Author: Gerardo Medina

[00:00:19] Nathan Wrigley: Welcome to the Jukebox Podcast from WP Tavern. My name is Nathan Wrigley.

Jukebox is a podcast which is dedicated to all things WordPress. The people, the events, the plugins, the blocks, the themes, and in this case, exploring AI’s impact in WordPress agencies.

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 wptavern.com/feed/podcast, and you can copy that URL into most podcast 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 wptavern.com/contact/jukebox and use the form there.

So on the podcast today we have Matt Schwartz. Matt runs Inspry, an Atlanta WordPress and Woo Commerce agency. He started it back in 2011 and has been working with WordPress even longer than that. In addition to his agency work, he also has a product called CheckView focused on WordPress testing. He’s got years of experience in the WordPress agency world, and recently he’s turned much of his attention towards the growing impact of AI.

If you’ve been hearing a lot about AI but a feeling fatigued by all the fragmented conversations, this episode might well offer a different perspective. Rather than focusing on how AI creates websites or content, Matt shares a different angle, how AI can be used inside a WordPress agency to enhance processes, improve workflows, and deliver more value to clients, with much of it happening behind the scenes.

We start by talking about how Matt stumbled into web design and how that led him to running his own agency. We dig into agency life, and why so many freelancers and agency owners are constantly iterating on their processes. From there, we talk about the big shift that’s happening, not just in building sites, but in how agencies can use AI to streamline their SOPs, client communication, and internal operations.

Matt explains the need for intention when adding AI to an agency. He introduces the idea of an AI vision document, that helps set guardrails and guidelines for where, and how, AI should factor into your business. He also shares real examples of ways AI can save time and stress in things like meetings, proposals, debugging, support, and even helping you expand your service offerings. We also touch on the risks, ethical considerations, and the importance of keeping a human in the loop during critical agency moments.

If you’re running a WordPress agency, or are curious about how agencies are adapting to the rapid pace of change, brought by AI, this episode is for you. This is part one in a two-part series, so listen to this and tune in next week for part two.

If you’re interested in finding out more, you can find all of the links in the show notes by heading to wptavern.com/podcast, where you’ll find all the other episodes as well.

And so without further delay, I bring you Matt Schwartz.

I am joined on the podcast by Matt Schwartz. Hello, Matt.

[00:03:45] Matt Schwartz: Hey Nathan. Thank you so much for having me today. I’m excited.

[00:03:48] Nathan Wrigley: Yeah, you’re very welcome. We’re on the podcast today to have a chat about AI. Now, before you hit the stop button, dear listener, because AI is all the rage everywhere, we’ve talked about it a million different ways. I think there’s something a little bit different about the conversation that we’re going to have today, because it particularly plays into the WordPress agency, kind of the stuff that you are not doing with the website directly, but all of the bits and pieces that allow you to have an agency, and how AI may or may not be best placed to insert itself in those different scenarios.

But before we begin that, Matt, do you mind just giving us your little bio? Maybe tell us a bit about your situation regarding WordPress agencies and whatnot.

[00:04:31] Matt Schwartz: Definitely. Yeah, so I run an agency called In Inspry in Atlanta. We’ve been around since 2011. We’ve been using WordPress since 2013, and also have a product called CheckView, which does WordPress testing.

But yeah, in the agency space specifically, you know, I’ve been talking to a lot of different agencies about AI. I’ve been pretty involved in it. And you’re totally right, Nathan, our goal today is not to make everyone just have to experience the verbal throw up of the word AI, AI, AI over and over again, which is, I feel like I’m sick of the word. But really going into how agencies can use it in, I think, really interesting ways, and also being candid about what AI is, and some of the pitfalls I think of it that, you know, aren’t always talked about, especially if you go on LinkedIn.

[00:05:15] Nathan Wrigley: Okay, so we’ll get into that in a moment, but just before we do, there’s a couple of interesting bits that I want to throw at you. And this is something that I heard in the British press not that long ago. And it doesn’t in any way, shape or form reflect on WordPress, it was just more generally about AI, and the fatigue that the general population are experiencing around that term.

And it feels like we have reached maximum capacity to just hear those words, and hear the overpromising and the potentially under delivery of AI. So I’ll throw that little bit in, but also, just to say that what we’re going to talk about today is not going to be how to get the pixels on the page, and how to use AI to turn the website out. This is much more going to be the background to the agency that you run and all of that kind of thing.

So before we begin, did you intentionally get into web design all those years ago, or were you more like just about everybody that I talked to, did you stumble into it a little bit more?

[00:06:13] Matt Schwartz: So I stumbled into it in the sense that I started when I was basically a kid. You know, I was like obsessed with building websites for like clubs, and middle school, you know, we had tables and HTML. I think Template Monster was around then. And I would just go to the website and look at these beautiful designs that I knew I couldn’t make.

So then, from there I built websites all through middle school, high school. Got paid, I think from my first one it was my Mom’s work. She worked at a dentist. It was awesome that he let me do that. And, you know, he paid me a couple thousand bucks, which was a lot in high school. And then from there I just built sites through college. We were in Drupal land over at University of Georgia. So that was a little harsh reality for the first CMS I ever used actually.

But I really just enjoyed building websites through that process. And I remember graduating in Information Systems in the Business School and being like, I think I’m just going to keep building websites. I think I like doing this. So I didn’t go the consultant route or anything like that, I just stuck with websites. So I stumbled into it when I was a kid, but I definitely chose to stay in it after that.

[00:07:17] Nathan Wrigley: Yeah. And what’s curious about that, and it maps very much what I did, almost every word that you said could map into my own life. Is that you, not working for a company, you are never sort of given the SOP. You have to do the SOP. You have to figure it out as you progress on your journey over the years. So every process that you’ve got, every thing that you do, every price point that you make, every email that you create as a template, you’re probably generating that yourself.

And so that kind of leans heavily into what we’re going to do today, because I felt that journey never ended. Part of being an agency owner was always this constant exploration of not the website itself, that kind of handled itself, more the, what’s the process? How do I get new clients? What are the backend systems that I’m going to use to make it all work?

And so I think freelancers in particular in the WordPress space have got that. And so they’re probably constantly looking around, very much beguiled in the more recent past by what AI can do to them.

And so let’s start, you’ve listed out very kindly a whole load of show notes for me. And the first point that you wanted to get into was, well, the big shift. So let’s start there.

[00:08:26] Matt Schwartz: Yeah, definitely. So I think one thing that we’ve seen as agency owners is, oh, websites and content now can start to be built by AI. And everyone’s talked about that, like you said. But I think what is more interesting is what you’re bringing up, which is around more the process of using AI. Which, if you are a freelancer and you have not looked at your process, please do. I didn’t look at my process for like years, and I would just repeat the same thing over and over again. It wasn’t until I actually started hiring people that I realised that was even really a thing. I know that’s sad, but that’s the reality. So if you haven’t, definitely look at that.

But when it comes to AI, I think being able to use it for process and your SOPs and automation, that’s really where I think it’s actually going to make the biggest impact for agencies that do want to use AI.

Because essentially, not every agency’s this way, this is a generalisation, but as a customer, or a client of an agency, they don’t see the difference between one website and another typically outside of the design, right? They don’t really know the technical know-how. But what they do see is, what is your workflow? What is your process? What is your touch points with them? And that’s ultimately what ends up being the product to your clients.

So I think as an agency owner, being able to use AI to make that process easier, and more clear, to your clients is what will really allow you to thrive. Not necessarily, just the content and executing building the website. Sure, AI may be able to help there, but that actually goes into the bigger process in my opinion.

[00:10:01] Nathan Wrigley: Okay, so we’ll definitely get into all of those, but I think basically the case you are making is that there is a there, there. There is something behind the AI that could definitely improve things.

I think it’s fairly unlikely that anybody listening to this podcast hasn’t dabbled in some way with a little bit of AI, but maybe there’s a handful of people out there who genuinely haven’t. And the last 20 years have been marked by fairly gradual improvements in things. You know, SaaS apps came along and they gradually improved and one superseded another. But again, it was incremental.

But over the last three or four years, I think that’s all gone out the window. Incremental’s no longer really a word. It’s seismic this week, seismic next week, seismic the week after that. Keeping up is going to be difficult. But anyway, needless to say, you are going to make the case that there are areas where AI smuggled into your business is going to be useful.

Can I just ask at the beginning, do you in any way show the AI to your clients? In other words, is there a moment where they get to see behind the curtain, oh, Matt, look, he did that with AI, or do you kind of have this curtain which protects you from the client, so that they never see that you are using AI? It’s a bit like how everybody who was a freelancer always uses the word team. They sort of pretend that there’s like nine of you, but there’s actually only one of you. So it’s a bit like that. Do you hide the AI from your clients or do you let them know that this is what you’re doing?

[00:11:30] Matt Schwartz: So when it comes to the product, we definitely let clients know if we are using it in their product. Because I think, at least from my ethical standpoint, I think you should do that. I don’t want to be in a case where we’re not doing that. But I do think when it comes to your process and internal workflows, no, we don’t typically need to do those things.

The only time we would do that is if we’re actually working with a client to improve their internal processes with AI. Then they may be seeing a parallel setup to what we’ve done, even at our own agency.

[00:11:58] Nathan Wrigley: Yeah. There are some people who kind of revel in the, no AI, if you like, so they make that a badge of honour within their business, whether it’s an agency in the WordPress space or anything else. And so obviously they would probably want to proclaim from the rooftops that they’re not using any AI. But I think yours is a fairly standard position. You know, if it doesn’t actually affect what they’re doing day to day, why would you need to use that? In the same way that you don’t need to tell your client, well, we’re using Salesforce in order to communicate with you. It’s just, there’s the URL, go to that and type your ticket in there and so on.

So your second point, why now? Why is it important right at this moment? So we’re recording this, I don’t know, towards the end of April, let’s say that, 2026. Is this like some sort of red line in the sand? Are we about to enter a Rubicon moment where we can’t go backwards?

[00:12:47] Matt Schwartz: Well, I don’t know. It’s seismic every week as you said. So I do think the gap is widening between agencies that are not using AI and using AI. But that doesn’t necessarily mean, in my opinion, you should just like hop on the AI train if you’re not currently deep in it. You do have to think about what makes sense to your agency and what you’re comfortable with.

But I think it really comes back to the fact that execution is becoming a commodity more and more, at least in the web agency space. If you’re building a brochure site, right, those tools are essentially becoming more and more replaced. Just like drag and drop builders came in and now this is kind of, in my opinion, the next iteration. It’ll be less about the execution of building a simple website. It’ll be more about, what is the true value of your agency to that client?

Which in a sense is not a bad thing, because this was always an argument before. You know, are you an agency that builds solutions for clients? It makes them money, or saves them money. Or are you an agency that just executes what they say? And there’s definitely a place for that. I think there will always be a place for that, but I think when you look at like a brochure site, it’s harder, I think, to make that argument than if it’s like an e-commerce site or a custom app, because the tools are just getting better.

So as an agency, I think there is an edge here with AI because clients are going to have higher expectations. You’re going to compete against companies that are using AI to do better touchpoint, to do more touchpoints, to having a better process.

Now, of course, that is dependent if they implement AI correctly, right? User error and AI is like any technology, that is definitely a major concern, guardrails, all that good stuff. But I think that is why this is the time, because if you’re not already looking at it, your competitors are definitely looking at using it in some capacity.

[00:14:40] Nathan Wrigley: Just something you said really struck home there. You said execution is a commodity. I’ve never heard that phrase, but that encapsulates so much, so well. I think that’s really interesting.

And I also share your moment in time analogy because I think we are at some moment where the seesaw, I don’t know if you use that word where you come from. The seesaw is definitely tipping to the point where, in the part of the world where I live, virtually everybody is aware of it. We mentioned that maybe there’s fatigue about it, but certainly almost everybody has had some exposure to it. They’re now aligned with what can be done, and at what cost, and for what amount of time.

And so it does feel like if you were to go and say, I don’t know, I’m going to build you a $5,000 brochure website with two pages, maybe a few years ago that was much more credible than it seems like now. And so this horizon of expectations is opening up. And it’s not just because we can do it, it’s because the clients, they know we can do it. And they know that things are going to be cheaper to produce en masse in the future.

So I think you’re probably right. So again, you’ve made the case for, this is the time. So not just that this is a good idea, but this is the time. Anything else to add onto that before we move on to your next one?

[00:15:56] Matt Schwartz: The only thing I would add to that is, you know, AI could be an edge for you. It could also be not using AI at all, because ultimately it’s about the value you’re providing, again, to your client. So you may be able to build a two page, $5,000 website using AI, but essentially if you’re able to provide value to that client in some other way, whether it’s your sales process, your overall process, your personality, whatever it is, that all plays into this. So I would keep that in mind.

But overall, you are correct. I think the floor is rising for everyone. And this is real dark, but AI to me only showed us that a lot of the work we do day to day, it’s just not that special, it’s execution. And that just means we need to be spending more time on the strategy and the value to the client, whether that’s using AI or not. But I think using AI to at least look at that is a good idea, if you haven’t done that up to this point, I think it’s the time to at least look.

[00:16:54] Nathan Wrigley: Yeah, I expect the calculus that’s going on in most, and I’m using air quotes here, normal people’s heads. So when I’m talking about that, I mean non-technical clients who might be coming, looking for a website for their bricks and mortar shop or whatever it may be. The calculus of AI is just this shrinking of time. The thing which probably would’ve taken a week to do, you know, okay, I’m going to phone you up, we’re going to set up a meeting, we’ll have that meeting, we’ll back and forth what we might want. And then a week or two later, you’ll show me a few wire frames or something like that.

That all seems now to have been crunched into literal minutes. That’s no longer a secret. I think at the beginning of the AI, movement, let’s call it that, a few years ago, I think there was a call that you could basically say, make more money, because the client’s expectations would be the same in terms of time, and the amount of expertise that was needed.

But that seems to be shrinking as well, because now the clients are aware that the AI can do that thing. Look, you just knocked it up with AI in three minutes. No, we’re not going to pay you for six weeks for that kind of thing.

Okay so, right, there we go. So that’s the now. Then you move on to something that I’ve not even thought about before, which is creating an AI vision document. Now you’re going to need to explain what you mean by that I think.

[00:18:08] Matt Schwartz: Yeah, definitely. So the idea really with this is being more purposeful about adding AI. At least in the past, at my agency, you know, I’d wake up one day freaking out and be like, we’ve got to try to see how this works. Does this make sense to add AI to this process? And it wouldn’t be very purposely built. It would just be like, hey, let’s try this.

And to some extent that experimentation is good, but at the same time, I think if you are a lot more methodical about that process, it will be better for the long-term use of AI. Because as I’ve said, I definitely have, maybe I didn’t say, I have hesitations about AI and I use it, right? I think that’s the paradox of what’s happening. A lot of people are using it, but we’re not all trustful of it. But we can see that there are potential gains and you want to be on the cutting edge.

But ultimately, the vision document, the idea is that you will create a document that outlines all of your processes at your company, at your agency. And see what are the things that you want to add AI to, what makes sense, things that are repetitive, that the team is losing time on, compared to things that really require human judgement.

So it’s not just like shove AI into everything. This AI document sounds like that’s what it is. But it’s actually like, there may be many places we don’t want AI at all, or we may want to have a guard for human judgement. And I think that’s actually a really good idea to protect your agency from risk and really just your reputation. Because otherwise, I think a lot of agencies are just kind of, you know, yolo, adding it everywhere and not really thinking about it from a high level.

The other neat thing you can do if you’re building this is, obviously you can use AI to help find patterns in your business from using your time tracking software to see who is working on what tasks, and what’s taking the longest. And being like, are these good places that we could use AI? Like I can connect Claude to Everhour that we use, and it can spit out who’s working on what in the past month. And I actually can get a good pattern, because one thing you can, I will say with AI is it’s pretty good at pattern recognition. That’s what it was built for. So if you just need like a high level idea, again, grain of salt, but at a high level, it’s pretty good at that.

So I think for a vision document, pulling all that data in, using AI and then setting these guardrails, figuring out what in your team’s processes you can build into a vision is a good idea. And that goes into the high level point I made, which was really, I think AI being used for more process and agency is the big thing here, more than anything.

[00:20:42] Nathan Wrigley: I suppose if you’re the agency owner as well, and obviously agency could be like three people, two people, right up to, you know, several hundreds, maybe thousands, who knows. If you are at the, towards the top of that pyramid, let’s put it that way, knowing when and where it’s going to be used is really important. You need to know that, okay, our support, 80% of our support is going to be handled by AI. That’s the thing that we’ve leaned into. We’re going to do it that way.

Or maybe you are exactly the opposite. You know, we’ve learned, our customer base are very dissatisfied with the kind of answers that they get, because of the nature of our company and the expertise that we need to bring to bear. So we’re not going to do any AI for support.

But also development, to know, okay, this is the moment where you must stop using AI. When you run into this snag, we’re going to deal with that as humans. We’re going to huddle together, figure it out as humans, and maybe take it back to the AI at that point.

But having that overarching understanding, and writing it down. Having an SOP, if you like, for AI so that everybody’s on the same page and knows where it’s permissible and not permissible. So you mentioned you’ve got a whole laundry list of possible things. So it might be in the sales process, the delivery process, the proposal stage, project management, QA, launch. There’s a whole bunch on here.

Yeah, that seems like a really neat idea, and not something that I’d figured out. And it’s kind of like, keeps you honest in a way. It means that this is what we’ve agreed to do as a company, these are the boundaries that we’re going to set ourselves. And they can change, but for now, this is what they are.

[00:22:14] Matt Schwartz: Yeah. I think that even if you’re not, again, jumping in the deep end of AI, just having a doc like this will protect you so that I think you do have these guardrails with your employees or your contractors. You know who’s using what, and you can really protect your agency even if it’s not implementing more AI, right? I think it’s just a good idea.

And like you said, writing it down, it’s funny, it’s kind of like when you build your agency, you write your mission statement and your values and that really does do something in, I think, the human psyche when you do that. And I think that can be applied here with the AI vision document too.

[00:22:47] Nathan Wrigley: I love your fourth point, which you’ve entitled, AI as a new core service offering. Because this feels like a really nice sweet spot. Because with the best will in the world, you and I, and probably a lot of the people that are listening to this podcast are very much into technology. We deliberately put ourselves in front of new tech, new features, new widgets, new gadgets, whatever. So we’re beguiled by it. But the truth is, we know there’s a lot of people out there that aren’t, probably don’t really want to get all that close to it. And so I think what you are suggesting here is, why not offer your AI expertise that you gain as an actual service to clients? Have I got that right?

[00:23:24] Matt Schwartz: Correct. So essentially, like you said, if you’re already building up these new technology skills, being able to apply this directly in a, I would say in the proper way, right? Like we’re seeing, again, AI thrown in everywhere. You have to know your clients and your customers. They may not want to hear the word AI. What they may want to hear it instead is, hey, I can fix your business workflow and I can save you thousands of dollars, and we can automate this. They don’t want to hear the word AI, and that’s okay. But it’s essentially AI at the end of the day, right?

So it may not be that the product offerings actually use the word AI. If anything personally, I’m kind of avoiding that, at least at our agency. Of course I’ll tell them it’s using AI, but it’s not what I lead with. I think it’s more about going in on, okay, what solutions can we provide clients and using this as a new offering, especially as a way to handle and mitigate what’s happening with brochure sites, right?

Brochure sites I think are going to continue to drop and you need to provide value to clients. And I think getting closer to their actual processes, there’s a couple different ways you could do this. Like I know some agencies that are using AI to build custom web apps, like lightweight internal ones. Which I think can be helpful, but I have concerns around the risks and security of that because I do know some agencies that are, again, are just yolo building it. I don’t think they’re doing the due diligence. But I do think there’s a way that you can build, let’s say an app that used to cost 50,000 for 10,000 now, right? Or 8,000 and do it mostly like the right way, do human review of the code. So it’s still something that they couldn’t have done at all before. They couldn’t have had this custom internal app.

And I think that is the argument for people that say, hey, I’m going to replace all my SaaS products. It’s not, in my opinion, you replace all your SaaS products. If you can build a SaaS internally that is built specifically for your business, and you feel like you can maintain and build it properly at the right cost, sure. You’re willing to do that. But if there’s a SaaS product out there that does exactly what you need, I’m going to pay the $30, and then go yell at that company. I’m not going to build it internally. So having these conversations with clients, if you’re going to build custom apps, I know I went on a little side tangent, but I think that’s really important say.

And then the other one I’ll mention as far as AI core offerings is using more automation with tools like n8n or any of those Make type tools. n8n, I would say is a little more advanced, but the benefit is clients are hearing about AI, they realise it can do a lot, and starting to ask them, well, how can I help save you money or make you money in your processes? So productising or creating SOPs that are more automated. Even using those tools for your own customers, I think can be huge. Because then you’re really getting to value directly with them.

Like, brochure sites, I think the problem is, it’s almost subjective sometimes the value, which I’ve always struggled with, depending on the client. But things like their processes and them seeing you automate this stuff, they see the value immediately. So it’s an easy sale that you can make. And you can provide that value, and potentially even get recurring income off of that. Because maybe you’re hosting the automation for them or you’re tweaking the automation. So those are some ways you can mitigate, I think what’s going on with AI.

[00:26:50] Nathan Wrigley: The next one, I’m just going to skirt over quite quickly because I think everybody can kind of grasp this. One of the things which AI is obviously superior, let’s go with that word, to the typical human, is its capacity to wrap its arms around a massive amount of data, and kind of make sense of its straight away.

One of the areas where I think you are saying this could be deployed pretty effectively is in things like marketing, where having an understanding of, I don’t know, geography, spending power in different geographical locations, what kind of products are going to service the market that you are launching into, and therefore how to build websites, pages that kind of react to that and will work well.

That’s the kind of thing that was always off limits to me. I wasn’t interested in the marketing side. Looking at that data, trying to digest that data, it was just never of interest to me. And now, I think everybody can understand that you point an AI in the right direction and it can draw conclusions, which are just so much more credible than somebody like me could summon up in six months of hard work, really.

[00:27:53] Matt Schwartz: Well,I mean I think you could sum it up, but I think you bring up a really good point, which is that with AI, it can pull in all this data and it can give you, I would say, summaries and next points that you just wouldn’t have done before. I actually think that’s the sweet spot with AI is, are we using this to replace a really good existing setup, or are we doing something that we literally couldn’t even do before because the client couldn’t afford it?

So I think that’s what’s really neat is I can be like, okay, client, we looked through your Freshdesk, we looked through all the data you gave us. Here’s what we saw your personas. And before, there’s just no way, as an agency, I would be offering that at the budget that they could afford, or maybe the interest as an agency to do that. So I think that is, a really neat thing is, especially for small businesses, we can offer them services that they just wouldn’t even be able to have in the past at the budget that they have.

[00:28:46] Nathan Wrigley: Okay, so really you are kind of broadening the product offering that you can have. I mean nobody here is going to advocate that you just use an AI and regurgitate whatever it says without some background knowledge that what you are saying makes sense. There clearly needs to be a bit of that. But the amassing of the data with some common sense, heuristics around what it is that the data is showing you.

Okay, that’s interesting. So maybe there’s some sort of low hanging fruit that previously you would’ve said no to and, look, we just don’t do that. You can now not only retroactively sort of say, yes, we now do that, but maybe even proactively say, look, we’ve got these other things that we can discuss as well. Okay, that’s interesting.

Right, here’s the next bit, and this is, I think if you are not an AI expert, and I definitely would consider myself in that bucket, I think this next one is some really great low hanging fruit to get you started. So this is, your number six, is AI inside agency operations. So this is using AI to make work easier, I guess would be an easy way to say it. So just run us through these points.

[00:29:51] Matt Schwartz: Definitely. So this is probably, if you’re familiar with AI, the most common uses. But essentially it’s going to be, you know, things like your meeting summaries, right? I think everyone has seen the bots that join in and, you know, there’s like 10 bots and there’s like two people and we’re like, are we in dystopia? Or it’s you and like 10 bots, and the other person doesn’t show up and you’re like, am I supposed to just talk to this bot? I think Mark Zuckerberg actually says he’s starting to have a bot fill in for him at meetings. Anyways, very dystopian.

But when it comes to meeting summaries and that sort of thing, I think where it can be really helpful if you’re not using it, again is, in the past, if I was having these discovery calls where I may not actually land this client, I don’t want to spend 20 hours trying to figure out the perfect proposal for them. It’s just not worth my time, basically, right?

So what this lets you do is it lets you, as an agency, do things you couldn’t do before, or you didn’t have the budget and resources to do. One would be on discovery. I can now take all the meeting notes, I can have it go to the client’s website and I can also have it look at my previous proposals. And I can have it put together a solution for this client, in terms of like what proposal makes sense for them.

To your point, I’m still going to review it. I’m still going to edit it. I’m still going to make sure that this makes sense, but I think that’s a perfect sweet spot again for AI. I know I keep saying it. Something I just wouldn’t have done before. I would’ve like, either I just spent 20 hours on it or sent a very generic proposal just to get something out the door. Now I can make it really a lot more nuanced because it can go through all that data.

So if you’re not using it for summaries or proposals or SOWs, I think a draft version of that, it’s really good at those sort of things with combining all the data.

[00:31:36] Nathan Wrigley: I am so surprised by how quickly that remarkable technology became utterly mundane. That is say that three years ago, the first time somebody dropped in a Zoom meeting with an AI bot, I thought, okay, that’s really unusual, what’s going on here? And then within three minutes you get the email after the call is finished and you see this perfect summarisation of exactly what you talked about, including correctly labelled next tasks for each of the individuals on the call.

That to me was, I was living in Star Trek. And now that just seems so pedestrian. And that’s remarkable. That’s the speed at which we’ve become adapted, and it’s become part of our modus operandi.

And if you haven’t used those, it’s really worth a try because you will experience the amazement that I had three years ago. And then you too can become completely numb to how amazing it is really quickly.

It literally will take an hour of audio and spit out a basically perfect summary in 150 words or whatever it may be, and it will capture it perfectly. I suppose the rebuttal to that is, well, what do you do with that? If nobody does anything with that then, well, you haven’t really lost anything. You’re in exactly the same place as you were before, but at least you’ve got a written record of it.

But like I say, that’s the low hanging fruit. They’re definitely things. SOWs, SOPs, meeting summaries, that kind of thing. Great idea.

Okay, next one. Number seven. AI for support workflows. What’s going on here?

[00:33:04] Matt Schwartz: Yeah, so this one’s a little bit more about the actual operations. But I’ve talked to some agencies that are starting to really build into their support process AI tools. For example, using things like n8n, the automation platform, where it can digest your help tickets. And we’re not necessarily going to have it solve the problems, right? But what it can do, again, is it’s going to have access to a lot of data about that website. It may have access to your project management software, all the other tickets that came in.

And unlike a human where it would take hours to do this, so we just aren’t going to do it, it can do a really good job of essentially making sure that we can have all the information we need for the support person to do what they need to do, the support team, right? So it can even give good initial resolutions for the team to do, so that they can work through tickets faster.

That’s a good example of, we’re not replacing the human, we’re not trying to automate it so it emails back the customer. But what we are doing is we’re taking in all the context of, hey, it’s this client, they’ve had these other tickets, it has access possibly to the WordPress site, so it can even see the error logs. It may have access to the server APIs, so that it can actually see what’s going on with that server. And then it can basically come up with a resolution that is likely the issue.

And you are seeing a lot of, even hosting companies going that route, where they’re starting to have agents inside their hosting so that you can pinpoint issues in WordPress a lot faster than you could in the past. And again, I still want a human to review that, but I do think by doing that, you can get a speedier response to your customers, and you can cover more tickets without alienating your customers or making it seem like it was, you know, written by a robot with em dashes everywhere, right?

[00:34:54] Nathan Wrigley: Yeah, you know what? I think this is a real area to tread carefully, certainly from my point of view, because I have definitely got AI bot fatigue. In that, there is some button that gets pushed when I find that I’m in a chat bot, and that is the only route that I’ve got through this whole system. I really dearly love to get in front of a human quite quickly. And I think a lot of people are learning that technique of, you know, the first thing you type is, speak to a human, or something equivalent to that. I think it’s really easy to misstep here, and misjudge people’s capacity to take AI only, or AI mostly or whatever.

This I think will be an interesting area to watch. And maybe this will be at the vanguard of when people express their frustration, you know, how much of this can you take? And monitoring that and keeping sight on when people’s, I don’t know, anger boils over because they’re not getting the service that they paid for or the service that they’ve come to expect or what have you. So, yeah. Anyway, that’s my 2 cents on that.

[00:35:50] Matt Schwartz: Hundred percent agree. It’s the most sensitive portion in my opinion. I mean that’s your touch point with your customer when they’re most frustrated.

[00:35:56] Nathan Wrigley: Right, that’s the pain moment. And introducing additional pain at the moment of pain is fraught with problems. And we’ve seen this play out in all sorts of other ways. I’m sure it’s the case where you are in your part of the world with telephone systems where you end up in this just infinite loop of, press three for, and then press four for. And then eventually you get back to, oh, well, I’m back to pressing three am I? Okay. And the anger boils over.

It feels like such a win. We’re saving time. We’ve got the AI to answer because it’s read all of our documentation. I’m going to guarantee that somebody will not be able to get what you think they ought to be getting with it.

And dare I say it, what about all those dear people out there who really are unable to access the technology in the way that you anticipate, or the way that you can. Maybe they’re elderly, maybe they don’t have the capacity to do it. Maybe they’ve got accessibility needs or something like that.

Okay, number eight. AI assisted debugging and WordPress management. I like this. This is a good one.

[00:36:53] Matt Schwartz: Yeah, so we covered this a little. It goes along actually with the above point which is, one thing that I see other agencies, and we’re also doing this internally is, you know, you can obviously connect AI agents now to WordPress sites directly, obviously with guardrails in place. But it can connect to the REST API you have the Abilities API with Automattic. There’s third party solutions like Novamira out there that can actually work with the PHP code side of things. Your hosting companies often are actually building their own tools as well.

So doing all of that, debugging has been, I will say, has been dramatically improved, at least at our agency. Because it can do all of that and it can really find a nuanced solution where, you know, we could spend 10 hours trying to work on some weird PHP issue because, again, it can look at the whole picture. And I think that is where AI is very good, is when it’s a one-off thing, right? Where it’s just like, this is a one-off troubleshooting task. I don’t want to spend 10 hours learning exactly what this was. It’s likely going to get you there, and then you can obviously finish it up if it’s not able to get you fully there.

But you can use these tools today to really reduce the amount of debugging and management you’re doing. And you can extend it. We’re not going to spend a lot of time on this, but doing edits on websites, a lot of page builders now are starting to build in syntax for agents so that it understands Gutenberg blocks. It understands how to edit and edit nested blocks. I’ve had struggles with Claude, where it would try to write nested blocks and it would just mush the whole page.

But as these page builders are becoming better, and as WordPress becomes better, essentially WordPress becomes the infrastructure, right? And Claude is actually doing the work. You’ve heard that. And what I get out of that with the infrastructure is WordPress is the platform, it provides all the capabilities, but then the AI tool, mixed with the human, is essentially going to be managing the WordPress site. And it’s much easier to tell AI to do that than to go into the backend and make edits.

But I am a little hesitant on just making free flowing edits, not checking the work on the actual website, or letting AI check the work. Some people are doing that. I’m not doing that. We’re saying, give us the link after every page you edit, and I’m going to go click it and I’m going to look at it.

Some agencies, they’re saying, okay, Claude’s going to go to the Chrome link and do that. Whatever you’re comfortable with, but in our opinion, there still needs to be human review. And I still don’t think that’s going to change, even if it gets better because until AI is as good as a human being, in the sense that we can trust it and it won’t lie. I give this analogy, right? You hire a developer, they lie to you twice, you’re probably going to fire them, right? But with AI, we just keep giving them a second chance. And, why?

[00:39:41] Nathan Wrigley: Free pass every time.

[00:39:42] Matt Schwartz: Why are doing that? And I think the way to mitigate that is you still have to have human review based on the risk factor. That’s really what it’s about.

[00:39:50] Nathan Wrigley: Yeah, I share your sentiment there. I think it’s very important to have a human in the loop. And usually at the end of whatever is going on, there needs to be a human just to do the sort of final summary and checking and what have you.

But the point that you mentioned there is, WordPress really has done an awful lot of work in the background to make itself AI ready. So a lot of the capabilities inside of WordPress, a lot of the things that you would normally have had to engage with the admin, with a mouse, or with a keyboard or what have you, a lot of that has been taken over.

And we are very much entering an era where WordPress becomes almost like the scaffolding for the website in a way. And you can talk to the website through these AI agents, but in many situations, I think in the next five, six years, there’ll be a lot of people who will be never visiting the WordPress admin and clicking around and trying to find the menus for things because they will simply ask an AI.

Can I change the clock to the 24 hour clock? Sure, done. And that will extend into everything. You know, I want that block to be, I don’t know, I want the text in that block to be bold, and have this particular font and yada yada, on it goes. And WordPress is doing a really incredible job at an incredible speed of laying that foundational work.

If you haven’t looked at what the Core AI team are doing, there’s definitely some interesting stuff.

[00:41:06] Matt Schwartz: It’s really neat.

[00:41:06] Nathan Wrigley: Yeah, it’s, and I think that an interesting and commendable approach as well, because rather than trying to, I don’t know, hold everything into WordPress, it’s very much the opposite. It’s, we’re just allowing everything to communicate inwards to WordPress. And WordPress will just be the foundation upon which the whole thing resides.

Okay, so we’ve got through 8 of what turns out to be 16 points in Matt’s comprehensive show notes. And just looking at the clock, Matt, we’re at it’s kind of 40 odd minutes, which is about the sweet spot. So I’m going to recommend that we split this up into a second episode. So this in effect, will be the first of a two part mini series, if you are okay with that. How do you feel? Is that all right with you?

[00:41:45] Matt Schwartz: Definitely. You know, I didn’t know we were going to dive this far into it, but I’m so glad we are. And I hope, you know, the audience is interested in staying around for part two.

[00:41:52] Nathan Wrigley: Okay. In which case, if you are happy with that, what we’ll do is we’ll knock it on the head, as we say in the UK, here. We will return next week with the second part. And I will advise people at that point to listen to the first part so they can keep up to date.

So we will see you in a week’s time. I guess all it remains for me to do, Matt, is to say thank you very much for joining me today. Part two next week. See you soon.

[00:42:14] Matt Schwartz: Thank you so much. Look forward to it.