Generalists, Pricing Advice, and AI’s Role in Development

Zach Stepek:
Hello, I’m Zach Stepek and I’m here with Carl Alexander as usual for another episode of DevPulse. Expanding the stack. Carl, how have you expanded the stack this month?

Carl Alexander:
I was at Work Camp Europe. I don’t think we did a show since then, so I was at Work Camp Europe. I was there for the, the fair announcement. I been reading more about more AI stuff constantly. AI stuff. So that’s like the. Expanding the stack and watch some good videos, but otherwise just business as usual. Expanding the stack, talking to people, getting their stories, knowing what’s happening. That’s what I love doing.

Zach Stepek:
That’s awesome. Well, it’s really hard to keep your guest a secret when everybody’s on video, so.

Carl Alexander:
That’s true. I never thought about that.

Zach Stepek:
Yeah, you know, it kind of changes the paradigm a little bit. Right. Normally we, we kind of keep the, the guest a surprise when we’re recording until we introduce them, but you’ve been here the whole time. Jason, how are you?

Jason Cosper:
I am doing well. Hey, Carl. Zach. Thanks for having me.

Zach Stepek:
Yeah, yeah, it’s, it’s always great.

Carl Alexander:
Absolutely. We’re excited.

Zach Stepek:
It’s always great to see you. I mean we’ve, we’ve known each other most of my time in the WordPress space, which is kind of crazy.

Jason Cosper:
Yeah.

Zach Stepek:
You know, all the way back to a little hip hop virtual machine action, I believe. Way back.

Jason Cosper:
Oh my goodness. Yeah. I remember when everybody was like, oh, HHVM, it’s gonna be the future. It’s really pushing PHP forward. There’s, there’s just gonna be so many. Yeah. Oh, and gosh, remember there was a minute where people were like, oh, are, are people gonna start writing plugins? Are we’re, are we gonna start rewriting parts of Core and Hack? And it’s like I, I’m, I’m actually really thankful that that never happened, but. Me too.

Zach Stepek:
Me too. And the necessity of an immutable file system, the necessity to reboot the, the machine once a day because it just stopped responding. Like when you’re in a, when you’re in a clustered environment like what Facebook operates. Rebooting one node on that cluster doesn’t really matter.

Jason Cosper:
Right.

Zach Stepek:
But when you’re hosting a single website and on a single VM and that’s all you have and you’re an E commerce store and you’re rebooting every night, you’re losing 15 minutes a night basically.

Jason Cosper:
Exactly.

Zach Stepek:
And that was fun. And then yeah, having an immutable file system, have everything on S3, it was a good time. And that’s kind of how we met. It was in the middle of that experiment when I was working with Oscar Mike. So it’s been a few, few days.

Jason Cosper:
You were doing that. I was over at WP Engine at the time and they really threw in on, on trying to make HHVM happen or at least sniffing around that. Fortunately, the PHP project got adequately scared by like how HHVM was eating our lunch and we managed to get a lot of really solid improvements in.

Carl Alexander:
I mean there was a gap, right? Yeah, it sucked because it crashed, but it was a significant performance until seven came out. Right?

Zach Stepek:
Yeah.

Jason Cosper:
Right.

Carl Alexander:
I think the whole thing kind of fell on itself once seven came out. But yeah, I put a fire under their ass to do the just in time compiling and all of that work. But yeah, I think there was a solid one to two year gap where I think I had it running. Yeah, it wasn’t always super stable. Like, I mean what we did, at least on my end was like we had monit tracking the process and then it would restart it if it crashed or something. So it was largely automated. But there was other issues that besides the crashing. Like it wasn’t actually like a hundred percent drop in replacement. Like there was weird things that would break. That was like. I think that’s what I remember more than the process crashing. It was just. And the errors were not useful. It was just like it was because it was a hack. It was HHVM crashing. It wasn’t giving you the proper logs or telling you what was quote, causing the problems. So yeah, I’m glad it’s gone. Like I remember that. That was like. Yeah, that was like 2016ish, you know, that was when I was breaking into the US market. Like community. I shouldn’t say market, but community. Because before that I was Canada.

Zach Stepek:
Only now we have a kind of interesting dynamic here on, on the show today because all three of us in some form have worked on infrastructure and hosting.

Jason Cosper:
Yes.

Carl Alexander:
Or still do.

Zach Stepek:
Yeah, well, yeah, I mean I’m Yeah, I’ve definitely, you know, doing.

Carl Alexander:
I literally have a database migration tonight for four hours. So I, I’m still in it.

Zach Stepek:
Yeah, yeah.

Jason Cosper:
I’m implementing a stack for effectively a consultancy that is doing some work around. Basically every time they build the site, instead of going to someone like Big Scoots, WP Engine, Kinsta, you know, anything like that, they just say, okay, we’ll host it in house. And they have been hosting stuff in house on effectively cPanel on AWS and they’re still running Apache.

Zach Stepek:
Oh wow.

Jason Cosper:
And it’s the default like WordPress, like cPanel Stack. It’s not even like the nginx cPanel Stack. And I was like, we can do so much better.

Carl Alexander:
Yeah, yeah, we can. There’s a lot of room. There’s a lot of room without getting to the fancy stuff.

Jason Cosper:
Right. Yeah. I can get you spun up on something that will work a lot better for you. Fortunately, the person who their CTO, because it’s a small team, is like, oh yeah, I’m comfortable using the command line. And I was like, oh great. And you know, you just saved yourself me having to build you some like rudimentary UI to like, you know, reboot services and you know, all this other stuff. I’m like, okay, okay, I can build you some tools. But yeah, no, we’ve all done infrastructure and really, I mean you’re, you’re still like, Zach, what have you been doing infrastructure wise lately?

Zach Stepek:
So right now I’m the director of partner programs at Big Scoots.

Jason Cosper:
Yeah.

Zach Stepek:
So I am helping agencies pick the right infrastructure and the right hosting for the projects that they’re bringing to us and making sure that they have a trusted partner that is not just a vendor. We never want to be a vendor. We don’t treat anybody like a number. That’s kind of what I’m doing at the moment. I’m also doing some other projects, you know, internally at the company. Because you wear numerous hats no matter where you are. Right?

Carl Alexander:
All the hats, all the hats, all the time.

Jason Cosper:
That’s a running. That was a running joke at my last job at Dreamhost. They said that I wear a lot of hats, mostly Dodgers ones. And I am never, if you can see behind me for the folks actually watching, I’ve. I’ve got a, a nice collection of big baseball guy and. Yeah. But also a lot of technical hats for sure. Skipping around between teams.

Zach Stepek:
Absolutely. Yeah. I mean I’m doing, I’m doing a design project right now. Right. That’s just. It’s part of the job. Right. Marketing is part of partner relationships and so doing some design work and still doing some coding here and there. But primary goal at this point is helping agencies do what they do really, really well. And so that’s where I’m at now. Carl, we know you’re still working on a mirror.

Carl Alexander:
Yeah, I’m still doing Emir. I mean it’s just. Does a thousand. But. But I do. I mean I always. My consulting is always WordPress adjacent, I call it. Although I got at Stern talking to about not doing that as much and just doing everything. But. But yeah, no, it’s like cloud infrastructure, auditing design, product development. I did fractional engineering leadership, which I really enjoyed. It’s just whatever, you know. For me, I’ve always had this expert generalist mindset. So it’s more what’s your problem? And sometimes I’m not the right person to fix it, but half the time it’s just like you need somebody, right. Like to do all these things and I’m really good at doing all the things. So I would say for Emir’s challenging me more on the marketing side. But like I wrote a book that did really well, so I know how to market stuff too. So it’s just. Yeah, it’s just. I think for me it just helps me like having all these hats. Like, I think it’s more of a personality thing. I don’t think I’d enjoy being a specialist as much as I enjoy being a generalist. So it’s just. It fits also, but it’s useful for companies. I. Sometimes I feel like companies aren’t as aware that they need that kind of work. I don’t know what your experience has been for YouTube, but I find that sometimes they. They don’t look at it when they’re hiring. Once they have the person, they’re like, oh my God, you’re great. I love being able to stick you everywhere, but they never think of hiring somebody that they can do that with. I don’t know if that makes sense.

Zach Stepek:
Yeah. I frequently tell people that I may not be able to fix the problem you’re having right now, but I definitely know somebody who can.

Carl Alexander:
Oh, yeah, that too.

Zach Stepek:
And if I can’t and if I can’t find somebody who can, I’m going to figure out how to fix it. Like that’s. Those are the two options. Right. I’m either going to connect you to somebody in my network who already knows how to do the thing, or I’m going to help you figure out how to do the thing.

Carl Alexander:
Yeah, yeah, exactly.

Jason Cosper:
Absolutely. That. Actually thinking about that. I am also. I guess all three of us are kind of generalists. I don’t know if either of you have read a book that came out within the past few years called Range. It is effectively. And I’ll send y’ all a link so you can throw it in, show notes or whatever. It’s Range. Why Generalist Triumph in a Specialized World. And I remember someone recommended it to me.

Carl Alexander:
What’s the subtitle again? Can you say it again? I didn’t hear it.

Jason Cosper:
Well, why Generalists Triumph in a Specialized World.

Carl Alexander:
Oh, cool.

Jason Cosper:
Yeah. And I. It was recommended to me when I was between careers last time around and really ended up kind of like, boosting me a little bit because I was like, you know, watching people, you know, get positions and pick up new careers pretty easily. Things are, you know, four. Four years between the last time I was looking for a job and now things have gotten a lot crazier things. The market is a more competitive. But it really kind of, like, hyped me up a little bit where I’m just like, yeah, man, My. My generalist skill set is actually much more of a commodity than I. I think it is. Have either of y’ all ever seen Sports Night, the Aaron Sorkin show from the 90s, early 2000s?

Carl Alexander:
No, not that Aaron Sorkin.

Jason Cosper:
I love.

Carl Alexander:
I’ve watched a lot of Aaron Sorkin, but not Sports Night.

Zach Stepek:
I have seen a couple of episodes.

Jason Cosper:
Okay, did you catch any of the run where Bill Macy is. William H. Macy is the consultant?

Zach Stepek:
I think that’s the one episode that I think I have seen was in that arc.

Jason Cosper:
Sure. Yeah. That really, I’ve found, especially over the past few months as I’ve been doing consultancy work in Little Room, the consultancy that I’ve run for several years, but just finally, like, you know, started doing a big push, like me and everybody else who was let go from their job in the past six months. You know, it’s time to focus on my consultancy and get a little money coming in the door. But another friend of mine suggested that I go back and watch the Bill Macy run on Sports Night because, like, as a consultant, as, like, you are a generalist, you know what needs to be done. Sometimes you may not have the technical skills to implement it, but you know what needs to be done. You have a set of, like, very, you know, serious and complete, like, basically recommendations for, like, here is how we move forward. And my friend was telling me, he’s just like that. He goes yeah, you can make money implementing stuff. He goes, the crazy thing is you can make more money walking in and telling somebody, here’s what you need to do. Pay me, please. He goes, you don’t even need to actually do the work. The second you start doing the work, they see you as somebody that they already pay in house to do the work. So the purse strings start to close up. He goes, but if you walk in and say, here’s what I need, or here’s what you need to do, go ahead and take your people and implement it. That will be 300, 500, whatever, an hour you want to charge. Then the people in charge will just be like, oh, okay, well, this guy knows what’s going on. Let’s go ahead and pay him. And yeah, it’s wild.

Carl Alexander:
Yeah, that’s a lot of how my consultancy works. But I will push back because that’s where I got the Stern talking to actually. By at work Camp Europe by Alain. Like, I never know how to pronounce his family name. Like Schlesser. He’s at YOST right now. Like, he maintains. WPCLI and I had been in talks where a company to do consulting work for them, and then things fell true and they had new leadership. I talked with their leadership and they were like, oh, we need. Like your. We might need you. But it was completely different. It was like I was going to help them win infrastructure. Now they’re talking to me about building a plugin. And I told them, well, I’m $300 an hour. You should probably not use me to build a plugin. And then Alain got really upset at me, actually, and he said, that is not your job to tell them. You know, if, you know, you’re a really smart guy, what you can pull off in three at $300 an hour is not the same thing. And it’s up to them to make the business decision whether they want to invest in that, because, like, the plugin you would design for them. Like, I have code in WP Rocket that’s attributed to me. It’s like just stuff I wrote. But it runs in millions of sites. You know, I can write code that runs on millions of sites that’s worth something. You should not be telling them. That’s not worth it. That is not your job. So there’s still something. It’s tricky. The point is that it’s tricky because like you said, I. I will just say that my experience around consulting is that once you hit a certain amount, it doesn’t really matter what you do. For them like they will treat you differently.

Jason Cosper:
Yeah.

Carl Alexander:
Like and I think that’s really important and what I shouldn’t have done and the lesson I took from Alain’s conversation is that I should not be telling them what they should be spending their money on. They’re the business people and you just tell them what you’re worth and they can decide whether they’re willing or not. And I think that was me because I consider WordPress work to be not worth that much money. But that’s my problem. It’s not something that’s really reflected by the market. It’s just my prerogative because I feel like it’s. It’s beneath being paid $300 an hour but actually it’s not necessarily right and that’s.

Zach Stepek:
No, not at all Carl that you.

Jason Cosper:
And everyone else honestly with the I don’t feel like WordPress work is worth a lot of money. I because I mean yes I am running consultancy like it’s you know helping keep a roof over my head helping you know not burn through my savings and and everything else I’m getting work etc here in the US like though unfortunately having a job a 9 to 5 is is attached to health insurance for a lot of folks so I.

Carl Alexander:
Know that’s the craziest don’t like this is not a place to go onto this tangent. I can’t right. I cannot. I could not. I cannot for the listeners I’m in Canada so that’s why But I cannot.

Jason Cosper:
Still looking for you know jobs looking for places especially like I’ve I built an almost 20 year career. I’ve been 20 years 20 years in WordPress and 19 of that 20 years I worked in web hosting. So like I’m out there looking for another day job. If it was in web hosting. Great. Because that is effectively like where a lot of my domain knowledge is. But just Even looking for WordPress jobs online, the disparity of how much people are willing to pay, you look at somebody who is, you know, hiring for somebody who does, say, infrastructure and you know, they’re, I mean, I, I interviewed at. Oh gosh, Extendify. Not in the WordPress world, just expense tracking software.

Carl Alexander:
Yeah, I know, I know Extendify from oh my God. But yeah, pre Covid I was dealing with somebody with that.

Jason Cosper:
But yeah, their salary range was like 160 to 200, 120,000 a year for an individual contributor. Yes. At Extendify and versus you have somebody who would be doing a similar role. But the second you attach WordPress to that because it’s seen as free software, it’s seen as inexpensive, it’s seen as all of these other things you run into, these roles that, you know, oh, that are. They either don’t list a salary or they say like, oh, our salary range starts at like 110, $120,000 a year.

Carl Alexander:
Yeah. I mean, that’s why I frame it as WordPress adjacent consulting.

Jason Cosper:
Yes.

Carl Alexander:
Like, it’s just clever things I’ve learned. But also it’s just, you know, a lot of the work that I do is not directly in WordPress. It just involves WordPress in some way, whether it’s like big WordPress sites that I am helping or auditing cloud infrastructure costs for somebody hosting WordPress. But yeah, absolutely. I mean, salary discrimination is really interesting. Like I, I’m in Canada. Right. Like, we might get the same salary range, but in Canadian dollars for the same job. I know like recently Pantheon had a job opening, but it’s in Canada. Like, I don’t know if they’re hiring us, but it was like same kind of decent salary range, but it’s in Canadian dollars. And I’m just like, why? Like, you know, like I know how much you’re paying in the U.S. like, I want the U.S. dollars. Like, please. You know, so. And you, and it’s less, you know, they don’t have to pay for health insurance and all that stuff too, because there’s just not of that in Canada too. Right. Like your, your costs are different. But yeah, I mean, it’s, it’s really tricky. Basically. It’s, you know, there’s some high salaries in WordPress. It’s just people want to pay less. They think it’s worth less. And we reinforce that too. Right. Which is why Alain got upset at me. Right. Because he’s like, actually, no, like you are. You will build something for them that will work, that will be worth that. And he’s right. That’s why I didn’t challenge him on that. It was a mindset thing. A lot of this stuff in consulting is mindset.

Jason Cosper:
Yep.

Carl Alexander:
You know, everybody asks me how I get the charging this much, and I literally tell people. I just. I was. I didn’t know how much I was worth. I would grandfather all my clients, but basically every new client, I would charge 25 an hour more. I slowed down eventually, but basically that was just. You don’t know, like. And we’re not necessarily wired for that too. So it’s just like, how do you present yourself? How do you market yourself? And. And people pay. Like, I’m here telling. I’m here telling people that are listening that people will pay $300 an hour.

Jason Cosper:
Absolutely.

Carl Alexander:
They exist. They exist. It’s not easy. I have lifestyle decisions that allow me to not work that much. Like, you know, I don’t have a. I don’t have a mortgage, I don’t have kids. So, like, I’m just really upfront about that stuff. But you can negotiate that. Like, how do you keep a pipeline for making, you know, 120, 130,000 a year doing that? I’m not sure. That’s not my. I don’t have experience with that, but I could tell you, you could charge that much and people will pay.

Jason Cosper:
Right. I had. I ran into a friend of mine, sent me a referral, and it was a small business. They needed. Effectively, what was happening was they were running into this issue that was driving them nuts. They were using the event calendar to display effectively all of, like, they run like, a site for, like, family events and things like that where, like, family friendly events and, like, the Los Angeles area, they do a good job of, like, actually tidying up. And like, they’re like, once an event has passed, like, they pull it out of the event calendar. Like, they do all the appropriate garbage collection and everything else. But there was something about the way a previous developer had implemented the pagination. They did a custom one for whatever reason, I think they were doing. Yeah, they were doing a custom query to, you know, pull particular data and have some penned stuff. This is like a legacy site that had been built like, 10, 12 years ago. So, like, really digging in. Yeah, really digging into a lot of that stuff. But they had this pagination issue. Once they hit, like, page three, four, or five, at some point, it would start displaying page one results over again. And you could never get past, like, page three, four, five, whatever. And they’re like, nobody. Nobody can see any of, like, the events that are happening, like, more than two weeks out. They had paid developers. I think I was the sixth developer that they had worked with. They had paid developers a few thousand dollars effectively in. At, like, you know, $30 an hour, $50 an hour, whatever. I was like, I do not charge this. Like, because. Because they were a small business. I went and I said, you know what? I will charge you 150 an hour. And I said, this is what I’m going to charge you. It will. This will not take me long. I know the internals. I think I see what your issue is. Like, I. I can help you here.

Carl Alexander:
Yeah, but even then, you would communicate with them if you were going, like. Like, a lot of it is communication.

Jason Cosper:
Absolutely.

Carl Alexander:
Like, I always. Zach’s heard me say this before. I think I might have said it on the podcast before, but I can charge a hundred dollars an hour just because I answer emails.

Jason Cosper:
Yes.

Carl Alexander:
Like, yeah, I always tell people it’s 300 an hour, but I will tell you every hour. Almost like if I feel like things are slipping, you will know way before anything happens. You won’t arrive and get a $3,000 bill for me. We’ll pause. I will wait for you to decide whether we proceed or not or how you want to proceed, and then we will continue. Because that comes with a certain responsibility. You charge more, communicate more, communicate more, be more on top of, like, what’s happening. Because that. It’s a sign of professionalism. Right. Like, yeah, you know, charging a lot is great, but it also comes. It’s an image. Right. Like, you’re not going to be. I’ve had less issues with payments and stuff like that for charging more, but there’s a certain expectation of professionalism that comes with that. And if you’re not doing that, then it gets really complicated, because then you might get paid, you might not. But you’re burning relationships, you’re burning clients, you’re not getting referrals. There’s a lot of stuff that happens because a lot of times I get referrals. And Carl’s expensive, but he’s worth it kind of thing. You want to get that. You won’t get that if you, like, burn them.

Jason Cosper:
Absolutely. And here’s the. Where that story was going is that when I took care of this issue for them. I said, it’s not going to take me long. I even said they were a little reluctant. I said, I’ll tell you what, I am going to time box three hours. I said, I will fix your issue inside of three hours. And they’re like, I have paid multiple thousands of dollars to other developers. You’re telling me at 150 an hour that you can fix my issue for $450 if this works. Yeah, sold. Okay. And I said, if I cannot fix your issue in three hours or give you an estimate of what the rest of your work costs, I will walk away. You don’t even have, like, I, I was just like, you know what? You’re a friend of a friend. Like, I want to help you out. I fixed their issue. I fixed their issue in like an hour and a half and then spent the rest of. Actually, I probably fixed their issue quicker than that, but I spent the rest of the time actually testing the issue because it was a pagination issue. I sat there and like spun up some browser test that sat there and just kept progressing through the pagination until it hit the end and was taking snapshots for me. But like, I spent more time writing the test that I ran to fix their issue than actually fixing their issue. And they were so thrilled. They’re throwing me more work now at that rate that I asked them and they were just like, I’m never going to pull another, you know, $20 an hour person off of Fiverr again, because I described my issue. You knew what was going on and.

Carl Alexander:
You fixed it and you communicated, yeah.

Jason Cosper:
The, the infrastructure thing that I’m working on right now. I specifically, I was like, okay, I’m doing this over a set amount of time. Like, here’s thing, how things are going. Once I get within about five hours of completion, I will let you know my remaining time estimate. It’s probably going to be five hours, but if I have a little more time to work on stuff, I’ll let you know. Like, and I said, if you guys have like a slack or something you use, it’s a lot quicker than email. Like, let’s just like spin up a channel, temporarily invite me into your slack. I’ll leave after the project is over. But effectively, like, I’m going to go ahead and spin this thing up for you. And they were just like, yeah, sure, this is great. And anytime I run into an issue, I was emailing them. And yeah, no, emailing is great, but sometimes you run into a situation where you email and you’re just like, I am literally in the middle of trying to like put an API key, put your API key into a form somewhere or into a config file. And I cannot move forward until I actually have the API key in the form. So can you please like give me the API key? So I was just like, yeah, put me in a slack or something and I can actually get a hold of somebody like at the time that we work. But yes, communication is key.

Carl Alexander:
Yeah, exactly. It’s just communication, proactiveness.

Jason Cosper:
Oh God, yes.

Carl Alexander:
Like, there used to be a meme of me and where I would talk about hiring but soft skills to pay the bills, like with the Beasty Boys. But it’s just like that stuff matters. The more you charge, the more matters because it. The things change, dynamics change, right? Like at lower end you’re like worried about getting paid. And I’ve never done the thing where you have. I always invoice at the end and I’ve never been burned, basically. But that’s like a rule that you hear because, like, if you undercharged, you get what people that are price sensitive and things like that. But again, there’s just these dynamics that change when you price more. Certain expectations are more important and some of them, you know. But yeah, it’s great time. And generalists are really good at that kind of stuff, that kind of consulting, because you’re basically often brought in to look at things that have no root cause or no obvious root cause, or they’re not really sure. They’re just like, we have a problem, can you help? And you have to get to work and wear your all your hats to figure it out, you know, which is why I love AI. I feel like AI for generalists has been the best fucking thing, you know, Like I’ve like, I use it like so much because of that. I feel like it’s not as useful as for a specialist. But for generalists, it’s like insane. Like the stuff I can. Like how much more stuff I can do now.

Zach Stepek:
Yeah. I find that with AI, the thing that matters the most is the ability to direct the agent. And so a generalist who has enough understanding of every part of a system has better luck having AI create a complete whole and working end result than somebody who’s a specialist and doesn’t understand how those other systems work quite as well.

Jason Cosper:
I haven’t been as, as bullish on AI. I know that there are a lot of people out there who, you know, are using it especially for stuff like Copilot. But I’VE called it spicy autocomplete, basically because that’s what it is. Yeah. Especially when it comes to copilot and things like that. You know, that’s all it is. And as far as like I know that there are people who are like vibe coding and everything else with it. If I’m, if, if something is going to write bad code, that somebody writing bad code is going to be me and I’m going to stand behind the bad code that I wrote because I can understand what that bad code does. However, I have seen some use cases and actually the last place I was working at Dreamhost, they were pretty heavy into like. Oh yeah, just have chat GPT do like a summary of this for you or whatever. And I don’t want to read a, you know, 30 page report on, you know, projections for the next fiscal year or whatever. Yeah, if, if you can figure out how to give me the CliffsNotes version of this, great. I do want to make sure that sometimes, and I know that it can like say things that weren’t even ever in the report, like when it does summaries and things like that. So like I’m always a little bit leery of it, but for the people that AI is working for, I love it for them. Like I think if there is a useful tool out there that, that you can make work for you, great. So I’m not like fully against it, but at the same time, like I, I just need a little bit more and I need it to not be so, just drastically bad for the environment. The amount of like fresh water that is required to run that. I, I don’t love that. I don’t love that. Asking it to, you know, someone asking, oh, can you proofread my Blue sky post or whatever to make sure, you know, or write a blog entry for me means that like my nephews won’t have any clean drinking water. Not really a fan of that.

Zach Stepek:
Yeah, I think the future there lies in on machine multi agentic workflows. So sure. Rather than having to rely on these large servers that can handle millions of requests from millions of people, you know, having on machine small, you know, LLMs that can handle specific tasks. So for example, I’ve been playing around with this agentic tool that will do QA as you’re writing code and it just, every time you build it QAs in the background, it uses it like user and it finds issues and reports them to you. And so just being able to automate those processes that we’re doing kind of manually or with tools we have to drive and being able to tell an agent to do that every time a new build happens or automate that every time you save a file just really is helpful. And so there are some multi agentic workflows that I think we can automate away. Some of the things for low environmental cost on machine that we don’t necessarily, you know, always remember to do thoroughly. That’s my problem, right. I, I have shiny object syndrome. It’s a side effect of the ADHD. When I see something that, that I want to do more than the stuff that needs to get done but is more boring, I try to avoid it and, and do the other things right. So I avoid the boring. So if I can automate away what the things that don’t, you know, that I don’t enjoy and have an agentic workflow take care of that just automatically at 95% accuracy even it’s a huge win. And it’s not all there yet. Like the tools aren’t quite there yet, but we’re getting there. And the new generation of Mac processors, all the Apple processors that you know have this great built in chip for running AI tasks, it’s really pretty amazing what a Mac studio can do if you throw an LLM on it.

Jason Cosper:
Yeah, I mean I even have a, just a M4 mini sitting at my desk here and it’s just the stock model but. And I mean I’m not ante. I have, I installed Olama and put a couple agents like on there just to kind of play around a little bit and. But yeah, that lives on my machine. I love that like I kind of. And having these models that are like a little more like specialized and trained in specific domains. I find that that is one of the problems with like a co pilot with a thing like that is like even though PHP and even WordPress like runs you know, so much of the Internet, like they’re like PHP is not a sexy language. So they are not necessarily giving it and giving, you know, these LLMs training on PHP. And if they are giving it training on PHP it’s modern PHP. And unfortunately as we all know WordPress does not really use modern PHP all that well. So trying to like force modern PHP into, into that, you know, funnel, trying to be like. And I’ve seen people do it, I’ve seen people say like I think Matt Medeiros a few weeks ago was like, oh, I had it write this plugin or something for me. And I was like hey, that’s cool. Like I, I hope it Keeps working for you. But like a lot of these models were trained on like Python and stuff like that. And maybe try to use, I’ve seen in just experimentation and maybe it’s just the. Because it’s I’m using one of these like lower powered on the machine models or whatever where I’m just like hey wait a minute, that’s not how we would do this in PHP, let alone like even modern PHP, Laravel, something like that, let alone like WordPress PHP.

Carl Alexander:
I think I still need to do this AI talk the more I listen up. I, I don’t use, I still don’t use AI to vibe code, but I use AI like all the time basically I uninstall Copilot. I don’t use Copilot anymore. I canceled ChatGPT actually and Claude, I’m using Gemini right now. But either way you were asking how it helps. So for example, here’s a simple example. I have to manage a Postfix server. Do you ever have tried to find information on Postfix? They’re an old Usenet post from pre, circa pre 2000. Terrible experience trying to debug or fix LLMs. Can ask a question, have a discussion. It might be wrong, might be on the right track. It’s going to be 10 times 10 I think is an understatement. Probably a factor of 100 times faster to debug an issue where you have a general idea of where the thing is but no idea how to find it and Google it. Impossible. WordCamp EU I was talking with somebody, I have a lot of architectural discussions with AI.

Zach Stepek:
Sure.

Carl Alexander:
Because one of the problems that I think people have, like at least you know, my caliber of programming, especially in the WordPress space when you work for yourself is you have no peers to talk to, no peers to do rubber ducking with, no peers to do that kind of work, which is important intellectual work and creative work. I do that with AI. I have really long discussions where we’re doing things. My favorite one is AI solves one of the fundamental problems of programming naming things. Sure, I use AI to name things.

Zach Stepek:
Yeah, I get that.

Carl Alexander:
No, it’s honestly I tell people and they never think about it. But it’s like I, I, I’m. My autism really struggles with naming stuff. Like I want really good names and sometimes I don’t like the name I’m coming up with. I’m like can you find me like five alternatives? And then I’ll say I don’t like this because of this and that and eventually just through the discussion I will be like, this is the one I like. I’ve debated it with somebody even if they’re not a real person. And I have something. There’s just a lot of things like usages like that where I think people are too focused. Like there’s a whole discussion about the vibe coding stuff that’s too long for here. But I think people don’t think enough about AI just as peer communication, right? Like the. You know, there’s a reason they called it Copilot, which I love the name is because it comes from pair programming. I feel AI as a pair programmer is amazing and I’ve wanted to do a talk about that for a long time because every time I have these discussions it’s always about vibe coding. It bombs for me too. Like I try with different tools and it bombs. It doesn’t do it the way I want. I have to spend time doing it like telling it what I want. But at the pair programming side of things, since GPT4, like ChatGPT4, I’ve been using it that way and it’s changed my life, you know, like it’s been the most important thing I added to my toolkit since the IDE. And I feel like every time I talk to people about doing something like that, they’re always like, oh, I never thought of just having a discussion, you know, because they. You want that you. Because everybody’s focused on the autocomplete and vibe coding. And sure, that’s great and has really long important ramifications for the job of programming. But there’s still your day to day. Like you’re a consultant, you’re working by yourself, you know, at least in my case. I need people like I need somebody to talk with and the smart people I know are busy with their own jobs, you know, like to get that kind of feedback and get their full attention is hard. So I feel like people should think about that more because there’s a. That applies regardless. You’re a programmer, it applies to anything you do. Like if you want to have somebody and you’re working on your own and you want somebody to like really have an intelligent discussion with about something, just to make up your mind to red team yourself or whatnot, it’s really useful for that.

Zach Stepek:
Well, that’s where that multi agentic workflow I was talking about comes into play, right? You’ve got multiple agents that are acting as a team to assist you. So you’ve got your QA agent, you’ve got your code review agent, you’ve Got your linting agent making sure you’re following code guidelines. You’ve got.

Carl Alexander:
There’s already tools for that though. I use a lot of linting, static analysis and all that. I have plenty of tools to yell at me. I have no tools that can tell me am I going in the right direction with this architecture. Do you see anything? They might be wrong, they might be hallucinating, but at least shit like somebody to push back against me, you know, like a lot of that, like people don’t understand, at least maybe not consciously, that this push and pull is an important aspect of design.

Jason Cosper:
Yeah. I’ve had plenty of conversations with folks that they’re wrong, but at least they got me thinking on a new way of pushing on something.

Carl Alexander:
Absolutely. Yeah. And that’s important. So that’s where I feel, that’s where a lot of my interest and usage of AI falls under.

Zach Stepek:
Yeah, I think there’s a lot of possibility in the field of AI and LLMs right now. Personally, I hate the term AI. It’s not what it is. Right. It’s a large language model that happens to know things because it’s consumed data. They say that they have reasoning models now.

Carl Alexander:
It doesn’t actually know things.

Zach Stepek:
Right. They say they have these reasoning models now. All they are is deep research tools that are, you know, in an LLM, we’re not at the point where this is intelligent thought. Right. But it is a really good way to surface ideas, concepts, things related to what you’re doing. And in that way I think that, you know, will these LLM based tools that we’re calling AI replace tools like Google and Google searches in the future? Yeah, definitely. Is that how we’re going to discover things in this new future? Absolutely. So I think that that’s where we’re headed. I think we’re still very early days and I think that the future is going to be a very interesting place as these tools get better and as we get better at not destroying our planet with them.

Carl Alexander:
Yeah. And one my favorite saying is whatever we have right now is the worst that it’ll ever be. So whatever it is today is the worst that it’ll ever be. So that’s great.

Zach Stepek:
Well, and it’s funny you say that because when we were talking about pricing earlier, the pricing advice that I got early in my career that has always stuck with me is that the best price is the price that someone is willing to pay. So, you know, it really does boil down to make the decision based on where you can get work and if you’re starting out. Yeah. We understand you’re going to start out at a lower rate than somebody who’s got 20 years in the industry should be charging. Right. However, value yourself. Absolutely value yourself. So I think this has been a great conversation about. About how I might fit into some of what we do in the future and always great, Jason, learning a little bit more about, you know, your career and where you’ve come from. It’s always been a pleasure having you in my community and I hope that we get the chance to work together again soon. Any final thoughts for everybody before we wrap up here?

Jason Cosper:
Oh, man. Well, I am about to just on the precipice of relaunching. Well, not relaunching, but launching a new podcast, the Query, with my former WP Water Cooler co host, Sei Reed. I’ll make sure also to send a link to that to get in the show notes. It’s a. The Query show. And yeah, just kind of bringing back some of those old WP Water Cooler vibes for kind of everything that’s been going on in WordPress lately.

BobWP:
That’s great.

Zach Stepek:
Very cool. Yeah, I’m excited to hear that. Sei’s great. I got the chance to speak with Sei in person finally at press conf. That was great. So. And honestly, the Water Cooler Discord is one of the coolest places on the Internet that people don’t know about. But now everybody that’s listened to this episode does.

Jason Cosper:
Oh, yeah. And we’ll definitely. I’ll scrounge up an invite link so people can come join and hang out. Because even though that show has been gone for almost a year now, like, we’re still hanging out and having fun and chopping it up, talking about WordPress and also not WordPress very much.

Zach Stepek:
Yeah, it’s a great community because we’re not. We’re not afraid to speak our mind there, which is, I think, a really important thing. Like, I think there are some communities in which people are afraid to be open about how they feel. And that is death to a community. Fear of conflict is a horrible virus that destroys things. And so I’m really excited to be a part of a community that still is speaking openly even when we are wrong, because we’re human, we’re going to get it wrong sometimes. So, yeah, we’ll. It’s been a pleasure having you here, Jason. We will see you all next month with another episode of Expanding the Stack.

Jason Cosper:
Thanks, fellas.

Carl Alexander:
Yeah, see you then. Ciao.