Skip to main content
wordpress supportwordpress support services

#109 – Ben Ritner on How Kadence WP Managed the Pivot Towards AI

[00:00:00] Nathan Wrigley: Welcome to the Jukebox podcast from WP Tavern. My name is Nathan Wrigley.

Jukebox has a podcast which is dedicated to all things WordPress. The people, the events, the plugins, the blocks, the themes, and in this case, how one company managed to pivot towards AI.

If you’d like to subscribe to the podcast, you can do that by searching for WP Tavern in your podcast player of choice, or by going to forward slash feed forward slash podcast, and you can copy that URL into most podcasts players.

If you have a topic that you’d like us to feature on the podcast, I’m keen to hear from you and hopefully get you, or your idea, featured on the show. Head to forward slash contact forward slash jukebox and use the form there.

So on the podcast today, we have Ben Ritner. Ben is the founder of Kadence WP, a company that specializes in developing themes and plugins for WordPress, with a focus on enhancing the front end of websites.

The team have developed a suite of blocks that enable users to do more with the core Gutenberg editor, and in 2021, Kadence was acquired and became part of the StellarWP umbrella brand. Where Ben now serves as part of the senior leadership team. Under StellarWP, there’s a suite of popular plugins, including Learn Dash, the Events Calendar, and GiveWP.

In this episode, Ben provides us with an insight into how the Kadence team managed the integration of AI into their products, shedding light on the impact that it has had on their website building process.

We discussed the challenges of migrating content between systems, as well as delving into the role of AI in content creation and design adaptation.

Ben talks us through what guided their technical innovations to accommodate AI generated content and images. Ben’s perspective is pretty unique, working for a company he started right through to today, where they have an install base of over 400,000 websites, all of which could potentially make use of AI, but not at the risk of breaking backwards compatibility with their own products.

We chat about the importance of integrating marketing and SEO knowledge for effective website development, and how the recent demand for AI might cause freelancers and agencies to think again about the products and services they offer.

We also get into the way that the team decides what trends in AI to follow. AI is still new, and it’s possible that what’s all the rage today will be superseded at any time. Where does Ben find inspiration for which AI platform to use, and which implementations of AI are dead ends for website builders?

If you’re interested in the impact of AI in web development and are intrigued by its potential, this episode is for you.

If you want to find out more, you can find all of the links in the show notes by heading to forward slash podcast, where you’ll find all the other episodes as well.

And so without further delay, I bring you Ben Ritner.

I am joined on the podcast today by Ben Ritner. Hello, Ben.

[00:03:40] Ben Ritner: Hey. Nice to be on.

[00:03:42] Nathan Wrigley: Yeah, thank you for joining me today. We’re going to talk today about AI. We’re also going to talk about Ben’s history with WordPress and Kadence, which is a product that Ben launched quite a little while ago, but he can tell you all about that. But first, Ben, just so that the audience know who you are, and what your relationship is with WordPress, a moment to give us your potted bio, if that’s all right. So over to you.

[00:04:03] Ben Ritner: Yeah. So I’m the founder of Kadence WP, which is a company that develops themes and plugins for WordPress, primarily focused on the front end of your website. So, how do you make that website look good, responsive, fast?

So we’re, you know, kind of in that page builder space, but we’re doing it from a blocks perspective. So we’ve got a package of blocks, that enable you to do a lot more with the core Gutenberg editor, than what comes in core. And at this moment, I’m part of the senior leadership team of StellarWP, which is a umbrella brand that Kadence is under.

Kadence was acquired in 2021. And so now I’ve moved in with the Stellar org, and we have a number of popular plugins in our suite, including LearnDash, and the events calendar GiveWP. So, yeah, that’s me.

[00:04:54] Nathan Wrigley: Okay, great. Thank you so much. We are going to dwell a little bit on AI. In fact, I think the primary focus of this will be to touch on AI, but we’re going to come at it from different angles. We’ll talk about the products that you’ve got, although that may not be the primary focus of what we talk about. But we’re going to talk a little bit about ethics maybe, or, you know, how AI works, and how it might affect the job market, and whether we’re doing ourselves out of work in the future, and so on.

And so one of the first questions I’ve got for you is surrounding the software that you build. Now, you can go into the details of it if you wish, that’s absolutely fine. Get into the weeds, and get all technical. I’m going to encourage users to go to the following url. So it’s If you want to find that in an easier way, go to WP Tavern, search for this episode, and it’ll be in the show notes there.

But over there you demonstrate how you can create, more or less an entire website. So completely from scratch. You’ve got a vanilla install of WordPress, you install all the bits and pieces that Kadence requires, and moments later, and I really do mean moments. I think you could probably achieve this in under three, four minutes. You’ve got yourself a functioning website.

Now, I’m interested to know, why you went down that route? What were the sort of, the things that you were thinking about? Why website building? Why not, oh I don’t know, just simply block building? I’m curious to know where you are getting your inspiration from. Because if we rewind the clock two years, nobody was talking about AI in the WordPress space, and those that were, were very prescient.

Now everybody’s talking about it. There’s this landslide of products. And you must feel the commercial pressure to get these things right. So, how do you know? How do you judge that what you are building is the right thing for the company’s future? As opposed to just, well, guesswork really. Difficult question, but let’s begin there.

[00:06:47] Ben Ritner: Yeah, yeah. So I think with Kadence, the decision making comes down to, we were really listening to our user base. And we had a product, we have a product called Kadence Starter Templates. Which is essentially a site that is designed, and has placeholder text, and allows you to import that as you’re like, I’m going tolaunch a site quickly. Maybe it’s in my correct niche, because we had over 80 starter templates.

So maybe the images are close to something I’m going to do, or maybe the text is close. But I’ve got to go in there and edit it, and make it my own. That starting place gets past that white screen of, what do I do, I’m staring at a white screen, and somehow I need to build a website from this.

And that was immensely popular. We saw all kinds of users using it, from the agencies using it as a, I want to get started quickly on a site, that they had already had designed for, but they wanted to use the starter template because they just wanted it to be built out to where they could edit.

And we saw, you know, freelancers using it. We saw hobbyists using it. And essentially, when we went in to look at, what’s the easiest way to implement AI, to make that faster and better for users? How do we make that white screen problem of, I don’t know what to do next, easier for users? And there’s a lot of focus right now on prompt based communication with the AI, where it’s really on, how good are you at writing a prompt to an AI to get stuff back?

And we kind of just wanted to approach, can we take our starter templates product, and just make the content, the images, and the text, be filled with AI data? Images, we’re still a ways out from, because AI generated images are pretty poor at this time, and expensive. What we do use, is we use AI to search Pexels, and then pull in images that are in that same vein for what you’re doing.

So it’s a good starting point, but is it a launch site? No. Like, there’s nothing about our product, or even where we’re trying to go, where we’ll say, you can launch and don’t need to edit. Part of it is because AI is not that good. But there is a sense of, are you 80% there? And that, I think, is really intriguing.

And I think, in our minds, this was step one. Ideally we can get to 90% there. Because I think once AI images gets better, and being able to tell the AI specifically, hey, we’ve got a hero here, there’s this text , we want an image for that background, you know, we will get there eventually. But as far as like our decision making into what are we going to build, it really came out of, what are people using and like, and how do we make that experience better?

[00:09:22] Nathan Wrigley: So you were in the, well, lucky isn’t the right word, but you’re in the enviable position, that you already had the stack of things that really were required. And it was just a case, well, just a case, it was a case of tweaking the things that you already had. So the starter templates, and what have you. And that’s the decision that you made.

It’s curious, because I’ve seen quite a lot of commercial products come out of nowhere, and they’re trying to do exactly the same thing. So in a sense, you had a lot of that stack in place, which must have made it, well, again, I don’t mean to use the word easy, but it must have made it slightly easier for you to build on top of what you already had.

[00:09:57] Ben Ritner: Yeah. It definitely made it, it made the story easy to tell. Because I think that’s part of the issue when you’re talking about AI is, how do you explain this to your users in a way where you can show them how this is going to benefit, and make their lives easier?

And we were also like, we know the problems that come with the current way that we were doing starter templates. Which was essentially, we have a dummy site that actually lives on a server somewhere, and you’re just copying that to your local site.

We had thought about, what if we broke this down to be a section by section thing? Where we don’t actually have a dummy site that lives somewhere. What we have is a library of sections, that can go throughout any kind of pages, and then we’ll live on your site, put those sections together with content, to build out the pages.

So we took the idea of starter templates and just said, let’s rechange the technology behind it, so it’s better and more flexible, and then add in AI generated content and images that can be changed on the fly, and are specific to the user.

We do have some other AI tools built in. What’s interesting is, you know, we started with, can we get starter templates, and then we backed off. But what we need to do is figure out the sections piece. How do we get AI into sections, and then we pulled that back out to like, okay, now we can do pages and then websites.

I think it’s really interesting to think through where we can go with AI. And I think what your original question was getting at is, how do you choose where to go? And we had a path of like, here’s an easy story, and here’s how we can implement it.

But when I think about, where is AI going, what’s the future, what’s it going to look like? AI, to me, is a very good tool at doing iterative work. When you ask an AI to improve something, it generally does a pretty good job. And when I’ve, you know, in my life, and experiencing with using AI tools, the best results usually come from a feedback loop of, it presents you options, and you either choose, I like this one, or you ask for edits.

And we don’t have a tool that does that very well. We have some inline text options, but it’s pretty basic. And what’s interesting for me, just because of using AI for myself, where I feel like our next realm of where we want to go to, is a more iterative thing, because I’ve seen that AI does a better job when we go there.

So part of that decision making, of like, where are we going? It’s based on, I use ChatGPT every day, pretty much. And when I get the best results from it, is when I’ve said, okay, that’s close, but edit this, or change this. And I think that iterative approach, if we can bring that into a UI, can be really interesting when we talk about, how do we build very custom and interesting layouts inside of WordPress?

So our challenge right now is, what does a page wizard look like that’s iterative, that allows you to say, I want this page to be about this, show me some examples? We show you some examples, and then you choose the one that’s like, I’m feeling this, don’t like that section, I would like more on this, and then generate.

And that, to me, could get really interesting, in terms of like getting you to that 90% of like, this page is 90% there, and I’m going to go in and tweak some of the text. But yeah, I think that’s all kind of how we’re choosing where to go is, trying and playing with the AI ourselves.

[00:13:16] Nathan Wrigley: It almost sounds like you are envisaging a future, in which you have a conversation with a page. So you’ve got something in front of you, maybe that was generated by something, using your templates. But maybe it was something that you created yourself. And then the AI is somehow in a conversation where you discuss, and you say, I would like the headline to be shorter.

Okay, right, give me several options, we’ve done that. That image is in the wrong place, can we move that round to the other side? And that picture of a cat, replace it with a picture of a lion or, you get the idea. So you are talking, in effect, to the page. Am I describing your intuitions there?

[00:13:53] Ben Ritner: Yes. Although I am scared of anything that relies on us to prompt the AI. So, I do want to have that talking, but I want to really focus on, rather than doing too much of, you can just raw text in, because that’s that same problem with the white page of like, I don’t know what to ask the AI.

I want to do a lot more of, we’re going to give you options, choose the one you like, and then mini select options inside of that. Or like, where we can guide that to where it’s really just a bunch of clicks, instead of relying on you to communicate well with an AI.

[00:14:27] Nathan Wrigley: The pace of AI is moving so quickly. I really meant it when I said that two years ago, nobody was talking about it. I have no recollection of it being an important subject in the WordPress space. And now it seems to be almost everything that I hear, be it on mainstream media, or in the tech press or in the WordPress space.

Somehow AI manages to get itself in there, because everybody’s so fascinated by it. I just wonder where you get your inspiration from. Now it may be that there are commercial rivals inside of WordPress, you can mention things or not mention things. Or maybe there are things happening in the wider world that, you know, being where you are now in the tech space, and the company that you work for, there’s probably tendrils that you can use via that company.

Or maybe it’s just other things that you’ve seen in the SaaS space, or things that you’ve watched on television, I don’t know. I’m just curious, because I genuinely feel like the whole thing is a bit of a storm, and you’re being blown around, and some things seem to be landing, and other things seem to, we have a go and it fits.

You are telling me that the metrics that you are getting are saying that what you’ve built is working, which is great. But I can imagine that six months from now, something else will have come along, spun the whole thing around, and you are going to have to get, well, not back to the drawing board, but you might well have to reimagine the things that you’ve done. So where do you even begin to look for inspiration in this space?

[00:15:42] Ben Ritner: It’s interesting. I think there is always a fear that someone will come and top you, in such a way that will make you irrelevant. I’m not worried about that personally, but I get it, and I definitely know that it exists. I think there’s a lot more involved in like someone just being able to come and trump you that hard.

We announced Kadence AI last March, and it’s honestly annoying to me that it wasn’t until January of this year that we officially launched it publicly. We had a beta version last summer. So we’ve had our cards on the table for a long time of like, this is what we’re building and this is how it’s going to work.

And I know that there’s other plugins and products that have done other things, I mean, Jetpack itself released an AI. And that does a lot of inline, you know, editing great for bloggers, where you want to generate content. I know that there’s other WordPress AI generated page builders, ZipWP is one. Great product in their approach.

I think that there’s definitely different ways to go about it. And for us, we put our cards on the table last March, we had to get that done. And the trickiest parts were getting the AI to consistently respond, in a format that allowed us to input that into the content that we had. And then trial and error, and things like that really came into play.

Another big piece for us, which was kind of annoying, we had to spend so much time on it, but we have a large user base. And we needed to create a credit system for our licenses, that allowed us to give everyone access to this tool, but also protect ourselves, because every time this communication happens with OpenAI, it costs us money.

That was an annoying, massive overhaul of a product licensing system, that had to have backward compatibility, and be implemented across our product line. And that delayed AI by several months, frustratingly.

You know, as far as like choosing where to go from here, we have a channel at work that’s like AI stuff. Essentially, someone in our org is posting in there every single day with, have you seen this tool? Have you seen this tool? I mean, I’m curious. I’m all in on, I’m watching the videos, what are they doing? I’m really interested. I want to know what people are doing, and how they’re thinking about it.

And then I’m also using ChatGPT everyday, because I want to make sure that I’m really staying engaged with it, and then allowing myself to think, how do I improve upon what we have?

Which I think is a lot different than like, what if you had a blank slate? You had no customer base, and you’re trying to create a product that’s unique. That’s a little bit harder to do versus like, here we have 400,000 active installs with Kadence blocks.

What do these customers need, and how do I give them the tools? And, how does AI benefit them? And to me, AI is a tool, it is not a solution. And I think that that’s really important in my mind, in terms of thinking of, how can I use AI, and how can I bring AI into the Kadence world?

[00:18:35] Nathan Wrigley: Yeah, it’s kind of interesting. Over the last few years, I think it’s fair to say that you, or Kadence, we could use those interchangeably perhaps, have achieved, and I’m doing air quotes, success. You know, three years ago, I suspect the install base was almost nowhere near the number that you just quoted at us. And it’s been a real rise. So congratulations for that.

That was something that I was going to ask you about, whether or not that growth forces you to put the brakes on a little bit? In the sense that, you know, you’ve got this existing user base, you can’t possibly ship something which breaks 400,000 plus websites.

But also, you have to get their buy-in, in a way. And I know that they’re your customers, but they’re also the people that you need. And so if you do something shockingly different all of a sudden, we know how that goes. That typically doesn’t go well. So just an intuition as to whether or not you can’t be quite as nimble as you would like to be, or if there’s the breaks that you’ve got to put on. Do you have to be ultra cautious? Difficult question, and maybe one that you’re not willing to answer, but i’ll throw it out there anyway.

[00:19:33] Ben Ritner: Yeah. I would say that the speed of development has changed, because we have improved our systems. Which, on one hand, slows us down a little bit more, in terms of some of the rapid releases I was doing early on, when it was like me and no one else, or like me and my sister and a support person.

We’ve definitely slowed down from some of that, but we’ve also done a lot less hot fixes, and bug fixes, because we’ve implemented more QA and testing. You know, obviously like we’ve helped that system a lot.

We’ve also done a lot of stuff, and Kadence over the years, that has improved the code quality, and the structure of the code significantly. But backward compatibility is really tricky, and there are lots of times in a developer’s life, man, this would be so much easier if we could just start from scratch. Because so much changes, and Kadence Blocks was built originally before Gutenberg was part of Core. We released it in September, and it became part of Core in December of 20 or 2019, 2018. Gosh, I can’t even remember the year that Gutenberg became part of Core.

But we’ve rewritten product multiple times, to try to keep it up to date, and it’s always a challenge. And that backward compatibility is always a challenge. And just even, I mean, when we originally wrote it, Grid CSS and Flex CSS were not fully supported as like things that you could use.

And so like, just even the structure of our blocks were so different, because we couldn’t rely on some of this new CSS, and that challenge is real. And that desire to just, give me a clean slate from a developer. Every developer will tell you, they would love that iterative, fixing of older code is so hard.

But that’s the hard work. And that has slowed us down, certainly it has slowed us down. But I’m also really proud of that work, because we’ve really fought to do as much as we possibly can to keep backward compatibility, and to continue to improve the code base, so that we’re thinking future. And so yes, you know, we’ve had a lot of success, and things have changed for me and, you know, my day-to-day in a large part.

But what’s causing the slowdown is not necessarily that we’ve been successful, as much as all the other things that come into play, when you talk about a product that’s been around for more than a couple years. And that’s the, we got to do a major revamp on this code base, and we’ve got to prepare all these blocks for what we’re doing in Gutenberg today versus, you know, four years ago, that kind of stuff.

[00:22:00] Nathan Wrigley: Prior to the whole advent of AI and all of that, did you have a roadmap which looked really different than what it is now? Because I don’t build software, but I can imagine that if I did build software, I’d want to be staring into the future as far as I could go. I would want that roadmap to be, you know, agile, but at least have some North Star to be heading towards.

And I’m just wondering, if AI kind of upended that, it was more of a case of, oh okay, there’s this, right. Throw all that away. Did any of that happen? Did you have to sort of retool the team, or take new people on, or just, like I say, throw away the old roadmap and begin again?

[00:22:33] Ben Ritner: Roadmap for me, or generally like Kadence, is usually not too much more than six months out. Mostly because we learn a lot from our users. The users determine the roadmap more than any other factor, and we’re continuously updating what we’re going to be working on in six months, because things change and WordPress changes. So it’s not as long term as you might think. Like, I don’t know what Kadence will be developing next January. I have some ideas, certainly. Some back of the mind thoughts, but nothing that we have committed to or decided.

And so the main shift that happened for us is ChatGPT came out, and within a month we were like, we’re not building starter templates anymore. And that was the big shift. We were kicking out starter templates, at least two a month, of these, you know, niche sites. And we just decided then to pull the plug.

And we were like, the future is going to be AI generated starter templates. We have to solve that problem. Let’s stop spending any resources that we can, on the old way of, it’s old and we just had released them, right? But like that old way of thinking. And that was, what did AI do in terms of shifting our mindset? It was like, we’ve got to put everything into getting AI starter templates out the door.

Thankfully, because I’m part of Stellar, we didn’t have to go to the Kadence dev team, and I’m one of them, right? And say, you guys who are working on all this stuff that we were doing, we were working on some major dynamic content stuff, some advanced query block stuff, big rewrite of Kadence blocks. We didn’t have to go to that team and say, everybody stop what you’re doing, we’re doing AI.

We were able to pull resources from our parent company, and that was the team. We formed a whole new team to say, let’s see if we can solve the AI problem. Which I was a part of, but I wasn’t the entirety of that team. And so I spent the last year, half on Kadence and half on AI. And that’s probably not accurate time, but like basically I’ve been on two teams. Then I got our Kadence team involved, as we integrated the AI stuff we had done into Kadence.

But how we were able to do that, is we were able to think, okay, we have a use case for how Kadence is going to use AI. But from a Stellar level, all of our products could benefit. LearnDash, for example, has already implemented AI into it. You can get a course outline done using AI.

And so we went and developed backend tools, and a server that does all the proxy communication with OpenAI, and all the caching, and all of that licensing piece. That was work that a different team, not the Kadence dev team did.

So it really helped in us being able to keep releasing things with Kadence theme, and Kadence blocks all of last year, while there was another team working on this whole AI piece. And then towards the end of last year, we spent that time integrating it really into, well, I mean we’ve been doing it the whole time. We had a beta version out last summer.

But, you know, that work then bringing in more of the Kadence dev team, and support team, and all of that to kind of get it out the door. That came at the end where it’s like, when we announced, hey, we’re going to do AI, there was a lot of fear that, well, what about your roadmap? What about you guys developing this or that?

Like that was the feedback we were getting, is everyone was very concerned, hey, AI seems like it could be a fad. And we were like, no, hold on, we’re still doing that. We were just fortunate enough to be able to have resources from outside of our brand, come in and work on a lot of the AI stuff. And so that helped a lot with the roadmap. But obviously the starter templates piece was a shift.

[00:26:07] Nathan Wrigley: It’s really nice that in the background you’ve got a bigger company, and the AI piece, a lot of the technologies, there’ll be a lot of commonality there. And so if you build that backend, like you say, the tooling that you need for LearnDash, and all of the different bits and pieces that make up the StellarWP Suite. And you can all share in the benefits of all of that rather than, I don’t know, five or six different companies all trying to achieve the same thing. So the economies of scale work really well there.

Pretty brave decision though, I’ve got to say, because I think, if you were to go back four or five years, everybody was, do you remember NFTs and Crypto and all that? There was that same euphoria, and it was going to be the future. And I think there’s quite a lot of people who down tools, and just moved in that direction. And fast forward to today, that now looks like it was a mistake. I think we can probably consign that to the dustbin of history. Perhaps I’m wrong about that, but it feels like that’s the way it’s gone.

And intuitions a couple of years ago about, let’s throw everything at AI, that could have gone the same way. It didn’t. I don’t think it did. It really doesn’t look like it’s going to go that way. It looks like it’s going to continue into, you know, for the next decade or however. Again, I could be wrong about that. But it’s a brave decision for you, the company to make. And it looks like you’re on the right side of history, so it’s a sigh of relief.

[00:27:22] Ben Ritner: Yeah. I mean, I think the fact that NFTs, so many people were like, what is this, and how does it work, and all of that? It’s part of the reason why you could get on board with AI a lot easier, because everybody could open up ChatGPT and go, oh wow, this isn’t the annoying Verizon Wireless, or whatever, large company chat bot that I get at support.

This is very different and improved. And I think that’s part of why it wasn’t easy to be like, this could be implemented and used, and be very useful. And I would say we did the bare minimum of, we’re doing text replacement on layouts. The future of having an AI do more creative, and I’m using air quotes when I say creative, work with you, is going to be really interesting.

Because I think that it’s a phenomenal tool to work with you. Like, as far as you being able to guide it, influence it. I think it’s going to take a lot of the time consuming, exhausting work of making this all look exactly right out, which is going to be great.

[00:28:31] Nathan Wrigley: I am going to get into the nuts and the bolts of the actual stuff that you’ve built over at Kadence. I dropped a URL at the beginning. It’ll be in the show notes. If you go over to there, you’re going to find a video right at the top of the page. Caveat emptor. I don’t know if that will be there in the days, weeks, or months to come.

But it’s a video. It’s a few minutes long. I think it’s you, Ben, it sounds like you. And you are running through the process of basically building a WordPress website from scratch. And I’m going to paraphrase it, and if I miss a step out here, forgive me, but I’ll just, for the sake of brevity, I’ll just paraphrase it.

Basically, you begin this wizard, and you enter some basic details. Details about your company. You describe the business, it spits back at you some possible keywords, and you select those. Once you’ve done that, you do what you often do with AI, you pick a tone, you know, this sort of feeling of how you wish the website to be. And then you go into the image selection process.

And then all of that gets consumed by the AI. It has a think for, I don’t know if that’s the right word. It does what it does. And then it brings back some layouts for you. And you’ve got this selection of layouts, and your intuition says, that’s the one. So you pick that. You then start adjusting colors, and it disallows, or at least it prompts you that that’s not going to be a good selection, in terms of accessibility, which is nice, thank you for that.

Then you mess around with the fonts a little bit, and you import some pages that you wish, that you think would be appropriate. So you’re nearly there. You, at that point, launch the site, if you wish. All the pages get imported, and your site is ready to be launched.

And then you’re into the process of tweaking it, and making it your own. I think anybody using AI would advise you to do. Edit with the block editor, go in, use the AI to amend things if you wish, or just do it yourself. And then you can begin the process of saying, actually, this row is just way off message. Let’s just swap that one out, and just pick a different one. And again, the AI can step in and what have you.

So that took me longer to say than it takes to do, I suspect. That is pretty incredible. And it’s leveraging all the different bits and pieces that you mentioned earlier.

Is this in any way bound to the products that you release? So I know that you’ve got a theme. I know that you’ve got a block suite, and there’s a pro version of that block suite.

Having played with it myself, I can see that you’ve definitely used the blocks that are at your disposal. But, how does that work if I don’t have access to your pro suite? Would I even be able to use this? Is this available? Is there any kind of free plan? Can I just pay for the AI bit? And what if I decide that I want to back away from it and, I don’t know, amend things? Is there any lock in, is essentially what I’m saying?

[00:30:59] Ben Ritner: Certainly. There is a free tool. Kadence Blocks is free, and Kadence Theme is free, and Kadence Starter Templates is free. You can get a free account to use AI. You’ll get 250 credits, which is more than enough to build a website. You could build a couple with that many credits. And so if you’re using free, you will have slightly limited choices into which layout you can choose.

So when you get to the point where you’re shown layouts, we’re going to give you, let’s call it six. Two of those will be pro only. And that means that when we design those layouts, they’re actually using pro blocks that we have. And so if you imported them, it wouldn’t work, because you don’t have pro, so you can’t access those.

If you use one of the free layouts, which I’ve already looked at one of our, you know, our more popular one is a free one right now. You can do the whole thing for free. And you don’t have any cost to you. You can go and try it out. We have a way for you to try it out in InstaWP as well. And that’s free. So if you just want to like play around, that is on our site.

As far as content lock, and this goes into a really interesting conversation of, how avoidable is content lock in WordPress, and in websites in general? And, how do we get away from that? Because the blocks are part of the core editor, if you delete Kadence blocks from your website, the HTML markup will still exist on your website, because it’s just block markup.

But the rendering of those blocks won’t function, because they depend on Kadence blocks to be rendered. And that’s how all custom blocks work in WordPress, is that if you have a custom block, you need that code to essentially handle the rendering.

There is work, and we’re scoping some of this out. There is work, and Matt Mullenweg talked about this State of the Word, to do more data liberation. Where you could say, I’m using Kadence Blocks, but I want to convert this to Core Blocks, or I’m using Spectra Blocks, and I want this to convert to Kadence Blocks, or whatever.

The issue with that, is that there is almost never a one for one with any of these tools. Elementor to Kadence, Elementor to Core. All of these kinds of ideas of being able to migrate out of one system into another is fine, if what you’re trying to get is the text and images. Not fine, if what you’re trying to get is a one for one of this to that, because Kadence Blocks wouldn’t exist if we didn’t massively expound upon what’s in core.

No one would use it, if it didn’t take it a whole lot further. And so to convert these blocks to core blocks, you will lose functionality. You will lose the ability to have the accordions, and the tabs, and the sliders, and the certain types of image overlays and animations, and all of that stuff.

So, is there content lock? Yes. There’s content lock with core blocks as well. If you want to take core blocks to the classic editor, that doesn’t really work. So it all depends on what your definition of content lock is, and whether or not you feel like this is the system that’s going to be around for a while, and supported.

And so I think, when you look at Kadence, we’re backed by a much larger company. There’s a lot of longevity in using Kadence Blocks, in terms of being able to render out those blocks for you, and the focus on backward compatibility.

Do I think data liberation is really interesting? Absolutely. I think where I think about data liberation is, how do we get, and this is what AI, I think, could be really good at. How do we get out of the mindset of, can I move from HTML to WordPress, or Wix to WordPress, or Core Blocks to Kadence Blocks, or Elementor to Beaver Builder, or whatever it is that you want to move?

How do we stop thinking about tools and go, you have a website, it’s HTML and JavaScript? Can we train an AI to convert that to some kind of standardised information? Where it basically says, here’s the JSON markup of this site. This is the type of content it is. Everyone builds importers for that type of thing, so you could say, I want this site in Kadence Blocks. And we use AI to convert that site into some kind of standardised markup that says, here’s all the content, the type of content, the context for how that content is. And then Kadence Blocks is an importer for that, or WordPress Core has an importer for that.

That wouldn’t still be a one for one, because, again, there’s just no way you’re going to get a one for one, unless you convert straight to HTML. You could move it, but you wouldn’t actually be able to use it down the road. But what could be really interesting is, if you could then import that markup into whatever tool of your choice.

Now, it might not be a one for one in design, but in terms of actually usable, and now those blocks will render, and are editable, and you can tweak the styling. And you’re that much closer. That to me is really, really interesting.

That’s part of the exploration for us is, can we create a system that exports? You know, where we think about grabbing the front end content, getting that into some kind of standardised system, importing that, and then exporting that in and out of that standardised system. I think that there’s a lot of really interesting stuff, in terms of just being able to process that and say, could this be a thing that we work toward?

It would take a lot of resources and energy, but would solve all of the weird platform specific stuff. Today what you need is a Wix to WordPress thing. Okay, like I totally get that. But I think down the road as AI gets a lot better at scraping sites, and understanding context, and being able to create, say, hey, this is an area that has media and text, that are related to each other. And we could conceive of a way of like, here’s all of the way that AI’s going to convert that content into a standardised format. And that could be really interesting for data liberation of your website.

[00:36:49] Nathan Wrigley: That’s absolutely fascinating. The data liberation thing, really is about CMS to CMS at the moment, isn’t it? You know, it’s Shopify to WooCommerce and it’s, I don’t know, Joomla to WordPress, or whatever it may be. But that more granular piece of, could I come from, let’s say Joomla, through WordPress, and the tools that I’m using inside of WordPress, that’s Kadence. So, can we go that extra step? Not just import it into Core Blocks, but can we then go a little bit further, because I’ve invested in Kadence, I know what I’m doing over there. Let’s give that a go.

But yeah, that’s really fascinating. And obviously, in everybody’s interest, if there’s some standardised way of doing that, and everybody could jump on board, so even your commercial rivals. You could access it, then the import isn’t the problem, it’s the quality of the product that is the reason why people would wish to import it. Yeah, that’s a really interesting idea, and not one that I thought of, the data liberation thing. Yeah, we’ll see how that goes in 2024.

In terms of the AI, the way that I deployed it was into a vanilla site, to test out what you’d got. Also I noticed, I could be wrong about this, but on the video that you posted on the website, it felt like that was a vanilla site. How does the AI work if, because most people are not starting from scratch, are they? A lot of the audience listening to this, will wish to try things out on sites that they’ve already got, and push the boundaries of pages that they’ve already got.

So I know that the blocks themselves, they can be adapted. You’ve got a bit in the, you know, the bar that appears when you interact with Gutenberg. There’s something in there which you can interact with the AI and prompt it, and what have you. But does the whole wizard that we described earlier, is there any utility in that, if you’ve already got a site? Or, is it just for beginning blank slate stuff?

[00:38:22] Ben Ritner: So that Wizard, you don’t need to generate a website, although you can, and you can import that into an existing site, if you want to. That’s totally fine. But Gutenberg, we have what’s called a design library. And if you have Kadence Blocks installed, you’ll see the design library tab in the backend, like in a page editing thing at the top bar.

And if you open that up, we can generate contextual stuff, using that same information from the wizard. So if you are building a page, for example, and you’re like, I need a call to action section, that’s a context that we’ve created. You can go in and say, I want all of your sections with call to action context.

So the AI generated text that is for your website, about call to action. And then you can pick from the layouts. And there’s hundreds, we have over 600 different patterns in that design library. And all of those can be empowered by AI, and many of them can be used multiple times. So you might see a pattern and say, well, that’s a great pattern, that I like.

I can put about context into it, or I can put value proposition context into it, or call to action context into it. So you can choose the context, and then you can choose the pattern type, and then you can drop that pattern into your website. So if you have an existing website and you’re like, I even have an existing page, and I just want a section, AI can help you there. Or if you’re building a new page, and you want to piece together patterns, there’s a lot of different ways that you can use that tool.

[00:39:48] Nathan Wrigley: So you can go back through and kind of retrofit, and just swap things out. And presumably, rows and patterns kind of used interchangeably there. You can just save those, put them somewhere else for later keeping. But the idea is that you can put these different contexts in, so that you can modify the things that you’ve got. Okay, that’s really interesting.

What’s the stack that’s in the background? I mean, I’m guessing it’s OpenAI, because everybody seems to be using OpenAI. Are you keeping an open mind about OpenAI? Now there’s a phrase I should coin. Because it does seem like hot on the heels of OpenAI, you know, they really were first to market with this interactive chat like capability. Which is very beguiling, but it does seem like hot on the heels are Anthropic and Google, with all of their offerings. Are you wedded to OpenAI, well, first of all, is it OpenAI?

[00:40:32] Ben Ritner: It is open AI, yes.

[00:40:33] Nathan Wrigley: And second of all, are you wedded to that, or do you keep exploring the different models? Because it does feel, in certain contexts, that OpenAI might be feeling the pressure, let’s say. I’ve tried a few different ones out, and in a variety of different contexts, OpenAI excels. In a few, very niche contexts, there are some rivals, which absolutely, you know, they just seem to do a much better job. So anyway, I’ll just hand that one over.

[00:40:57] Ben Ritner: Yeah, for sure. I think we are very much open-minded, in terms of what we do. Part of us building this whole proxy server to handle, and this whole prompt database, and the way that we communicate with OpenAI. The whole time we were thinking, this structure, all that we’re doing, if we need to switch to some other system. Because OpenAI changes their prices, or they discontinue doing what we’re using the for, or for whatever reason, we need to make sure it’s movable.

So we have experimented with Google’s, and right now we’re using OpenAI. But in a year from now, we might not be. And I think that was a key part for us of, like we need to make sure that we have the ability to shift and move, if we’re seeing a new company come up that is significantly better.

[00:41:42] Nathan Wrigley: Yeah, you never know. Maybe there’s somebody who’s developing something which is really adjacent to the web design industry, and an AI which is specifically for that.

So what you’ve built, this conduit between the website and the AI. You’ve built this middleware, which sits in between. And that middleware could just be repurposed at a moment’s notice, to talk to the, I don’t know, the Gemini or Anthropic API. And so, really, you wouldn’t have much disruption to service. Ah, that’s interesting.

[00:42:08] Ben Ritner: We’ve got that server in two different places, so we can experiment too. So we can experiment by just locally pinging saying, hey, use the backup server, and then, let’s experiment with that. So we have a lot of ability to do things, to test things, without disrupting our user base.

[00:42:24] Nathan Wrigley: You mentioned that there was some sort of credit system, whereby you acquire credits, which then you spend, in order to do the transactions with the AI. How does that work? Is that like a monthly thing that you add on top? Is that a subscription service on top, or do you get a certain package when you purchase it? Just run us through how that works.

And you mentioned that 250 credits was available on the free version. You also mentioned that that would probably get you through the process of building a couple of websites. So yeah, just talk us through the typical spend on doing things, and how the credit system works, and how do you top things up if you’re an agency, and a blowing through your credits.

[00:42:59] Ben Ritner: Yeah, so we aren’t charging anything extra for the credits. They come with whatever you have. So if you’ve purchased from us, you get credits automatically into your account. And if you have an existing purchase, and if you go and purchase today, our pro products have a certain amount of credit. So Kadence Blocks Pro alone is 1200. If you go up to our essential bundle, which is Kadence Blocks Pro and Theme Pro and Starter Templates Pro, that is going to give you 4,000 credits, and then up from that is 8,000 credits.

We largely created the credits as a deterrent to prevent abuse, not as a system to sell you something more. And our goal is that you don’t need a separate subscription. And we want to make it so easy to use, or you’re not really worried too much about your credit count. To that point we’re talking about single lifetime credit bundle packages, that you’ll be able to buy for that agency, that’s burning through their credits.

We’re going tomake that very, very cost effective. Because, again, we didn’t ever consider that credits could ever be a revenue driver. We have no plan for it to be a revenue driver. We literally just want to protect ourself on cost. So going forward, there’ll be some way to buy credits very cheap.

The whole communication and stuff with OpenAI is relatively cheap. That’s part of the reason why it’s so usable, so many people are building on it, is because being able to communicate with the AI is cheap. When you take it out to a scale of 400,000 active installs, and things like that, you start to go, oh wow, this could get expensive pretty quick.

So that’s where, you know, we need to limit some of that. But in general, the idea is that we give you more than enough credits. And if you are an agency, or you’re burning through a lot of credits, we’re going to make it very cheap for you to buy more. Current plan right now is a $20 package that’s going to have like 4,000 extra credits. So we’re going to make it really simple for you to just say like, oh, top me up, I need more credits.

[00:44:51] Nathan Wrigley: Let’s just, before we wind it up, let’s close the lid on Kadence itself. And let’s just talk about your intuitions, Ben, about AI in general. Now, I’m not trying to force you to say anything here or to, you know, imply that you’ve got a certain opinion. But I’m curious as to what you think.

So there’s two phrases that I’m just going to raise, and one is that, you know, a rising tide carries all boats. In other words, we throw AI into the mix, and everybody’s experience of life, and website building, and all of that, just becomes better. And then there’s another phrase which is, just a race to the bottom, which is entirely the opposite, you know.

And I’m just wondering if you have any intuitions about that. Because, certainly amongst the people that I’ve met, who are jobing website developers, you know, they’re using tools, much like you’ve got, and rivals, to build websites.

There is this sort of impending sense of fear amongst some of them, that the very tools which seem so beguiling at the moment, and seem so brilliant because they save a bit of time here, you know, save a minute there, and an hour there and what have you. That all is great, except that there’s this fear that maybe these tools are going to hollow out the very industry that we’re serving.

To the point where, well, in a few years time, you’ll just go to some webpage and tell it, I want a website to market my cat food, off you go. And, you know, three minutes later you’re done, and it’s all finished. So I wonder if you’ve been thinking about that, if you’ve got any guardrails in place, or intuitions there. I know that’s a fairly ephemeral question, it’s hard to get your hands on. But just a few minutes on your thoughts around that.

[00:46:15] Ben Ritner: Sure. I think, if you are a freelancer who’s building brochure websites, and not offering anything else, other than, here’s your website, that would be a concern. I think that’s a valid concern of like, will these customers keep coming to me, for this very simple brochure website, if that’s all I’m offering?

I think that that’s been going away for a while, and that freelancers, if you’re paying attention, you’re probably already looking at, how can I help people have an effective website, not just a website?

And I think that there was a time where everyone was like, I just need a website. There’s just so many local businesses that were like, I just need a website. I think we’re coming to the end of that time, and most businesses are not just saying, I need a brochure website. They’re saying, I need a website that helps me do business, even these local companies. That is able to capture on SEO, that’s able to adequately supply a convincing marketing message for someone to choose them over their competition. That represents them well online, and that allows them to be effective, and continue to serve their local communities, or broader.

So to me, I think if you’re a freelancer, you should be really thinking about and learning about marketing. And really thinking and learning about SEO. Because what you want to be able to offer is, yeah, this might look like a brochure website, but let me show you how it’s actually effective. Let me show you how we positioned your marketing messaging to make it effective.

And we are running the ability to do social media ads, and everything else. Like, there’s just so much more that most businesses need, most anyone who’s going to be online. I feel like, if all you’re offering is a brochure site, and done and move on, that is at risk. Because I do think that AI can create a very basic brochure site, that someone will be able to say, that’s good enough.

But I don’t think that that’s what the majority of, especially WordPress freelancers, are really trying to pitch someone on. And I think that you just need to make sure that you are continuing to grow yourself. To say, I understand the web, the intricacies of marketing, and SEO. And I can help you, not just to make a good looking website that has your content, but make it so it’s actually something that people will be able to find, and that’s going to actually grow your business, or improve your business.

And I don’t think AI is going to be able to solve that. I think tools will help you solve for that. But I don’t think AI, on its own, or anything automated like that, is just going to be able to solve for that.

[00:48:49] Nathan Wrigley: It’s interesting. I think the only thing that I know that AI will do in the future, is be an agent of change. Whatever that change is, it’s more or less a question of just putting my finger in the air and guessing, I’m not really sure.

But it’s fascinating to get your perspective on it, because a lot of the people that I talk to are those freelancers, they’re the people building websites. And interesting to get a different perspective. A company that’s building the tools to enable those freelancers to do the work, and all of the concerns, and thought that you are putting into building out those tools. Really fascinating conversation, I enjoyed that tremendously.

Just before we call it a day, Ben, I’m wondering if you wouldn’t mind just sharing some of the places, or place, where people could get in touch with you, if they wish to. I will drop the URL, like I said, of the Kadence website. So if there’s anything aside from that, just let us know.

[00:49:36] Ben Ritner: I’m not super active on social media, but best place would be our Kadence Facebook group. That is the one place on social media I will come in and out of. I’m not a huge social media person, but that’s one that I, that community is super fun, and I love that.

You will find a lot of very engaged people in there. So you can go in there and ping me in there. And that’s probably your best, outside of, obviously you can come to the Kadence website, and just click on our support.

[00:50:02] Nathan Wrigley: Great. Thank you so much, Ben. It was really a pleasure chatting to you today, all about Kadence, but also about AI. Really appreciate it. Thanks so much.

On the podcast today we have Ben Ritner.

Ben is the founder of Kadence WP, a company that specialises in developing themes and plugins for WordPress, with a focus on enhancing the front end of websites. The team have developed a suite of blocks that enable users to do more with the core Gutenberg editor, and in 2021 Kadence was acquired and became part of the StellarWP umbrella brand, where Ben now serves as part of the senior leadership team. Under StellarWP, there’s a suite of popular plugins, including LearnDash, the Events Calendar, and GiveWP.

In this episode, Ben provides us with an insight into how the Kadence team managed the integration of AI into their products, shedding light on the impact it has had on their website building process.

We discuss the challenges of migrating content between systems, as well as delving into the role of AI in content creation and design adaptation.

Ben talks us through what guided their technical innovations to accommodate AI-generated content and images. Ben’s perspective is pretty unique, working for a company he started, right through to today where they have an install base of over 400,000 websites, all of which could potentially make use of AI, but not at the risk of breaking backwards compatibility with their own products.

We chat about the importance of integrating marketing and SEO knowledge for effective website development, and how the recent demand for AI might cause freelancers and agencies to think again about the products and services they offer.

We also get into the way that the team decides what trends in AI to follow; AI is still new, and it’s possible that what’s all the rage today might be superseded at any time. Where does Ben find inspiration for which AI platform to use, and which implementations of AI are dead ends for website builders?

If you’re interested in the impact of AI in web development and are intrigued by its potential, this episode is for you.

Useful links

Next level site creation with AI






Data Liberation

Kadence Facebook Group