potted history of how come I've come across Andy, and it goes a bit like. This, I'm gonna read it into the record quickly.
Andy has been working on the web, I think for 15 plus years. Andy, I think it's fair to say, is one of the few people out there who really understands CSS. If you like me and can pretend to understand it, you can get away with quite a lot. But I think it's fair to say that Andy really does understand CSS.
If you were listening to this podcast many years ago, David Wamsley and I. Split off and did a different podcast all about using CSS to build websites. That then led to me finding Andy, I subscribed to his content. And, yeah. So I've summarized your life.
[00:04:50] Andy Bell: Thanks. Yeah. Nice and
[00:04:52] Nathan Wrigley: tell us who you actually are?
[00:04:56] Andy Bell: I think you've done a pretty good job there, but, so I got, into CSS, outta spite more than anything else, so I used to, yeah, I'm a traditional designer, so that's what I started as. And then, I used to design. Websites and hand over PSDs to developers and then they'd send me something back that just resembled something abstract rather than accurate.
And I was already messing around with flash and stuff at the time and I was like, I'm sure I can fix this myself. And then, ah, I was like, I'm sure I can just learn CSS. It can't be that hard, can I? And then the rest is history, really. I got really into writing CSS so much so that I can't remember the last time I did any design work.
So it is, Complete switch. Yeah, it does help that I've got a very talented designer that works for me, so I'm, I am fully redundant on that front, but it's, it's also, yeah, I find working in the browser just so much more rewarding than drawing rectangles in Figma.
[00:05:57] Nathan Wrigley: Yeah. Nice. I like it. Did you, properly obsess about it then? with the benefit of hindsight, did you look back and just really go down a bit of a rabbit hole with CSS?
[00:06:07] Andy Bell: Yeah, I was. It's a, it's something I've been sort saying for a while. I, see CSS as a design language, it's a programming language for designers and like I, it speaks to you like when, if you're that way inclined, if you think like a designer already, like it makes a lot of sense even down to the way that the cascade and specific to work a word.
I can't say by the way, 'cause I'm also northern, but, really, not helpful.
[00:06:35] Nathan Wrigley: of work.
[00:06:36] Andy Bell: Yeah, isn't it? Yeah. It's like a doctor not being able to say
[00:06:41] Nathan Wrigley: that's right. Yeah. Yeah. it's okay. They can't write,
[00:06:45] Andy Bell: Yeah, neither could I, but the, but that's it. Yeah. So it's, always it's always resonated really well with me as CSS and that was like in the old days as well.
The CSS we have now, is unbelievable versus what we had at the time. And yeah, I've just, it's just always been a interest of mine. I remember in a conversation with a friend who did, CSS consultancy about five, six years ago. And it was just after CSS grid came around and he was like, I'm gonna get out of the CSS world.
'cause I think the language is finished now
[00:07:23] Nathan Wrigley: That was just when it was getting interesting.
[00:07:26] Andy Bell: how wrong, can you be? it is, it is changed like almost unrecognizable since then. yeah. But it's great. I love it. Absolutely love it. It's such a fun sort of language to work
[00:07:38] Nathan Wrigley: So this is a WordPress podcast, and we've got this curious relationship with things like C-S-S-H-T-M-L, JavaScript and what have you in that a lot of that is taken away by bits and pieces that you add on top of WordPress plugins, themes, and things like that. It kind of abstracts that away and in, in many cases.
It's kind of encouraging you to kind of more or less ignore the CSS behind the website. It, you might have a slider or a box to fill in, put a number in here and that represents the margin or the padding or what have you. And everything takes place in front of your eyes. And I very much fell into that camp during the early days of CSS.
So when tech, so when we stopped using tables and CSS came along, I learned everything I could about CSS, but then kind of got into the whole WordPress thing and thought. Those UIs are now good enough for me. They'll do what I want. But this recent collaboration for this podcast I'm doing with David, it's called the No Script Show, and he's really fascinated in the same way that you were and probably still are.
He is just absolutely taken by CSS. He's fascinated by it. And so he's peeled back the curtain to me. and I honestly didn't realize that it was as exciting as it was. We are almost at the point it feels like that most of the JavaScript that you ever wanted to move pixels around on a page are completely unnecessary now, and it's changing.
Month by month, week by week. There's a few people that you can follow. I'll put their names in the show notes, Andy being one of them. But, it really is like this language which allows you to do everything without the need for, JavaScript necessarily. And do, can you, do you see this sort of trend continuing?
Is there much work still to be done or are we kind of,
[00:09:21] Andy Bell: Yeah. Oh, there's huge amounts to, to go. still. I do. So I think that. The browsers they're working in really work really well with each other. So the IOP stuff is great and the focus seems to be good too. there's a few bits that I'm not overly bothered about. there's been a lot of focus on Masonry, for example.
but it brings back with it some issues. So for example, if you shuffle the content. In different orders to what the source order is, then it's gonna affect the tab in order for keyboard users. So the need to address those sorts of things. And, but then on the, subject of JavaScript, I think it's CSS now is putting JavaScript where it's supposed to be on the client side, which is supposed to deal with interaction and events and.
Like you say, it is been used to move pixels around the page. Also, in the past it's even used to style UIs with CSS and js, which I've never, truly understood why you'd wanna do that. but. So I like to see it now as CSS is enabling JavaScript to be the one of the pillars in its own right as is HTML and then CSS.
So it is creating that proper separation of concerns, which I think we've been asking and wanting for decades at this point. yeah, I really like to see it and it's only gonna get better. it is CSS and it's only gonna improve as time goes on. So you're never gonna run out of things to learn with it because it's just so vast.
and capable too.
[00:11:05] Nathan Wrigley: yeah. Honestly, dear listener, if you haven't checked in the last, I don't know, three or four years on the state of CSS, it, it is transformational what has happened. It really has moved away from what you think it can do. Like margins, padding, that kind of thing, to being able to do so many curious, interesting things.
There are so many developments dropping all the time. There's lots and lots of real interest in, out there. There's, there's blogs that you can follow. There's people that you can follow. There's teams that you can follow. Mozilla and Chrome and all of those kind of people who are just pushing the boundaries.
And, Andy kind of feels it feels like you are. In that mix somewhere. Is it kind of surprising to you that people view you as some sort of C-S-S-C-S-S expert, or is it, do you feel like it's a, I don't know, a result of work hard work done over many years?
[00:11:58] Andy Bell: yeah, there's, been a lot of graft. I can never fully accept it. I think it's just, that's just natural imposter syndrome. I always like in these sorts of conversations, say there's better people, there's better people out there to listen to. But yeah, like I think one thing that's benefited me in my career is, having, it's not even really a niche, it's just having this.
Like focal point, like I've, I'm designer and front end developer, but so obviously deal with all of everything, but writing about CSS has always been more interesting to me than anything else. So I think it's just a case of volume and. velocity of the amount I've published over the years about specifically CSS stuff that you just become associated with it.
And I'm not mad about it to be honest, because it's the thing I'm most interested in anyway. So I'll keep doing it and it seems to be paying off quite well, so
[00:12:57] Nathan Wrigley: You've, you've got, you've got several enterprises that I know of. There's maybe more strings to your bow, but there's there's the sort of work side of things and you've got a very nice domain name. It's set Studio, which is pretty cool. is that a team of you? Is that like a collection of freelancers?
How does that whole thing work?
[00:13:15] Andy Bell: There, there's a full team. So there's four of us all together. so we've got me got, Jason, our designer, Leanne, who runs our tech stuff, and then we've got Vicky the producer, who's the nice person that clients get to
[00:13:28] Nathan Wrigley: That has to be one.
[00:13:30] Andy Bell: She, yeah, she's the friendly face. Yeah. But we will, organize them within an inch of their lives.
But, yeah, they don't get to see me unless they're in trouble, a lot of the time.
[00:13:43] Nathan Wrigley: But, you have had some really ha heavy hitting clients. We'll get onto the CCA lily stuff in a minute, but you've got, You've got some really, you've got a real like roster of some amazing clients. I was reading it earlier, the one that sticks out for me was, the NHS, but there's a whole laundry list of them as well.
A again, it, feels from where I'm sitting that you, you represent this kind of oh, we want to be like Andy. We want these big sort of clients. that is the goal, of most agencies to get these big clients. Is it all roses over there with the big clients or can that be kind of interesting and fun?
[00:14:23] Andy Bell: the summary I like to use for big companies, the big, ones is they're really useful for your logo reel on your website, but they're absolutely horror. They're absolutely horrifying to work with at the time. I, actually tend to pr like our actual target client now in, in the urgency is like mid-level SMEs, like people were, the clients that we work best with are the ones where we can actually talk to the decision makers in the organization, like the, high up decision makers, like founders and stuff.
I think. Big, companies are great, but the move at a glacial speed and when you're an agency, that's really difficult to work like that because you obviously work at a breakneck speed a lot of the time. So it's a, it is really complicated. but like I said, you get on that logo reel and it really does add to the credibility of the place.
So it's worth all the
[00:15:19] Nathan Wrigley: get the feeling that quite a lot of people use logo reels in a, let's go for fraudulent way. I'm not entirely sure that when I see some web designers websites that is in fact who they have worked with, but, let's assume that Andy's telling the truth seems like an honest guy.
here we go. Google, the NHS Harley Davison Ali. Ollie, why did I say Ollie? I have no idea what Ollie is. Oracle,
[00:15:45] Andy Bell: cool.
[00:15:46] Nathan Wrigley: B Unilever, the Natural History Museum, British Cycling, capita Vitality, vice Media. And he said, swelling. As we say, these say these words. That's so impressive. well done.
That is pretty remarkable. Yeah, yeah. however, let's move on. We've got, you've also got this other sort of more educational piece, side to what you do, which is called Pick a Lily. tell us about that. What are you doing over there?
[00:16:16] Andy Bell: So particularly started, first off, I thought, let's find a name that's really hard to say and
[00:16:22] Nathan Wrigley: Yeah, that's okay. Good job. Success.
[00:16:26] Andy Bell: tick. and then the reality is I spotted, I don't know how I found the domain, but I found the domain. I was like, that's a really cool domain I'm gonna make. So I bought it, just no intention of doing anything with it. I just bought it and then it just kept like burning in the back of my skull and I was like, I really need to do something with that domain.
So it started off as a newsletter.
[00:16:50] Nathan Wrigley: okay.
[00:16:50] Andy Bell: a really simple newsletter. And then, so like 2018, and then it evolved into like my blog about CSS. So instead of publishing as much on my personal site, I, published on that instead. 'cause I thought, oh, can I create myself a focal point for writing about the web and design and CSS and whatnot?
And, it's just evolved really. I went on a sort of break when I founded the studio, because obviously you. You start an urgency and everything that goes into chaos mode for a while. So I just didn't have the bandwidth to be writing as much content. And then as it, last year we, the start of the year, we had our little.
All hands. And I was like, I'd quite to bring CCA Lily back to life. Is anyone interested? And, everyone was like, hell yeah, let's do it. So we just did like full redesign, full rebrand, the full kit and caboodle. And then we just, I've always had in my mind, I wanna publish courses and I want to publish, more, much more sort of free education for people as possible, because I wouldn't be where I am without.
Other people doing the same. So as I see, it's just my duty to do the same now. and I, wish that was still a sort of key principle in the web industry because I think all of us who've been around the block, a long time can attribute our success to. Like blogs that we read and other sort of free resources.
So yeah, so that, that's a big part of what we do. And it seems to have gone down really well over the last, nearly two years now since we brought it back to
[00:18:37] Nathan Wrigley: Yeah, the URL, it, it really is quite good. I'm gonna get all the amount of I and Ls. Completely wrong. I'm gonna try and get it right. it's pic Lili PI double C-A-L-I-L-L-I. I think I did that
[00:18:53] Andy Bell: Yeah. Yeah, you
[00:18:54] Nathan Wrigley: it in the show notes. It's a bit like spelling Mississippi or so.
[00:18:57] Andy Bell: the only person on the planet who can spell picker Lily. First time. Yeah.
[00:19:02] Nathan Wrigley: just bookmark it, put it in the browser
and it somewhere.
But there's a whole load of other stuff that you do as well, or at least have done, like you've done courses for Google, you've got, books that you've written and what have you. And it sounds from the outside at least. Anyway, and this is kind of where this podcast is going, it sounds You have this amazing life where work kind of is unending.
There's this forcet of work. You turned it on sometime 12 years ago and it just kept on coming. Big clients, bigger clients, yay rah, everything's fine. But the bit that got me to contact you, and I'm not that, I don't normally contact people out the blue, so forgive me. but the bit that got me contacting you was a piece that you wrote not that long ago.
end of November, 2025. So 27th of November. It's on, Andy's sort of personal website. I'm gonna say it's Bell bz, again, links in the show notes and it was called, it's been a very hard year and the bit that drew me to this was. Like, I just, I'm just drawn to authenticity and people who are quite willing to say that was really bad.
we've had a bad time. Not this constant, I don't know, social media presence, nonsense, that we all seem to have fallen into the trap of never being able to say times are hard. Never being able to say times are tough, always pretending like everything's okay. So you can go and read that, but would you mind summarizing what the year 2025 has been like for you?
[00:20:36] Andy Bell: Yeah, it is been, so I'm glad you referenced that what or what you perceive is different to the reality because I think this is the case for a lot of people, especially based on a lot of people going to touch. And a lot of, some inquiries, which is fantastic. some really good inquiries.
But mainly it was people saying, thanks for writing it because it's exactly what I'm experiencing at the moment. But I didn't wanna broad, I didn't wanna broadcast that because I felt like it would affect my ability to let land work or whatnot. But yeah, the gist of the piece was like this year in terms of I've been doing client services for the vast majority of my career.
I've never experienced, including the pandemic, never experienced like a lack of inquiries, quite like 20, 25. and then, inquiries that do come in that they went against our sort of principles. So one of those principles is we will not do promotional material for AI companies. Like we just wanna do it.
we don't wanna be associated with that. And the vast majority of the inquiries came from like startups who were effectively just a wrapper for open ai, finding somewhere to exploit users for profit. And we just weren't interested in any of that. and it felt like a lot of clients, the spend, drop considerably too, and the spending power, and I think the economies, the, economies plural, have really affected that people, high inflation and.
Donald Trump's tariffs haven't really helped businesses spending power and all of these factors are combined, and it just, it became a very, slow year. So we pushed on Piccolo with that extra time that we had. And based on, so I released a course last year, and so we, all of our projections, based on how that costed, we were like, it'll be fine.
we'll be absolutely fine. It's all good. releasing two courses and it should be all good. But then that Black Friday event started and the numbers were telling a completely different story. So I just thought I, I've had that. So I had that draft for quite a while in my, obsidian that I use, for storing all these things and.
I thought, I don't really wanna publish it at the moment. and then I just saw that, the way that it was trending on that. So I thought, okay, something's definitely going wrong and I don't know what's going wrong. I don't know, if we can get over the noise or anything like that. So I thought, I'm just gonna publish it and I'm gonna see what happens.
and thankfully I did publish it because it really changed the trajectory of everything and. It's, yeah, it's one of those things that I recommend to people, like how do I, like what can you attribute success to? And I'll always say, without a doubt, blogging has, is at the root, like social media been, lucky to have a reasonably large following on social media, but that's got nothing on writing, writing blog posts and, being, getting good at writing and.
That's just the thought process I had really, it was like, what's always worked for me? Blogging, what should I do, blog? And yeah, it worked out really well. And I think it actually told people a story that they probably didn't realize on, like what our ethical boundaries were and what we were actually trying to do, with Pic Lilly and, Set Studio.
So yeah, it worked out good
[00:24:21] Nathan Wrigley: is such an interesting story because I, can totally empathize with you not kind of wishing to click the publish button because it, like we said, everything about the modern world seems to be always best foot forward. Portrayer, the kind of the perfect version of yourself.
Everything's fine, and even if everything's not fine, just stiff up a lit. Just crack on. But you, did exactly the opposite. And that is so interesting that was the lever that, that it turns out many people were, that's what they needed to get in touch. Whether it was just to commend you for your article, which is basically what I did.
or if it was to, okay, now we've got an intuition about the kind of human being that you are.
[00:25:09] Andy Bell: Yeah.
[00:25:09] Nathan Wrigley: Isn't, that I can't even get a purchase on that. I don't really know what I'm trying to say, but there's people got some intuition for you as a, as an individual perhaps, that then trans, that then prompted them to get in touch, leave a comment, or to offer work in your direction.
Wow. Because
[00:25:30] Andy Bell: yeah.
[00:25:31] Nathan Wrigley: it, I feel it, it could have gone either way. But it went the beneficial way. That's great.
[00:25:36] Andy Bell: And that's what the sort of, the, worry was that it would go the opposite way. but then, like I published it and then. But we had a, an inquiry come in and it was a really good inquiry that we pursued and they, read the piece and they were specifically looking for an agency that was anti AI because they'd been, they'd been burnt.
By a couple of agencies who'd fully invested. and so they, they were specifically looking for ones that, there was zero risk that their brand was gonna be, affected by the app. But because, it's probably not a surprise to many people that work in, have worked with agencies, but a lot of agencies have embraced it wholeheartedly because it's a shortcut.
As they see it. And that's the sort of sad, the sad state of affairs really, but a big part of running an agency is going against the perception of agencies. because so many do burn people. And so that's why we call ourselves a studio, not an
[00:26:50] Nathan Wrigley: Yeah. Yeah, yeah.
[00:26:51] Andy Bell: as an attentive that. Yeah.
[00:26:53] Nathan Wrigley: co. Couple of things I'm gonna just draw out, if you don't mind, from what you've just
[00:26:56] Andy Bell: Yeah, of course. Yeah.
[00:26:58] Nathan Wrigley: one is I'm gonna more or less guarantee that a bunch of people listening to this. Are exactly where you've been in the year 2025 and it, wasn't like 2024 or 2023, maybe the phone is ringing less or the email inbox is far less full than it's ever been.
And just a general sort of sense of head scratching. Like, why, what, what's changed? we haven't done anything different. Everything's as normal, but. Okay, just take note. it's not just you. People such as Andy, whose reputation is pretty stellar, this side of the pond or wherever you may be listening to it.
it happens, so you'd have to beat yourself over the head about it. but also that there were curious bit that came outta that I didn't actually know was, I didn't realize that you not only didn't work for people that use ai, but you don't. Utilize AI in your processes. Firstly, is that sentiment true?
Did I just say the right thing? You don't. You don't leverage AI tools to build the stuff you build.
[00:28:02] Andy Bell: No, we've experimented with it absolutely. Like we're technologists, so in the sort of earlier times, before it was stuffed down everyone's throats, like when it came about this feels like a useful thing because, you've got documentation at your fingertips and you've got, the boring things.
oh, gotta build a react component. I can just get this. Tool to scaffold it for me and do all the boring bits for us. And, we did give it a fair, a few different tools, a few different, providers, a sort of fair crack of the whip and like when I really do it, and it was probably right about 40% of the time, which at the time we were like, ah, it's very early, so let off.
And then all harms started coming to light. with ai, people, we talked about this before, like people using it for therapy, and the last thing you need is, and when you, if you're in the need of therapy, is an agreement machine. To be delivering that therapy to you. It's quite opposite of what an actual therapist will give you.
And, this said, there's, it's, you can, it's been directly attribute to suicides. It's been directly attributed to psychosis. And I don't think we're, one thing that I always saying in these sort of situations is, what you are reading is a small percentage of the actuality. Because a lot of people will report this and this won't be reported.
So if you are seeing a handful of cases in the media, there's a hell of a lot more cases in reality. And it really, the cat's already fully out of the bag. you cannot put the toothpaste back in the tube, as it were with ai. Like it is fully there now. And even if all of the AI companies went out business tomorrow, people could still run these models locally on the machines.
So it's, there forever. But the lack of responsibility is the thing that really, boils my blood as
[00:30:12] Nathan Wrigley: Yeah. Yeah.
[00:30:13] Andy Bell: the lack of ethics and the lack of safety mechanisms put in place is, scary because we're all people, work in technology. So we have the ability to look at technologies like this and.
Sort of question them and and really analyze the usefulness. But a normal person doesn't have that and they don't realize it's a lot of
[00:30:38] Nathan Wrigley: What's kind of curious though is that there's this, strange juxtaposition of you're into tech, therefore you must be into this aspect, this pathway in tech, the AI bit. if you like working with computers and you're building things for the web, This thing is, this is irresistible.
This is absolutely perfect, and I think it's totally all right to say, actually, you know what, there are bits of that really trouble me. And, anybody listening to this podcast over any length of time will know that, that is where I stand. There are bits of it that really trouble me. Just segue in slightly.
So have you, out of the post that you wrote and the responses that you've got, have you decided to rejig your. studio, have you decided to offer a different marketing spin so that is a more prominent bit of what you do? Do you feel that there's a market for kind of like the no AI version of a web development agency?
[00:31:39] Andy Bell: Oh, for, sure. I think, something we've been playing with internally, like messaging and stuff, but. something we're gonna be doing in, in, oh, sorry, that's not my mic. Sorry. Listeners. what we're gonna be doing in 2026 is, so I used to do consultancy more than anything, so it's like CSS consultancy, but one thing we're gonna start offering in the new year is basically giving companies a pathway away from it and surveying the damage for them.
That's already been. Being done on their, our business and code vases and helping them find a way out. Because with any technology, once you got a certain way into it, it feels impossible to get out again. I think React is a really good example of that, where a lot of companies would fully embrace it and then. They feel like the get out of that is too much work. So then they double down on it and then keep going. And I saw a lot of this in the CSS consultancy as well. Like people would, sorry. Teams would like fully embrace Tailwind CSS for example. And then they feel they'd get to a point where it was becoming more problematic than helpful for them.
And they'd bring me in because we dunno how to get out of this. So a lot of the time my, my work was helping them create that sort of transition plan and bringing a bit more education into the team and skilling them up on, on the cascade and specificity and, all that sort of stuff.
Like really nailing the basics into them and then giving them a route out. So it's like a transitional rollout of, the framework. I feel like we can do the same with a, with. Companies that have been burned by AI and, there'll be a lot more than you imagine that are, suffering from that.
So it's something that we are gonna do and probably position ourselves more and more as the anti AI studio as well. I imagine people listening and thinking Andy's has completely lost his mind. but I think one of the things that AI's got going for it is. The softest, easiest media, treatment that you could probably imagine for a, technology and a group of sort of billionaires that run it.
I think if the media actually did their job with ai, I think we'd all be. More terrified of it, and I'm much more critical of it. And it's, something that, it's a post I'm working on at the moment. I normally do like a wrap up post where it's okay, I did this many blog posts and this, that this year I'm going for the jugular with ai.
And I'm saying, look, we've gotta take this thing on. We've gotta take it on as a community and as an industry because. It's not, doing its job. It's actually dangerous. and we do it. And a, big part of that is countering the very easy treatment the AI technology system is getting in the media and actually prompting people, not that.
So prompting for, like prompting people to. Write about when it goes wrong for them because it does go wrong for a lot of people. It comes up with bad results. it just, it destroys people's code bases. I was reading one when it deleted someone's entire route directory on the computer.
[00:35:06] Nathan Wrigley: that was horrific.
[00:35:08] Andy Bell: We need more people to write about this stuff when it happens because I think if you like create this sort of huge volume of.
Reality. I think it will help to focus the mind because I do, empathize with people, like it feels like an exciting technology. It feels magic, but it's not. It's a probability machine that is, every response is a hallucination, not just the bad ones, all of them. It's effectively making things up in order to satisfy a prompt, and it's making those up off the huge amount of training data.
Most of which was stolen. so it's making, so it's, it is doing that in a qualified manner, but it will also. Always want to satisfy your prompt, hence why it'll just make things up in order to satisfy the prompt. I think one thing that would improve AI tenfold is if it couldn't find an answer in his training data, it would
[00:36:09] Nathan Wrigley: honesty, that's just
say,
[00:36:11] Andy Bell: I don't know the answer.
I don't actually know the answer, but here's like a, here's a link to some, like a Google search results that I've just tuned for you. So it's, it's a bit more specific in the search query and stuff like that. So just to give you a helping hand, but I can't tell you the answer to that.
I can't, imagine that ever happened though.
[00:36:31] Nathan Wrigley: did you ever hear, this is such a weird like analogy, but did you ever hear of the, this thing called tulip mania? in the Netherlands, and I'm gonna make updates here. but let's say it was like the 16 hundreds or something like that. There was this whole thing where tulips. Just became incredibly popular.
absurdly popular. The value of a tulip was probably worth more than its weight in gold. And it just went on and on, and, tulips became just you would ship tulips around the world rather than any other commodity, if you possibly could. 'cause their value was so high until this moment, this exact moment in time where it sounds like more or less, everybody just turned around and went.
They're tulips. They're just tulips. What the heck? And then suddenly their value kind of just descended. It went to more or less zero. It went to the value of a tulip. You know what it should actually be. And you can see the, I guess there might be examples, wall Street crash, that kind of thing. I have intuitions along those lines with all of this.
I have intuitions that it's, we're just wrapped up in this hysteria about it. We're kind of ignoring. All of the harms, often ignoring all of the mistakes. when it, when we ask it to do something and it comes back with something and you immediately find fault in it and say, hang on a minute, and then it goes, oh yeah, of course you are right.
[00:37:56] Andy Bell: You're absolutely
[00:37:57] Nathan Wrigley: absolutely right. Here we go. And, and you imagine that you employed a hundred people in your organization and reliably, a hundred of them every day lie to you, just did you do this work? Oh, yeah. That's all sorted. Yeah, I did that. Did you actually do it? No. No, I didn't do that work.
you, they wouldn't Last a minute, and yet we seem to have an endless capacity for forgiveness. Of this kind of thing. And I just, I genuinely don't know where that capacity comes from. I think it's exceedingly exciting, it, possibly more exciting than a chew lip even. but nevertheless, people seem to be beal as this promise that it's always gonna get better.
My understanding with LLMs at least, is that the, The capacity is the, horizon for it getting better and better is receding further and further into the distance because it's, no longer able to crunch any new human made data because it's done all that already. As you said, it's kind of taken the stuff off the internet and scraped it all.
So the only thing it's got now is training on its own output.
[00:39:07] Andy Bell: Yeah. it's unbelievable, isn't it? And, the, it is just a, it is a, you want a case study of a very good marketing campaign. And I think the AI industry is one of those, And I think it does help that you get sort of influential people in the industry will constantly gone about how good it is and how helpful it is and not show you the, not show you the scratch pad as it were of all the times it goes horrifically wrong.
and you're right, the LLM stuff is plateauing. and there isn't much more they can do unless. They built a lot more data centers. and it to me that sort of suggests that the technology is incredibly inefficient and, not as good as it, it probably could be and not being developed as well as it could be.
'cause you get the, I think one real eye opening moment for me was when the team behind deep seek in China, they matched and they enough matched, I think it's chat GPT-4 at
[00:40:15] Nathan Wrigley: Yes, I remember this story. Yeah.
[00:40:16] Andy Bell: process, on old processes. And you could even run it on a laptop if, I remember rightly. and to me that sort of suggested as always with Silicon Valley, especially that maybe.
Maybe the hype was slightly over, hyped as it were because the sort of, the quality of development that was delivered was probably not as high as you might imagine from Silicon Valley. Silicon Valley's always had this reputation of the best, working there, but I can tell you from experience, that's not the guess.
[00:40:51] Nathan Wrigley: Do, you remember? it's really, we have such short memories in tech, don't we? but it really wasn't that long ago that bitcoin. Bitcoin, everybody's gonna be using Bitcoin. We're all gonna be using Bitcoin to buy everything. And of course, we must buy as many as we can right away satisfy the need to make more Bitcoin.
And, and that just kind of silently went away. Nobody's talking about that. NFTs, pictures of monkeys. Terribly important to have pictures of monkeys, at great expense.
[00:41:26] Andy Bell: Only a sensible person spends 200,000 pounds on a
[00:41:29] Nathan Wrigley: very good. And then the web, Web3 0.0, that kind of thing.
[00:41:32] Andy Bell: yeah, yeah,
[00:41:33] Nathan Wrigley: but they're kind of quick, quickly forgotten, oh, we are moving on now we've got the next thing.
And I, I feel that AI has the capacity to be that, but you can imagine in two years time, you and I'll come back, there'll be robots everywhere. AI will
[00:41:49] Andy Bell: Yeah.
[00:41:49] Nathan Wrigley: taken over and be absolutely amazing and will be Oh, okay. Yeah, we slightly wrong there, but, I don't.
[00:41:55] Andy Bell: Willing to be proven wrong on this, but I feel I'm feeling very confident in my opinions. Yeah.
[00:42:00] Nathan Wrigley: I don't have those intuitions. I just, and all of the stuff that you said, the bit that never really gets talked about, the environmental impact, the fact that we do really genuinely seem to have, a user base of, in many cases, in, in the part of the world where I live, at least, young people are kind of getting into using AI for needs that would be best, best provided by humans, let's put it that way.
mental health and things like that. But in the absence of that, in the real world, they turn to these because they simulate that. But I don't, the outcomes don't feel great. There's this whole AI thing called the alignment problem. And the alignment problem is, basically is it aligned to our.
To humanity's, best interests. And I, don't really see much evidence of that. I see an alignment to billionaires, billionaires capacity to be wealthier than they already are, but not a lot else. I.
[00:42:56] Andy Bell: the thing is for me, is the, I think one of the fundamental problems with, LLMs in general, and I think this is what's really kneecap in them as well, is. Is the way that they're programmed to respond is to be agreeable, at all times. So there's the meme, isn't it?
You're absolutely right. because that's where it'll come back when you correct it, it'd be like, you're absolutely right. The thing that's always confusing me about that though is that people saw that immediate agreement and thought, wait minute, what else are you lying about? that doesn't seem to clock.
But then for like people that don't work in this industry. one, one thing is there's, no sustainable profitable model for AI at the moment at all. Like hundreds of billions of dollars have been blasted at this problem. And in order to make that sort of money back it, it has to be.
Incredibly popular in the mainstream market and paid for, by people in the mainstream market. But the vast majority of people in mainstream are actually using it for free. and the, and it is probably not given them enough use to warrant paying for it also. and I think the sort of the, era of subsidized pricing that's in place is not gonna last too much longer either.
So I think people now say you pay $10 a month, for example. You can live with that. But if that suddenly turns into 50 to a hundred dollars a month, that you starts to really question, is this giving me that much value? and I, just can't see a pathway forward for it. I listen to, Terron,
A journalist
[00:44:49] Nathan Wrigley: He's, fantastically entertaining, isn't he? When he speaks.
[00:44:53] Andy Bell: Yeah, yeah. Yeah. He seems to be hyperfocused on AI right now and not Yeah. Yeah. And hyper. Yeah. And, he's a AI marketer's worst
[00:45:03] Nathan Wrigley: Oh, he's got all the numbers. Yeah.
[00:45:06] Andy Bell: He has the stats and he has, he does unbelievable levels of deep research on this stuff.
And I think if you are starting to question ai or even if you're listening to this now and you're like, these guys are talking absolutely rubbish, like it's the best thing ever. Listen to the last, even the last six months of Ed's podcast and especially his monologues that he does. And then come back and tell me that we're talking
[00:45:34] Nathan Wrigley: Yeah,
[00:45:35] Andy Bell: because I think he'll convince you.
[00:45:36] Nathan Wrigley: He really has been looking at it. Like the financials of it, I think is fair to say. That's is, where he is most acerbic and most effective. He looks at the numbers and I have no capacity to keep up with that because that's not my thing, but this is what he does.
He digs into the numbers and if we're to believe that he's true and I have no reason to believe that he's not. the numbers really don't stack up. it is much like we've seen in the past, where we've had gold rushes that then kind of collapse. I'm thinking, I think he's used examples of the railroad in the United States where it, it, lots of people invested a huge amount of money off the strength of loads of people traveling by rail all over the place.
And these people just never did travel by rail. So the whole thing kind of collapsed and, it kind of, it. It struggled on and is what it is today, but there are examples in history where these kind of things collapse and it does appear that a huge amount of money has gone in. But all based upon future promises.
Of what will be achieved. And generally the gold standard seems to be a GI, this sort of artificial general intelligence, which can kind of do everything, which to me is kind of the most scary if anybody actually does do that.
[00:46:51] Andy Bell: Yeah, the, thing like, I used to be mildly scared of that, but then when I see the incompetence coming from Silicon Valley, I'm thinking, nah, it's never happening. Just I don't think quantum computing's ever really gonna happen, or certainly not in the next couple of decades, at least. I think it's, a, it's something that will come, I think.
But it's certainly not gonna come off the back of LLMs. that's for sure.
[00:47:18] Nathan Wrigley: am really interested by the, pivot that you've done though. I in the same way that, everybody could go to Ikea. And get bargain basement furniture. There's an awful lot of people who don't want IKEA furniture. they want the sort of craftsman built, bespoke, artisan approach to whatever it is that they're doing.
And maybe at the moment we're just kind of trying to figure out where humans fit into this jigsaw puzzle and, what the, pay rates are gonna be in the future. But I can't. I can't help but commend you for what you're, that, that approach, that idea of like humans first, no, ai. Like I said, we'll come back in two years time and we'll both be destitute, but, until that
[00:48:02] Andy Bell: Yeah.
[00:48:04] Nathan Wrigley: I, feel like you've got a, you've got a really interesting niche to speak to.
You've got a really interesting. Set of people that you'd wanna speak to and dare I say it, this is gonna sound awful, dare I say it. They kind of strike me as the people that I would personally wish to work alongside, if they kind of see the value in what they're providing to you and that you are providing back to them.
There's a nice kind of symbiosis there. There's a bit of humanity injected back into the work, which, which I think is being lost, gosh.
[00:48:36] Andy Bell: I think that, a big part of the thing that we do as an urgency is like we, we lean really heavily on web standards and. Progressive enhancement and building experiences that work for everyone. Like regardless of the device capabilities, the network capabilities, and that if you are a, company that's in a, you're turning over a few million a year and you can see there's a pathway to grow.
There's no safer bet. Than leaning into the web platform, because if you are gonna grow at an exponential speed, if you, grow on the back of, for example, like next JS or like heavy framework usage, you can introduce all technical debt. There's gonna be very hard to pay off. as your team grows.
It narrows your hiring pool too. which, is not very helpful. There's, always been this attitude in tech for the last few years of. We use React 'cause it makes it easier to hire, but it makes it easier to hire bad developers. in my experience where if you, we put people on this sort of pathway of, we're gonna lean into the platform because you can build in iterations and you can build in cycles.
And you're building on a really solid foundation, by leaning into it and simplifying things. And it seems to resonate with the right sort clients for us. Like I think clients that can see further than their end of the nose, it resonates really well with them. Like when you have that conversation with them and where they're thinking sort of five, 10 years ahead rather than what's happening this year and next year.
It's, what's coming down the line for them. And yeah, it always works by putting them on that
[00:50:25] Nathan Wrigley: It is so curious to me that all of these kind of frameworks and what have you came around, obviously, I'm using WordPress, which is a, you, is definitely a, sort of a, path that you go down and, you get into that ecosystem, you use it and what have you. But one of the things that came out of this podcast was, with David was the web as a platform.
Is fast approaching, it can do almost everything. Now that you imagined that you would wish it to do anything that you, not anything but a significant amount of the things that you would have relied on some framework to do is now being built directly into, browsers. The browsers are talking to each other.
They're building stuff in concert with each other. there may be a month drift here or there, but. Broadly speaking, they're adopting everything at the same time. And you can go and check out those open sources to see where they've got to with those kind of things. And, banking on that feels like the safest ground you've possibly got.
[00:51:26] Andy Bell: Yeah, so yeah. it's getting near 20 years for me now and. I've seen a, I've seen a lot of cycles in that time, and the only thing that hasn't changed in that whole time is that the web wins every time, and the web platform always wins
[00:51:45] Nathan Wrigley: Hmm.
[00:51:47] Andy Bell: so it feels. I think there's a benefit in that.
So say for example, your freelancer or an agency, there's a real benefit in learning a lot of these technologies because clients will come wanting these technologies and sometimes you just have to serve that. I think WordPress is actually a really good example of that. Like clients will come specifically for WordPress and their right to do And then with that WordPress though. I think any sort of, anyone who's worked WordPress for a long time will tell you that relying on a lot of plugins, for example, is not the best way to get a high performing website because plugins will inject all sorts of nasties. I think, I know the Gutenberg editor wasn't very popular in the WordPress community.
and I think it still isn't. but I think that sort of shifted the balance a little bit. where especially when the full site editing stuff came about, where there was a push towards lighter weight footprint. But so sticking with the WordPress thing, like building your own custom theme, for example, was a much better way of.
having a high performing WordPress website, like the Set Studio website used to be on WordPress and it was a lot faster than a lot of static sites because we we embraced using, building the theme from
[00:53:10] Nathan Wrigley: Minimal. Stripping it out.
[00:53:12] Andy Bell: absolutely minimal. Yeah, there's nothing in there. we off, it's from the olden days now, but Elliot J stocks his, stark esteem,
[00:53:22] Nathan Wrigley: There's a name. Yeah.
[00:53:23] Andy Bell: that, that was my gateway into WordPress, was the darkest
[00:53:28] Nathan Wrigley: you read the Net magazine by any chance for you, or did you.
[00:53:31] Andy Bell: yeah, I used, I mean there's a little side note here, and we'll get into this maybe another time, but I used to be a professional net mag troll also, which because me and Linberg were friends with each other, so I used to see like very weird quotes in the little tweets bit, and it was just me messing around.
[00:53:51] Nathan Wrigley: I remember once going into a shop and buying a copy of T net and the lady who was behind the counter, she was obviously incredibly smart. She said to me, you're buying a magazine dear about the internet. Can't you get it on the internet? And I remember thinking, she's right.
[00:54:09] Andy Bell: She's absolutely right. Yeah. there's a reason they didn't survive, but yeah. Yeah. And that's, so that's that era. And so we, came out that project like at the time, whereas Let's use Starkers as our base. And even though the theme was however many years old, since it had been changed, we didn't really have to do much to get it up to speed with WordPress.
I think that's one of the strengths of WordPress as well, is how backwards compatible is. So we brought it back in line with so modern WordPress standards and then, yeah, the site was unbelievably fast. the other reason that we moved. Platform, from WordPress test, it's now an astro is we just preferred the authoring process of working
[00:54:57] Nathan Wrigley: Interesting. Yeah. Yeah.
[00:55:00] Andy Bell: and, we wanted to just test some ideas that we had for that platform as well. There's no better way to test ideas than on your own company website. so we did that as part of a redesign effort last year and, but still It doesn't perform any better than the WordPress
[00:55:20] Nathan Wrigley: I was curious.
[00:55:22] Andy Bell: yeah,
Yeah. and,
[00:55:23] Nathan Wrigley: more so that's,
[00:55:24] Andy Bell: but it's just a, yeah, the template's nicer and it just, it feels easier to work with, but mainly because we don't do much WordPress anymore us, so we're not as in tune with it as we, we probably were in the, olden days. But that, that, that's always the case, really, is just lean into the platform and strip as much stuff out as you possibly can, and you'd be amazed at how fast things get all of a sudden.
[00:55:45] Nathan Wrigley: It is interesting in the WordPress space, this old full site editing thing. So we've got classic themes, which are the old version, if you like, of themes, and we've got the full site editing or site editing as it's now called it, it hasn't taken off in the way that I think the leadership had hoped that it would.
The idea, I think, was that it would, skyrocket and everybody had move over. However, the, chatter at the minute is. It seems to be pivoting slightly in that direction. I can't be confident that's the way it'll go, but it does seem that, that interface, that way of doing things. so templates inside of some sort of UI that you can interact with without having to open an IDE and things like that is, is definitely taking off a little bit, but it's, it is a big stretch for, to get people prepared for what that means and how
[00:56:34] Andy Bell: Yeah. It's amazing,
[00:56:35] Nathan Wrigley: and it's quicker, that's for sure. Yeah.
[00:56:37] Andy Bell: Yeah. And I'm glad they're doing that. I think the Gutenberg. Thing, was always slightly confusing at the time because. WordPress is like famous for its community, right? And, the effort that the WordPress organization put into the community and work with the community.
And I found it baffling that they'd, that they would force a feature into WordPress that was designed for people, probably not in the WordPress community that they interacted with on a daily basis. And it always felt to me like it should have been, It should have been an on by default because they were obviously targeting non-technical users with it.
And that makes total sense that, they were directly trying to challenge square space
[00:57:23] Nathan Wrigley: Yeah. Yeah. Yep.
[00:57:24] Andy Bell: there. But in the settings of WordPress, there should have been a configuration where you could say, I, do not want this, I want to carry on using just
[00:57:33] Nathan Wrigley: You had to go out and find third party or, you could write your own
[00:57:36] Andy Bell: had Yeah, exactly.
Had to install and Yeah,
[00:57:40] Nathan Wrigley: Gutenberg. They're still very popular actually.
[00:57:42] Andy Bell: Yeah. Yeah, weirdly enough.
[00:57:45] Nathan Wrigley: Those kind of popular. I'm a big fan, but it's absolutely not for everybody.
[00:57:52] Andy Bell: No, Cost
not.
[00:57:53] Nathan Wrigley: sadly, we've reached the sort of 55 minute mark, which is probably where we should knock it on the head, but I do want to just make a big call to, like the thrust of what we talked about. I think I would summarize it as follows, if you had a, if you had a tough 2025, firstly, you're not alone.
Even, the big hitters like Andy. It wasn't an easy year, but also, I suppose the second message coming out of it is, if you've got intuitions that are not necessarily a hundred percent laser focused on the future of AI and how that can make you more efficient and you wanna pivot into something a little bit more, the year 2020, before all of that funds started, you have a model in Andy, you could go and check out what he's doing and.
No doubt writing blog posts about it in the future. He did say before the call start, I hope this is okay to read into the record. He did say that he's gonna pull out the big guns, in the year 2026. He's gonna take aim at AI and start firing on all cylinders, trying to dismantle
[00:58:52] Andy Bell: be more Ed. Yeah,
[00:58:55] Nathan Wrigley: So watch this, and, we'll see how it goes, but
[00:58:59] Andy Bell: I've been playing nice for the
[00:59:01] Nathan Wrigley: that's
[00:59:01] Andy Bell: year or so, and a lot of people will still associate me with being a massive AI Craig, but yeah. I'm not gonna hold back.
[00:59:09] Nathan Wrigley: the authentic you. Andy, where can we find you, apart from the bits and pieces that we mentioned at the top of the show, is there any sort of particular place you push
[00:59:17] Andy Bell: Yeah. So I tend to, in terms of social, I spend the vast majority of my time on Blue Sky. so my handle on there is Bell dot. Bz. and that's also my personal site, bell bz. And then on there you can see links and there's just links to everything. So yeah, I'll make it much easier for you rather than me reading a lot of URLs out.
[00:59:43] Nathan Wrigley: Yeah, that, I'm glad that you've
[00:59:44] Andy Bell: you are. Draw me on reading out pic, Lily. Yeah.
[00:59:47] Nathan Wrigley: show with a quite a lengthy list of URLs, which nobody can remember 'cause it's an audio podcast and most people are doing vacuuming or washing the dishes when they
[00:59:55] Andy Bell: Yeah.
[00:59:56] Nathan Wrigley: they, yeah, exactly. Yeah. But anyway, I'll link in the show notes to that and Andy is no doubt available and you too.
Can join the, AI bandwagon and try to pull it off the rails in the year 2026. Andy Bell, thank you for chatting to me.
[01:00:12] Andy Bell: Thank you for having me.
[01:00:13] Nathan Wrigley: Okay. Thanks for listening to that. I hope that you enjoyed it. If you did and you want to leave us a comment, we would love that. Head to wpbuilds.com, search for episode number 457 and leave us a comment there.
Don't forget, we'll be back next Thursday for a podcast episode. We've also got our This Week in WordPress episode every Monday, 2:00 PM UK time.
And if you want to get in touch about advertising and getting your product or service in front of a WordPress specific audience, wpbuilds.com/advertise. Or just drop me an email [email protected].
Okay, we will be back next week. You stay safe. Have a good week. Bye-bye for now.
