Brice Ayres: There we go. I think it worked.

Jacob Luetzow: Maybe. ⁓ crap. One second. I see you. ⁓ I'm I think people can hear me. Okay, g guys, it's gonna take me a minute. All right, there we go. Now I think we're good. All right, what's up? I'm Jacob Litzo, your Elixir Mentor, and welcome to another exciting episode of the Elixir Mentor Podcast. This is where we discuss everything related to Elixir from interviews with enthusiasts and pioneers in the community to innovative projects and libraries shaping Elixir's future. Today's guest has led engineering teams, sold Elixir to leadership at companies you wouldn't expect. Hired for it and now runs his own fractional engineering shop, ⁓ Axio Intelligence. Bryce, welcome. It's good to have you. I'm excited to chat.

Brice Ayres: Yeah, likewise. We've been trying to connect for a little while here and so excited to be here.

Jacob Luetzow: Yeah. Unfortunately my life got in the way and now things are up and running. And yeah, sorry for all this mess, guys. ⁓ first time streaming with ⁓ Riverside. So I guess we'll see we'll see how it goes. And I don't know if anyone's tuning in. I don't know how to tell. But

Brice Ayres: Yeah, is there a live ⁓ feed somewhere? See live user?

Jacob Luetzow: There should be a live feed on like YouTube, Twitch, all the good stuff. I think on X as well. And then apparently everyone should be able to talk to us. So we'll see.

Brice Ayres: ⁓ I guess over there you'd see stuff happening. Yeah. Awesome.

Jacob Luetzow: Yeah. Well anyways. ⁓ yeah, I like to kick these off a little bit. ⁓ do you wanna kinda share who you are, ⁓ and like where you came from and how you found Elixir and all that good stuff.

Brice Ayres: Yeah. Yeah, definitely. ⁓ yeah, so I think I guess how far back to go. ⁓ I guess ⁓ back in high school, I took a web dev class, which would have been like in two thousand and one, two thousand two, you know, which is pretty cutting edge, I feel like, for a high school class at the time. And the teacher at the time he had no clue how to do web development. It was like an online course, which was like also pretty cutting edge at the time. And so you just show up this class, get on the computer. But this guy would come in and ⁓ he would just kind of ramble the teacher. ⁓ but he would tell to you about business he I guess he was a successful business guy before he, you know, kind of left the business world and came into teaching. And so he'd tell us about contracts, structuring deals, like a bunch of non-web related stuff. ⁓ and he's like, You guys can charge for this, you can go out and do work for this. And so kind of inspired me being entrepreneurial. I was like, Okay, I'll go out and, you know, I'll start building websites for mom and pops. And ⁓ I used ⁓ the skills I learned in that class to go build my own sites. ⁓ I think back in the day I was ⁓ hacking Xboxes. So if you remember the original Xbox, but you could mod it. So you could like solder chips into the board. ⁓ and you could do there was eventually a Yeah, yeah. And there's eventually a software mod you could do. And so that's really where I cut my teeth learning like networking and things like that. Cause you connect two Xboxes together to like transfer files. ⁓ so you learn the differences between two different, you know

Jacob Luetzow: I remember those days.

Brice Ayres: types of cables, this is a crossover cable, this is a d you know. and so that's kind of where I cut my teeth. But I I found I enjoyed ⁓ making kind of the software a lot of times was like the thing I would spend a lot of time on instead of the business sometimes. and so but it was it was pre iPhone days, right? And so software development wasn't super cool or wasn't I I guess I didn't know you could study that in school or college, right? And get a degree in computer science and ⁓ it's hard to believe now, but back then I just didn't know. And so ⁓ I wasn't sure what I wanted to do, what I wanted to be when I grew up. And so I had a buddy come back from the army, he was a medic and convinced me to join the army. ⁓ moved around a bunch and ⁓ I got stationed eventually in ⁓ South Korea. So I was up on kind of the border on the DMZ and the mountains there. And I noticed a lot of people were coming and going. It was a one year rotation. You'd show up, you didn't have your family where we were at because it was kind of a a ⁓ They call it a dependent restricted tour. and so it was just you show up for a year, get out of there. So no people wouldn't bring much with them. They maybe buywood c buy with buy a couple things, and then they would leave and they would just throw the stuff in the trash. And I was like, ⁓ I should build like ⁓ back in the day, Craigslist was really popular. So I was like, I'll make a Craigslist type website where people could buy and sell stuff ⁓ on the military base, but then nobody was using my website. And I was like, Well, how can I get people to use my website? And for Phone calls were really expensive at the time and they were really complicated. You had to go get a calling card off base. You had to like load it on this cell phone and put in this, you know, 20 or 30 digit code. ⁓ and so it was kind of a huge pain. And so ended up ⁓ working with a guy ⁓ who had like a voice or IP company and ended up building ⁓ basically a soft phone that worked from the browser, which was pretty cutting edge at the time. I think Skype was just kind of out around that time. And so End up building this ⁓ and I had a lot of friends that had deployed to Iraq and Afghanistan and they were like, Hey, this would have great to use on a government computer because I couldn't download Skype when I was deployed because the computer's all locked down and restricted, but yours works from a browser. So you could open up a browser and make a phone call from the browser. And this is pre, you know, Google ⁓ phone calls and G chat and all that kind of stuff. And so I basically started I guess they were it was called causes.org, I think was the original site. And it was integrated so tightly with Facebook, you couldn't tell if it was a Facebook group or it was some external website. And so I created this group called Free Phone Calls for Soldiers. And ⁓ it just blew up on Facebook. Yeah, and ⁓ and then there's another one called like f I started ⁓ it was ⁓ free ⁓ free internet for all deployed troops. And between these two groups, ⁓ ended up getting 3.7 million followers.

Jacob Luetzow: That's pretty that's pretty cool.

Brice Ayres: Which in two thousand nine might have been the biggest following on the internet at the time, which is pretty unreal. But like had no idea what to do with all these these people. ⁓ and people kept asking to donate money, and so ended up forming a non profit, ⁓ so we could take donations. And then we gave away ⁓ millions of minutes and calling cards and phone time, ⁓ to troops kind of deployed ⁓ overseas. And ⁓ was really cool experience. But I I realized the thing I was struggling with that I was more on the front end and building WordPress sites and stuff like that. And so I really wanted to learn the back end. And so after that experience, after I got out of the military, I went back to school, ⁓ studied at ⁓ a military college called the Citadel in South Carolina, got a computer science degree, ⁓ and, you know, learned kind of heavy back end. So I could con you know, 'cause at the time I was having to hire people to help me out on all this back end code. And I was like, I just don't understand how they're doing this or what parts ⁓ are happening there. And so ⁓ yeah, so that's kind of how I got into ⁓ software development from like a background. And then since then, ⁓ after college, I you know, worked at various different companies, ⁓ Top Golf, which I think some most people have probably been to, Teledoc, which is like a telemedicine company. most recently like was that Whataburger, ⁓ which I think if you're in the South, you might know, be familiar with.

Jacob Luetzow: Okay. I've been to one. They're good.

Brice Ayres: Yeah. Yeah, they're great. The in and out of Texas.

Jacob Luetzow: No, that's a that's awesome. What kind of got you onto the Elixir track then?

Brice Ayres: ⁓ so I first kind of came across it in it was like twenty fifteen or twenty sixteen, ⁓ working at Top Golf actually. And ⁓ it was interesting. We had a bunch of different services. We we used a bunch of different languages there. So you had stuff in Ruby, Python, some low level C for scanning like RFID chips and the golf balls and things like that. ⁓ but one of the engineers ⁓ spun up a small, small microservice in Elixir. And ⁓ it became my project. It kind of became my baby to take over. And ⁓ I just remember it being really challenging because I hadn't worked with the functional language before. And so I'm sitting there scratching my head looking for the for loops, you know, like I think most people do when they first get into functional like where how do I do this? ⁓

Jacob Luetzow: Dude, I struggled so bad. I yeah, because I came from, you know, object oriented programming. I feel like our our stream might suck, but I don't know. I yeah, I just tried to pull it up on YouTube, but at least when I publish this later, we should have solid recordings.

Brice Ayres: Exactly. Yeah. Hopefully, across our figures. ⁓ but ⁓ yeah, the it's interesting from ⁓ YouTube perspective or sorry from ⁓ functional functional programming perspective, right? Like everyone their first time seeing it is kind of confused, like, how do they get this to work? How come I can't reassign this variable? lots of interesting concepts there. But when I was all said and done and I kinda cut my teeth trying to figure it out, I was like, This is actually a lot simpler. Then, you know, a lot of a lot of the functionality that's in place in a functional programming language, right? Like keeps you on the rails where it's like, hey, you can't redeclare this variable. So like all these pitfalls or traps you would get in an object oriented language, ⁓ you don't have anymore. And so that was kind of an eye-opening experience from there. and then I think after Topgolf went to work at Teledoc, they're a big Ruby on Rails shop. And so they had some services and Elixir I got to work on there. ⁓ and then I worked for a company Blocket.

Jacob Luetzow: Mm-hmm.

Brice Ayres: out in ⁓ McKinney, Texas, ⁓ which ⁓ they've grown pretty large. I think they maybe they have a hundred people now in the company, but I was like the second or third employee there, I think. And it was a Elixir was our main main go to. ⁓ and so really kind of learned a lot there just doing Elixir full time for about two years.

Jacob Luetzow: That's awesome. I remember, yeah, like kind of same story. You know, I came from front end, JavaScript, Swift, object-oriented programming, just like all I got you so used to writing like real nasty. I got like so many code reviews when I was like a junior dev about just like ⁓ like the Christmas tree, the nested

Brice Ayres: Yeah.

Jacob Luetzow: You know, conditionals and loops and like all the nasty code you can accidentally write with you, don't realize like when you're new, you're writing a function, you're like, Well, I need it to take in this and output this. And you don't really think about breaking it up into you know all the transforming steps. And I think like functional programming just kind of like forces you, you're like, you write a function just to do that one thing. So it almost forces you to write that clean code.

Brice Ayres: Exactly. Yeah, a hundred percent. And you just think about things very differently. Like I think ⁓ I had one of these Elixir books, like ⁓ a Dave Thomas book, I think it was probably Elixir, you know, one dot three or something like that. But ⁓ remember reading it and it had a fizz buzz problem in there. And, you know, it's a classic, you know, very simple interview quite you know, I don't even know if people would ask it in an interview, it's so simple, but watching the way it's solved with pattern matching in Elixir and reading that, and I was like, Whoa, this is a very different way.

Jacob Luetzow: Yeah.

Brice Ayres: Where it was basically using that like modulo ⁓ and then matching it on the ⁓ you know, if it's divisible by five or divisible by zero with no remainder, then it like it would pattern match into these different things and print fizz, buzz, or fizz buzz. And I was like, ⁓ that's a wild way to solve that problem that I would never have done in object oriented.

Jacob Luetzow: Yeah. Yeah. I ⁓ once like pattern matching clicked for me, like my life changed.

Brice Ayres: ⁓ yeah. It it just it's it just makes you think about problems very differently and then allows you to solve it in cleaner patterns, I think, in your code, right? ⁓ that's just not available.

Jacob Luetzow: So you kinda got your your feet wet working for other companies that already implemented Elixir. What like first of all, what is ⁓ fractional engineering? What how would you describe that to someone that

Brice Ayres: Yeah, I guess some people call it fractional engineering or fractional CTO or ⁓ in my case it's some of these companies that need engineering help, ⁓ but maybe they're you know, they don't have a team or a large team, ⁓ and they need someone to come in and kind of really own a slice, ⁓ at least to get ⁓ started and then maybe they could transition it to a team. ⁓ but somebody kind of manages for them, but they don't necessarily need somebody full time. They don't need somebody forty hours a week. Maybe they just need somebody ten hours a week.

Jacob Luetzow: Mm-hmm.

Brice Ayres: ⁓ and espe you know, historically that was kind of the case, but now it's interesting with you know the age of AI, ⁓ you can almost be a fractional engineering team with one person, right? ⁓ one or two people, you know, and so you can go into a company and yeah, I'll I'll give you ten hours a week, but I I'm giving you forty hours of coverage essentially. ⁓ you know, or what would have been forty hours a week worth of of work.

Jacob Luetzow: Yeah. Yeah, I'm curious because essentially you're a consultancy and you know, you're ⁓ what has AI done for bringing in clients and things? Cause I've heard both stories now. I was just at ElixirConf. A lot of guys doing contract work, they're hurting, but then there are some guys that are having the best years of their lives.

Brice Ayres: Mm-hmm. Yeah. Yeah, that's and I think that's it's interesting. It's probably ⁓ and I think this is any business you're in, right? If you're a photographer, you're ⁓ you know, do some other kind of work, like ninety percent of what you do is probably marketing and sales, and ten percent of actually the thing you you have to do. And so a lot of work is just networking. And so like for my side, like I'm you know, I I take a ⁓ my friend, he goes he gets like laughs at me because he's like, Man, you'll always take a coffee. With somebody. Like, I'll always go talk to somebody, I'll always network. You just never know. And I've gotten so many referrals that way where it's like somebody knew somebody who knew somebody who's like, I need some help with software. And they'll name drop you because you're, you know, you took one conversation with them before. but yeah, it's it's an interesting cause I think right now it's I feel like it's a superpower where you can go into some of these companies that maybe haven't fully adopted AI yet, and you can come in and own a slice.

Jacob Luetzow: Mm-hmm.

Brice Ayres: And output what would normally take a team of ten people a year to do, and you could do it with one to two people in three months, you know, and ⁓

Jacob Luetzow: Yeah. It's pretty wild. And like I I c you know, everyone's like kind of doom and gloom. At least like I feel like in in the dev circles, people are either like really optimistic or doom and gloom about AI just like replacing us, right? I like tend to lean to the optimistic. I think it's a tool, right? And I think you can leverage it. We're always gonna need engineers. I don't care what you say, because you're gonna need some kind of mind behind orchestrating a plan, right? Like

Brice Ayres: Mm-hmm. Yeah.

Jacob Luetzow: Yeah, I think writing code, we're probably done doing that. But not putting everything together and solving problems, that's not done.

Brice Ayres: So do you you I guess where do you see the vibe coding? 'Cause right now it's the simple websites and stuff can be vibe coded, but it's like does that just keep going and going until the AI can now do advanced architecture and the full gambit?

Jacob Luetzow: Yeah. I think So and maybe it gets there, but maybe it doesn't. And I I think that's gonna always be the missing straw, right? Like like scaling and like infrastructure and what your tech stack is, those are complicated problems when you start growing a user base, right? Like you can build an app or a web app with any technology. It doesn't matter. If you only have a couple thousand users, you can make it with whatever you want and it's gonna run great, right?

Brice Ayres: Yeah.

Jacob Luetzow: And I think once you run into like scaling problems and growing problems, you're going someone put this a really good way ⁓ at ElixirConf. Like you're you you're basically like making all these, you know, two by four structures that are stick houses, right? But the second you have to build ⁓ a tower, you have to knock everything down and rebuild it so you can actually support having a fifty story building, right?

Brice Ayres: Mm-hmm. Yeah. Yeah.

Jacob Luetzow: So I think there's gonna be a lot. I think like the cool part is we're gonna see a lot of like solar, solo founders building really cool products that are gonna be like essentially really awesome MVPs. And then I think if they actually turn into a big product, we're still gonna need really smart engineers to, you know, scale it.

Brice Ayres: Yeah. Yeah, it's interesting. I'm seeing ⁓ jobs, ⁓ kind of these I wouldn't call them entry level jobs. I would call them more like ⁓ smaller scale engineering. So, you know, you go into like an office space and like we need help automating these spreadsheets, getting from here to this department to this department, or you know, which normally would been a normal engineering job. ⁓ but I I know personally a couple of people that have no engineering background or experience that are pretty much purely vibe coders. They maybe have dabbled or learned a little bit, but they now have engineering jobs. ⁓ right, and making six figures going in and working at these companies and taking a job that normally would have been, you know, s an engineer only that trained a long time do that.

Jacob Luetzow: Yeah. And I think that's I think that's really cool. And ⁓ the only thing is like you don't know what you don't know till you know. And I think like it's really cool that you know you can now like tr like transforming data or moving it and whatever you're doing and in-house tooling. I think vibe coding is gonna be awesome for that kind of thing. And then as long as you don't F up your data, things are good. Just make sure you do a backup, right?

Brice Ayres: Yeah. Mm-hmm. Yeah. Yeah, that's the problem. People forget to ⁓ patch all the security holes, make no mistakes. They forget to type that one line in ⁓ at the end.

Jacob Luetzow: Yeah, you gotta type that. Don't forget that prompt. Most important, your head of security. You're you're Michael Lubis at Paraxial IO and no vulnerabilities.

Brice Ayres: ⁓ huh. Yeah. And that's where I think people are getting in trouble is, you know, they're making these things that are you know, and I think fine for in house software or something like that. But the second you start putting stuff on the internet and people are like, Why is my my hosting bill fifty thousand dollars? And you're like, Well, 'cause it you wrote a background loop that just loops nonstop and does, you know. ⁓ and so I think people are going through those those pains, but ⁓ it's pretty cool to see. It's just inspired a lot of people that weren't going to take that leap into engineering that are now dabbling into engineering.

Jacob Luetzow: Yeah. You can just experience pain a lot faster now. Cause I think like all that pain was still experienced before you could like vibe code something, right? Cause I remember when I first got into tech, I joined a startup, like an education startup. And one of our students, it was right when Pokemon Go came out. And he made an app called ⁓ Pokemon Radar. But, you know, it was like

Brice Ayres: Yeah. A slow burn. Mm-hmm.

Jacob Luetzow: I don't even I'm just I don't actually remember the details, but it was hosted on I think like AWS and he might have used like Firebase and all these things that are great for like kind of hacking things together. But it went viral overnight. And he ended up like waking up to like a five hundred thousand dollar like AWS bill. And what really sucks is he sold his company like overnight, but it Was barely enough to pay his bills that he owed.

Brice Ayres: ⁓ wow. Yeah. I think you'll start seeing a lot more stories like that where I mean, just all kinds of I mean problems we've just been dealing with in engineering that don't come to the surface a lot of people don't maybe don't talk about outside of, you know, kind of the inner engineering circles, you know, mitigating DDoS attacks and things like that. ⁓ which is it's pretty pretty interesting. But I think really cool that you have new call it new blood coming into the industry, you know, and kind of pushing in

Jacob Luetzow: Yeah.

Brice Ayres: But I know some engineers that are kind of frustrated because you know it's taking some of those jobs away. But I feel like like you said, it's engineering still needs to be there. And I think you always need somebody to translate, ⁓ I would say translate crappy business requirements into code. ⁓ because like that's that's really the skill. It's never the writing the code by hand has never been the thing you're doing or selling, the skill set, right? It's like it's just a tool we need to solve the problems we have.

Jacob Luetzow: Yes. Right.

Brice Ayres: ⁓ and I think some people forget that and they they hold on to that very tightly, right? And it's like

Jacob Luetzow: I know a lot of people have a hard time letting go of that. I honestly I like programming, but I it I like solving the problem and having it the end solution more than writing the code to get there.

Brice Ayres: Yeah. Yeah. I mean, there's something nostalgic about, you know, some artisanal handcrafted code. But but yeah, really it was always solving problems and it was that was the tool you used. That was your hammer to build the house or right. ⁓ but ⁓ yeah, it's it's just an interesting time we live in now, translating those requirements. ⁓ 'cause a lot of people think they can solve some of these requirements, but have never built a product and ⁓ you know, product engineering I think is where

Jacob Luetzow: For sure. Mm-hmm. Yeah.

Brice Ayres: ⁓ those engineers are really winning right now. And you talk about like some people are having a hard time, some are doing great. And I think those ones that can see the problems ⁓ and convert those requirements into like a solution ⁓ are really going to be the winners coming out the other side. Whether you know the AI is writing the code or you're writing by hand or whatever it may be, just being able to help somebody understand their problem and figure out what solution they should actually do or build is going to be be the the s the secret sauce there.

Jacob Luetzow: Mm-hmm. Yeah. Yeah, I agree. I mean, AI's made like I was never a good web dev, like front end engineer. I just I don't know. I struggle. Like CSS, Tailwind, divs, the whole thing. Everything about it, I suck at. But now I can write I can create really good products. ⁓ because I'm like I'm not stuck. I'm not just, hey, I wrote the back end code. Now I need a front end dev to make the front end real pretty for me. Now I have Claude do it all for me.

Brice Ayres: Yeah.

Jacob Luetzow: Hm.

Brice Ayres: Well, CSS was always like a arch nemesis of engineers because it's like you just purely have to memorize some of it. It doesn't follow like engineering patterns, you know, and so you're like you have to remember this quirk about Chrome and this quirk about Safari or whatever it might be, right?

Jacob Luetzow: Yeah. Yeah. I can't memorize anything anymore. I don't like my brain my brain does not work. I don't know what has happened, but yeah, it's just gone.

Brice Ayres: Yeah, that the ⁓ kind of that loss of like using that muscle to code, right? Like I I tried to write something by hand the other day and I like, ⁓ man, I'm like really bad now. It's like

Jacob Luetzow: I think about that a lot and I'm not sure if I should be con like, do you need to worry about that? Like cause w the other day one of my buddies was like, Well, what if it is a bubble and tomorrow we don't have it? I was like, Well, I guess like I would just start writing code again and it'd probably come back pretty quick.

Brice Ayres: I think so. I think it'd be like riding a bicycle. Like you'd be like, ⁓ okay, I remember this. And you would just have to go through the slug of I mean, anytime you learned a new language or learned a new framework, you had to go read the docs and you'd be like, Okay, and then after a couple of weeks you you would get used to it. ⁓ but I ⁓ that's what I was gonna say. Yeah, it's it's interesting now in this age of AI, ⁓ because from my perspective, you know, I I started out doing software engineering, like I was writing code myself.

Jacob Luetzow: Right. Yeah.

Brice Ayres: ⁓ then I started managing teams and then I started running like engineering organizations. and as the higher you go, the less you get to see the code, but you're responsible for the outcomes that come out the other side, right? And so one thing I've been, you know, as I've been talking to different people and giving them my perspective is ⁓ if you think more like, you know, with AI, think more like an engineering manager or even, you know, a director or VP or, you know, CTO, you're not gonna get to see the code. But you have to manage the outcomes that come outside out of that process. And so as long as it meets the minimum requirements of like it runs in a certain amount of time, it performs a certain way, it does the thing you say it's going to do. Do you care? And I think this is the like big question. Do we care what the code looks like inside, right? Like the internals. ⁓ and that's a hard one. You're like, yeah, we do, because we, you know, we've been trained to care about code quality for so long. But

Jacob Luetzow: Mm-hmm. Right. Mm-hmm.

Brice Ayres: Now when you look at it through a lens of, you know, you're now call it upper management and you got a team of agents you're running and managing. ⁓ do you care what the code looks like?

Jacob Luetzow: But if you if you're benchmarking performance and things doesn't really matter what it looks like. Yeah.

Brice Ayres: Yeah, does it matter? Yeah. It's like the tree falls in the forest. Does it make a sound? Does it matter? You know? ⁓ I don't know. That's that's the hard question I think a lot of people are struggling with. And, you know, for me, I've I've kind of been okay with that because I've been kind of running these engineering organizations where I haven't been able to look at the code. And the only thing I get to do is I get to ask the the developers about the features. And I say, Hey, did you run these tests? Did you how much what's your test coverage? What's your did you think of this use case?

Jacob Luetzow: Yeah.

Brice Ayres: What happens in the future when the business decides to go this direction? How much of a lift is it going to be to pivot that way? And those are all the same like conversations I have with my agents, right? That are thinking about all those things. And so it's it's almost mimicking those exact engineering processes that you would at an engineering org, but now you're having those conversations with agents rather than, you know, ⁓ you know, an engineer on your team in a stand up, you're having your little mini stand-ups for your agents.

Jacob Luetzow: Yeah. Dude, Chris Chris McCord does not review any live view code anymore. He's like, it has the established patterns, and as long as it looks good at the end product, it's probably written correctly. And that's wild. Yeah. Isn't that crazy?

Brice Ayres: ⁓ wow, yeah. That's a that's a big endorsement there. Yeah. It's wild. Yeah, I've seen a lot of hot topics on X or Twit you know, formerly Twitter, ⁓ recently about like you should review every line of code and that that seems to be a very hot hot topic right now. But you know, you're seeing ⁓ a lot of people like, you know, Chris McCord's a great example, like he was he's an incredible engineer, and for him to say, I'm not gonna go read that code, ⁓

Jacob Luetzow: No. Mm-hmm.

Brice Ayres: And ⁓ I think that's where you put the guardrails in place. You put, you know, use framework like I try to use frameworks when I can or establish a framework or pattern. And then once you do, from there it should mostly follow those patterns. And you can have a, you know, a skill or ⁓ a background agent or something that's checking to make sure it's always using your patterns, right? And so it's your quality control.

Jacob Luetzow: Yeah. Yeah. I will say like with all my projects I have going, I'll I use Ash framework and Ash and LLMs, it's just amazing. Like most of the time, like I review my code, but I'm getting less and less like detailed about it because you just once you start seeing as long as the LLM has established patterns and you make sure they're there, you can pretty much assume all your resources or whatever it was just working on are correct.

Brice Ayres: Yeah. That's I've been a huge fan of ⁓ the Ash framework for years now. But I always had a hard time adopting it pre AI, ⁓ because it's it's a it's a lot to kind of take in. It's a you know, the DSL that it's basically a new language you almost have to learn. ⁓ and it's I the way it's structured to me makes a ton of sense 'cause you and I think ⁓ I've heard ⁓ Zach Zach Daniel ⁓ talk about it where he's like basically at every company you reinvent these same patterns every time. You end up like once you get enough code, you're like, Okay, well I I should have

Jacob Luetzow: It's a new language, you know. Yeah.

Brice Ayres: you know, this type of pattern or this folder structure or this. And you just took kind of an opinionated approach. in it but it was hard to learn and wrap your head around it. And a lot of times it was like a one line change will give you th so much functionality. ⁓ an AI that

Jacob Luetzow: Mm-hmm. W what's wild is a one line change gives you like multi tenancy. Like how cool is that?

Brice Ayres: Yeah, yeah. Yeah. It's it's unbelievable how much you can do with just so little code. And so that was always ⁓ kind of their secret sauce of their, you know, consultancy was they could move quick because they had this framework they could use. ⁓ and now, you know, with AI, I've been probably using AI coding for a couple of years now. And so initially it was didn't do great with Ash, but now that it's had more data to train on and the LMs are a little better.

Jacob Luetzow: S

Brice Ayres: It does great with it and it actually keeps it on the rails 'cause it's an established patterns and frameworks out of the box that get you going.

Jacob Luetzow: Mm-hmm. Yeah, I don't know why I'm blanking right now, but what is the the library that Zach made that like syncs all the library? Like if you have a basically like a clawed MD file, an agent MD file, it syncs them all. I'm blanking on it. But it's my favorite it's my favorite library and I use it and it's like a life it's a game changer for everything.

Brice Ayres: yeah. ⁓ yeah, I know exactly what you're talking about. ⁓ Yeah, what is that? ⁓ I know exactly what you're talking about. I use all the time. ⁓ and so it'll it'll pull all the especially the L L ⁓ you know, information in.

Jacob Luetzow: So I'm actually creating a podcast tool right now. I haven't really told many people about it. And I wish I had it running right now because I actually kind of just got like a proof of concept kind of working yesterday. I'm making like a a podcast producer tool where it's like a an app overlay. So like if you and I are talking, it listens, it right it it records our transcript. And then like if I pop them and be like, hey, can you fact check that? It'll go back and fact check it and then

Brice Ayres: ⁓ nice.

Jacob Luetzow: pop it up on the screen. But like, how cool would it be if I could just be like, go find this. What are we talking about? And it would pop up on the screen.

Brice Ayres: That's a great idea. 'Cause I've seen people do it like post ⁓ after you record, it'll go and then find all the links and build your link pages and stuff. But ⁓ to do it in real time would be great.

Jacob Luetzow: Yeah. Yeah, yeah. And I do that stuff after and I was like, how cool would it be have to have like the Joe Rogan Jamie, but on a budget?

Brice Ayres: Yeah, Jamie pull that up for me. Yeah. That's exactly right.

Jacob Luetzow: Hold that up. And then, like, what I'm working on too is ⁓ it helps you plan your podcast episodes. So, like I was actually messing around with you ⁓ because I was like, let's just see. So, like, I'll create my guest profile and then I attach my guest to an episode, right? So I have episode notes and a guest profile, and I can have it go and research all the links that I added to you. So, like I can add your LinkedIn, I could add your X.

Brice Ayres: ⁓ cool.

Jacob Luetzow: And it'll kind of pull together a summary of who you are. And so I feel like it'll be a really fun tool. We'll see if I ever like if it ever becomes like a production app, but it'll at least be a tool I use.

Brice Ayres: That's awesome. Yeah, that's You gonna call it young Jamie?

Jacob Luetzow: ⁓ I'm calling it right now it's called off mic. But but we'll see. We'll see what happens. Young Jamie. I might get sued if I named it that, you know.

Brice Ayres: Off mic, okay. Yeah. Just gotta get him to endorse it and then like use it on the podcast or, you know, Joe Rogan's podcast. Yeah. That's how you get it.

Jacob Luetzow: That would be awesome actually. But like what's really cool too is like you can like I ha you can like set up questions and segments of the podcast. So it could be like it'll pop up, you know, suggested questions you should ask as you're talking. Yeah, it'll be cool. We'll see how it works.

Brice Ayres: That's nice. That's a really good idea. Yeah, because I I I feel like that'd be useful too for ⁓ like producing YouTube, like long form YouTube videos. I don't know if you ever had this struggle when you were making your content, but for me when I first started, initially I was like, okay, I'm gonna write a complete script of everything I should say. And then like it felt like I was reading it, and I was like, Okay, well now I need to do bullet points and I'd do bullet points, but sometimes I'd forget where I was and I'd go off topic, and then I would touch on something later that's now covered and

Jacob Luetzow: Mm-hmm. Yeah.

Brice Ayres: So in your head you're trying to like reread your bullet points to figure out where you should jump to next. You're like

Jacob Luetzow: Yeah. I went through the same progression. And you can probably tell from my earlier videos, they probably do seem like I'm reading. And then you know what I ended up doing is if I was like going through a project, I would always write the project beforehand, right? Obviously, make sure it works. And then I would just throw comments to make sure I wanted to like if I wanted to mention something specific, I would just have commented code right there. And then as far as the code goes, I would just talk about the code as I wrote it. But yeah.

Brice Ayres: Did the same exact thing. Yeah.

Jacob Luetzow: I just ended up using my code editor.

Brice Ayres: That's that's so funny. Yeah. That my problem is I then I'd I'd pre code it and then try to ⁓ like I basically had two screens open. So one I'm like copying the code in, but then I couldn't I wouldn't necessarily do it in the logical order that you would actually s write the code, you know, 'cause I'm like trying to do file by file or something and sure enough people in the comments like

Jacob Luetzow: Yeah. Right. ⁓ well, you know how I do that? I just used git. So I'd be like git, my branch would be video one, two, three, four, and then and then basically I'm talking about everything that's now new in that branch.

Brice Ayres: Okay. ⁓ gotcha. Yeah. I was copying it all over by hand and so like sometimes I'd do this file and it was like, How did you know to put that function name there, you know, before you declared it in the other file? You're like, ⁓ yeah. It's you like get some funny comments. Yeah.

Jacob Luetzow: Yeah. Well, I mean, come on, it's planned. So you're like, well, you didn't write this function yet, but we're about to.

Brice Ayres: Yeah. 'Cause the first couple of videos I did like on the fly and it was rough. You know, it was like took me an hour to write a very simple like React app or something by hand or whatever I was working on.

Jacob Luetzow: D making content is a lot of work. It takes a lot of time. It kind of it burned me out. And that's why I like doing like the podcast because I don't really have to plan much for this because at least the the the format I do, I just enjoy kind of like chatting. Cause I feel like you get more out of people if you talk about their experience, but like naturally. And yeah.

Brice Ayres: Yeah. Yeah. It's definitely a lot easier with two people, like a one person show. You gotta bring all the energy, you know, and that's one thing I learned was ⁓ just being on camera. I thought I was being very like over the top energy, and I'd go back and watch the video and I was like, ⁓ that's kinda boring to watch, you know, like you gotta really amp it up. Exactly.

Jacob Luetzow: Pharrell. So many coding tutorials just suck because of that. And it's like nothing against the people making them. It's like one, coding is already is kind of a dry subject. And then ⁓ it gets really easy to fall into like a monotone.

Brice Ayres: Mm-hmm. That's why those guys like Theo or the Primogen, like they ⁓ crush 'cause they're very entertaining, you know, whether you agree or disagree with their hot takes, you know, but they're they're big personalities on camera, for sure.

Jacob Luetzow: Yeah. Yeah, yeah. Yep. Yeah, it's definitely an art form, that's for sure. ⁓ I'm curious to go like you kind of made a comment about having leadership roles and you think it's made it easier for you to adopt AI. Do you want to like dive into that a little bit more? Do you like because the way I look at it, like I agree with that statement because like building Kill Switch, ⁓ which is my little like document storage with dead man switches, it's been a fun project and

Brice Ayres: Yeah.

Jacob Luetzow: The way I have like AI set up right now, it's like, you know, I'm the solo founder, but I'm doing all these things. Like I'm using Grokbot for a lot of automation, and then I'm using Claude Code and bead work, which is an awesome context library if you haven't checked it out. ⁓ but I'm using all these things and like you feel more like the orchestrator that you just like you're checking in on things to make sure everything's still happening. But what is like

Brice Ayres: Yeah.

Jacob Luetzow: Did you adopt AI pretty quickly and kind of accept that you weren't writing code anymore?

Brice Ayres: Yeah, I think for me, I I mean initially it was I was I just like shipping stuff and so for me initially I got hooked on just like I can ship more stuff. Even when it wasn't great, it was just amazing how much it was producing, or just like write me some tests. ⁓ like all the stuff you hated writing by hand, but ⁓ yeah.

Jacob Luetzow: Yeah. Well, tests are like free now. So like if you don't have test coverage in your code bases, shame on you.

Brice Ayres: You have to actively try to tell it not to write tests, you know. And sometimes I will for like when I'm doing a test project or something. I'm like, I don't waste your time burning tokens writing tests because this is not this is throwaway code or ⁓ but it'll it'll try to add tests and make it super, you know, test coverage ⁓ friendly. But yeah, it from a from my perspective, it was really easy to kind of see the patterns emerge because it was all the patterns we've, you know, in engineering have developed over decades of experience, right? Like

Jacob Luetzow: Yeah. Ha ha.

Brice Ayres: Figuring out stand ups and scrums and you know, whatever you there's a bunch of different methodologies, but if you just take those same principles and you just treat the agents like as if they were, you know, engineers on your team, right? Like put somebody in charge of them. So maybe there's an agent that's in charge as the orchestrator, right? For that team of, you know, sub agents. ⁓ 'cause you look at like ⁓ I think at one point I had maybe thirty people under my like engineering org. And and so I had a mobile team. So I had four or five guys on mobile. ⁓ I had 'cause they're building some like native apps and stuff. You have a couple of back engineers, you have a couple of front end engineers, you have a DevOps, you know, small DevOps team, ⁓ then you have designers and product people and so you gotta I mean all these moving parts and pieces and you can't look and review all the code yourself. And so I'd have to delegate, you know, you're in charge of this team, you're the lead, you're the lead, you're the lead of this team. And that's like setting an orchestrator agent. ⁓ and so I commonly like I'll use ⁓ Claude's kind of my daily driver right now. But I'll use like Fable five point one as the orchestrator reviewer and have it spawn Opus five sub agents to do the coding implementation, right? And so I'll give it a task and have it go. And then I'm even, you know, in some of these pipelines I've built before, ⁓ I still haven't quite mastered this, or at least maybe I'm not doing it right, but I've used ⁓

Jacob Luetzow: Mm-hmm.

Brice Ayres: linear before and so I'd have different ⁓ essentially tags. So like a QA tag, a release engineer tag, a ⁓ back end engineer, front end engineer. And I gave ⁓ basically like a whole, you know, skill essentially for each one of those. And then I would use linear where ⁓ I would go back and forth with a product manager agent, just like I would in a normal engineering org. I'd work with product, understand their needs. Really like I'm kind of dictating, you know, kind of playing both sides a little bit, but

Jacob Luetzow: Mm-hmm.

Brice Ayres: Based on that, it would generate me a bunch of tickets in linear, and then those tickets would get picked up by there's a background agent that's delegating work, like the scrum master, you know, organizing and delegating. And then it would fall into the different columns on the, you know, different swim lanes for each ⁓ person, like the designer. So if this had tagged as a designer, it's like, ⁓ this ticket blocks this one. And so I'd built a whole system that was kind of mimicking exactly that engineering process you would see at an organization. But just replacing the people with agents essentially. ⁓ and it was really cool. And I think the the most underrated ⁓ part of that system, ⁓ surprisingly, was I had a documentation ⁓ writer, like a technical doc writer. And so when the ticket made it all the way to the end of the pipeline, the ⁓ just before it got released, a technical document writer would go and review that, review existing documentation and see if there's any notes we need to put in there.

Jacob Luetzow: Nice.

Brice Ayres: So that future agents could have a reference of why we made that decision. ⁓ and so and sometimes it was small, they'd add two lines of text changes, you know, based on the ticket that came out the other side. And then ⁓ I had one release engineer. ⁓ so you'd like ten work trees going on all at the same time, but you need to get all that code back together. And so I had ⁓ basically the way I solved it was having one release engineer that was in charge of just going through and merging the code together so you didn't get crazy merge conflicts at the end, which is kind of

Jacob Luetzow: Mm.

Brice Ayres: Kinda similar how you would do it in an engineering org, right? Like you don't want people just merging pull requests.

Jacob Luetzow: I'm not brave enough to let my agents do my releases or merges yet. No.

Brice Ayres: ⁓ no. Yeah. I've I've gotten even ⁓ like crazier with it sometimes too. Like I run I'll like Kubernetes in production and then I'll ha give the agent like a a read only so he can go in and read what's happening in the cluster and the logs and like goes into production and like will diagnose issues that are happening in real time. I'd be like, Yeah, and like that's been and I've seen some projects where people are actually installing an agent that lives full time in the background on the cluster.

Jacob Luetzow: Okay. That's pretty cool.

Brice Ayres: It will self heal. I'm not brave enough to go that far yet, but it's coming.

Jacob Luetzow: It's yeah. I mean it's definitely coming. I don't know. We'll see. I'm okay with like read only on prod, but everything else kinda scares me.

Brice Ayres: Yeah. That's it's a little wild to be like, I'm gonna give this agent full access to my entire system, you know, where it's behind ⁓ essentially the wall the the curtain of security, you know, where it's like and it's gonna go and make changes. So I haven't done that yet, but I've seen people doing it where it they call it like self healing infrastructure. So if your database disk gets full, it'll automatically just go and provision more, ⁓ write the request and change.

Jacob Luetzow: Yeah. What if it like accidentally just drops your database or does something dangerous though? Like that's where I would get scared.

Brice Ayres: That's true. ⁓ and I guess that's where maybe you put other processes in place or you need consensus maybe from multiple agents. could be like a strategy. So I've done some stuff like that before. ⁓ this is probably more along. Like I was making a I was trying to make a trading pot, ⁓ using like polymarket. I don't know if you've ever gone down this like rabbit hole of ⁓

Jacob Luetzow: Mm. I have thought about it and I've never done it. I've had a bunch of buddies make trading bots, but yeah.

Brice Ayres: ⁓ but I was trying to figure out I I wanted to you on some of these ⁓ markets like Kalchi or Polymarket, you can see the top fifty, you know, whales essentially that are making trades. And ⁓ my theory was I'll watch what trades they're making. ⁓ kind of like your the Pelosi tracker or something, you know, like follow these whales that must have some insider information and then like try to go ahead and ⁓ replicate what they're doing.

Jacob Luetzow: Yeah. There you go, yeah.

Brice Ayres: But I had basically I didn't want to just blindly trust and like make these trades. And so I created like a consensus pool of agents. So it was like three to five agents. And ⁓ they all had different personalities. So like one of them was looking at like, you know, how much money could you actually make from this trade? What's your and then one's like concerned about losses, maybe one's concerned about what's the realistic probability that this trade will happen based on volumes and ⁓ trading ⁓ trading volumes happening currently. And ⁓ But it it opened up some interesting things because I couldn't get my agents to make a trade. And I was like, why are they not making a trade? Like the consensus would always come out like only one or two of them out of the three would approve the trade, you know. And I was like, why can't I get a three for three consensus? Like, what's going on? And some of the ⁓ this is like maybe some insider information, maybe somebody out there could use to make some money on polymarket. But ⁓ I would find that some of these people were buying ⁓

Jacob Luetzow: Ha ha ha.

Brice Ayres: Trades and these are like on sports or like you know ⁓ different events that haven't happened and they're not gonna happen for up to a year out. So it's buying a position on a a sports game that's like a year out. You don't even know who's on the team in this case. And I'm like, why are and so like my agents were declaring that as a risky trade because they're like the volume that's happening, there's almost nobody trading this right now. And so I think what they were doing. Was they were buying trades essentially really far out, but like the ask maybe like was, you know, ⁓ 75 cents. They would bid low, you know, at fifty-five cents, fifty-nine cents. And it seemed like all their trades were trying to be in the middle between like forty five to fifty five cents. are you familiar with how it works? How it's like zero like a penny to ⁓ ninety-nine cents, basically. ⁓ so it's like a it's like a two sided market, and so basically there's like

Jacob Luetzow: No, no. Not really.

Brice Ayres: Kind this wave of like what the current ask is. And so if you hold it all the way to the end, so if you buy a position at call it seventy five cents, ⁓ saying the Dodgers are gonna win this baseball game, you can sell it before the end of the game based on predictions like somebody hits a a grand slam, like maybe it spikes for a second, you could offload it at eighty cents and you made f five cents per share. ⁓ or you could hold it to the very end. If they win the game, then you get a dollar, but it costs you seventy five cents to buy in, so you made twenty five cents.

Jacob Luetzow: Mm. ⁓ I see. People are everyone's just gonna be ⁓ just betting on everything, on the silliest of things too, aren't they?

Brice Ayres: Yeah. So that's what I think these whales are doing is just buying really far out. So there's no volume. And so whoever owns some shares are like willing to just like offload it to get out of their position. So they buy it at extreme discount. They try to buy it around the middle. ⁓ so that way in the baseball game or whatever sporting event it might be, at one point somebody's one c team's gonna go up or hit get a great base hit or something, you know, and all of a sudden you're gonna have ⁓ a spike and then they'll offload it and get out of their position before the game resolves.

Jacob Luetzow: Yeah.

Brice Ayres: ⁓ but I don't know. So my my bots are just like, This is a risky trade. There's no volume happening right now. This is events like a year out. And I'm like, What is happening? And it was it was a pattern I kept seeing with all these whales.

Jacob Luetzow: That's yeah, that's funny. So you had to readjust how who you were looking at then at that point.

Brice Ayres: Yeah. ⁓ and so you're trying to figure out like well Yeah. You trying to figure out like what are you trying to figure out or what are you trying to look at or what are you trying to what are your data points there. But with AI it makes it possible to do these like fun little side projects that I would have never invested time into otherwise, right? It's something I can just keep opening another terminal and just kind of chip away at it while I'm working on something else.

Jacob Luetzow: Yeah. Yeah, that's fun. I mean that's like essentially gonna be my podcast tool, right? Like at least I can I can just like vibe code it and test it when I do podcasts and we'll see how it goes.

Brice Ayres: Mm-hmm. And kill kill switch is your main, right? And then you may have like the podcast one. Yeah.

Jacob Luetzow: Yeah. Kill switch is like a product I'm very, very proud of. I like it's turning out to be a really cool thing. And like honestly, without AI, I wouldn't have been able to do it as a single person.

Brice Ayres: It'd been really tough, right?

Jacob Luetzow: It would have taken a lot longer, at least, you know.

Brice Ayres: Well, I saw recently I've been getting more into the marketing side of like building an app is like one thing and that's the easy part now, but it's like how you get distribution and I saw you had some marketing successes this past week.

Jacob Luetzow: Built so it's funny, we're in we're in a bubble, you have to remember, right? We think AI like no one's gonna need us to build things anymore. But leave a software developer circle, no one can use AI to build an app, or not nearly as easily, you know? But yeah.

Brice Ayres: Yep. But y it seems like you cracked the code this past week or two on some marketing efforts for Kill Switch.

Jacob Luetzow: Dude, I mean, so I suck at marketing. I yeah, and I might I should probably make a video on it because I think it's like super beneficial. Cause I think engineers in general, like building the product's the easiest part, at least you know, for a technical person. And I don't have it, I don't I still don't have it like nailed down amazing. Like I'm not gonna like guarantee you're gonna like make a million dollars.

Brice Ayres: Did you do anything special there? Yeah. Yeah.

Jacob Luetzow: But I'm getting signups and constant trials and sign up like I before it was like just stagnant. I would like get, you know, a paying user once a month. It was like really slow. It's kind of like a little like disheartening, you know? And then I started using Grokbot. And I was like, what is this Grokbot thing? Because it like hit X hard when it first got released. So I was like, I'll check it out.

Brice Ayres: Mm-hmm. Yeah. It's pretty cool. Yeah.

Jacob Luetzow: It's really cool. So I have a whole team of grok bots. Like I have a research and content bot that will just like one, it knows kill switch. It knows everything about it, all my blog posts, everything, and it will draft new blog posts. ⁓ and then that will always like make my social media copy, all of that stuff. And then I have a video editor who who Knows how to use the JavaScript ⁓ remotion framework. And yeah, it's really cool. Code-based, basically all JavaScript video editor. So my LL or my Grokbot edits videos for me. ⁓ I gave it access to like my stock image and video account, so it will go and find stock images, the audio, whatever it needs to.

Brice Ayres: Most.

Jacob Luetzow: And it's been making my sixteen to thirty second shorts every single day. And then I have a social media scheduler. Yeah. I have a social media media scheduler. And I was like, hey, just always make sure the next three days are scheduled. And so it does that. And then the video and they're all in a group chat. So the video editor knows he always needs three days of videos. And like it's cool, dude.

Brice Ayres: Really? Does it does it automatically post for you on social media too? Do you run you don't run any snags with like 'cause I know some of these platforms try to prevent bots from posting, but

Jacob Luetzow: Yeah, I give it. It's been so there are like I'll get logged out on a lot of things and then I I have to just re-sign in a lot of times, but for the most part, it's like pretty automated. It's kind of just like I'm the overseer, so I'm in the group chat and be like, Hey, ⁓ I couldn't post Instagram because I need to be logged in. So it'll, you know, it'll throw me a notification. So I re auth it and it's it's pretty cool.

Brice Ayres: Okay. So you reauthorize permission? Interesting. Yeah, that's what that was the thing I was trying to figure out if it could actually do that. 'Cause if it can do that, then you start to build these social media accounts or followings and profiles, right? That ⁓ would be beneficial.

Jacob Luetzow: Mm-hmm. And then like what's really cool too is I have a customer service bot and that monitors my hello kill switch email like a couple like four times a day. And it so it kind of like made a bunch of buckets. So if it's like a product or you know just sales crap, it goes in a bucket. If it's customer service related, it will draft a reply.

Brice Ayres: ⁓ cool.

Jacob Luetzow: Bump me and be hey, is this okay to send? And I'll say yes or no, right? And so that's been pretty cool. What's really cool too is I created an admin account on Kill Switch that Grok has access to. And so it will it has a read-only access to my Stripe as well. So like people that sign up but don't start a trial.

Brice Ayres: Yeah.

Jacob Luetzow: I have it like drafting all these emails and it does cold outreach and like it does all this retention stuff. It's really cool.

Brice Ayres: ⁓ that's cool. Yeah, it's only you've you basically replaced all those like drip marketing campaign platforms, right? ⁓ which you normally have to pay a hundred bucks a month for or something like that.

Jacob Luetzow: Yeah. Well, and like inside a kill switch, once you sign up, I have my own drip campaigns coming out of my own platform now. But then like I'm having Grockbot do like the more sensitive touches as long as I approve them. So it's like, hey, we saw you sign up, but you didn't start a trial. Like, you know, if they're if you're having an issue, just you know, reach out, more than happy to help. And like it's I don't know, it's been really awesome.

Brice Ayres: Yeah. Yeah, that's what I'm ⁓ that's my next kind of direction to go is working on kind of like ⁓ I don't know what to call like enterprise marketing tools or bots or agents, like so trying to get distribution or trying to like for you know fractional engineering, right? Go in and start emailing companies. ⁓ and part of like I was doing for a while, I built I built a website ⁓ platform. So you could you could basically stand up your own website ⁓ real easy. ⁓ and I build I build you a custom site, but really like what I was giving you was this really high performance site that would score like lighthouse scores would get you like almost a hundreds across the board. And so my thought process was just email people, like go find websites that were scoring not well. And this was like my sales pitch, like and I was doing really well, ⁓ just for small, you know, build a thousand bucks here, twelve hundred bucks here, and then I put ⁓ on a monthly retainer similar to like

Jacob Luetzow: Mm-hmm.

Brice Ayres: What is it like Wix or Squarespace or something, 20, 30 bucks a month, depending if you prepay for the year or whatever it might be. and so just recurring revenue that way. But the easiest sales pitch was go find these people that had these sites that had terrible lighthouse scores, and they'd be like, Hey, here's some example sites I've built recently that have like near perfect lighthouse scores. And ⁓ there's a new ⁓ I don't know if you've been in the light looked at lighthouse scores recently, ⁓ but there's a new ⁓ category called agentic browsing.

Jacob Luetzow: Mm-hmm. No, but I have heard that. I there's a website called is is agent ready.com or something like that. And I've I use that. And that's nice because it like everything you fail, it gives you an LLM prompt to like add to like fix, which is kind of cool. But I haven't checked out Lighthouse in a while. What's funny is I built a system similar. I actually built like a whole like it a s like a basic

Brice Ayres: Mm-hmm. ⁓ nice.

Jacob Luetzow: Like if you have a a business website with like just a blog or some simple things, I built a whole web hosting platform in Elixir that I could just like spin people up off of like templates and things. It was pretty cool.

Brice Ayres: Yeah. ⁓ did you? Yeah. Basically what I got. It's like a I built a skill that knows how to build these websites a certain way, publishes them to the platform. And instead of doing it like the WordPress approach where like you have like a database to go fetch your blog posts and stuff, I have a ⁓ portal where they can go write the customers can go write their own blog posts and upload photos and do all that stuff. But when they do it and they save it, it rebuilds it as a static build and then pushes it out to the CDN. ⁓ and so it's really fat, it's like a lightning fast build.

Jacob Luetzow: Mm. That's awesome, yeah.

Brice Ayres: You know, and so it's the performance is just off the charts compared to a WordPress or something like that, because there's no database call. It's ⁓ statically built. ⁓ it's a little more expensive, like depending if your people are like publishing like twenty times a day, that might be bad. But ⁓ I think they'd have to it'd have to be like an astronomical amount of blog posts they'd have to publish. Yeah. ⁓ you've hit your ten a day, it's gonna be a hundred dollars a month now if you want to write fifteen a day. Yeah. I haven't hit that problem yet. Yeah.

Jacob Luetzow: Mm. Just right limit them. Exactly. Like calm down, br calm down, bro.

Brice Ayres: But you but like I think your Grokbot thing's great, like where, you know, I've thought about now ⁓ using that ⁓ have it use the Lighthouse Chrome extension, basically go crawl websites, find ones that score really poorly, like in my area, and then just screenshot the Lighthouse scores, put in an email, be like, Hey, like we help improve these scores, like you know

Jacob Luetzow: Yeah. Yep. What's really cool is you can do like you can just do like a screen capture and you can teach grok the grok bot anything you want too. So like if it doesn't know how to figure out on its own, you can go and like record your screen and click through things and be like, do that, dude.

Brice Ayres: Yeah. Super super useful. But yeah, you could use it to find ⁓ broken websites like forms that are broken or whatever. Like so just like, hey, not even you know, don't let me rebuild your site, I'll just fix your existing site if you can give me access, you know, to it. And now with AI that stuff's really easy to change, but for some of these people who are you know, they're not super technical or it was set up one time a long time ago, they don't know how to go change that stuff and so you can come in and and leverage ⁓ instead of being doom and gloom with AI, use it to your advantage to make changes at lightning speed, right?

Jacob Luetzow: Yeah. Yeah. Yeah. What's the coolest thing that I've been seeing is I've been getting conversions from Chat GPT and Claude and like all these LLMs and like that's cool. So people are finding kill switch through LLMs now.

Brice Ayres: Yeah, and there's a whole I don't know if you've gotten into it. ⁓ I'll have to see if I can find it afterwards and send it to you, but I've seen ⁓ people talking about how easy it is to rank. Almost like back in the day, SEO for Google used to be able to just type all the keywords in your webpage and it would just show up on the front page. ⁓ but people are kind of doing that with the LLMs now. And so one guy I saw, ⁓ he basically went to Reddit, I think it was. So there's like five to ten places most of these LLMs get their content from. And so he Yeah, I would love to.

Jacob Luetzow: Yeah, yeah. Dude, I hate Reddit. We'll talk about it in a bit.

Brice Ayres: ⁓ but he goes to Reddit and he ⁓ basically goes and just makes a post and says like, What's the best ⁓ you know, company in this city that does this or whatever? And then he goes and he answers his own question with like another account and says, ⁓ the best company, yeah, the best company for this is this. And then a day later or two days later, it's indexed in the LLM. And so when he goes to the LM and types in like, what's the best company for this? ⁓ it's now showing up.

Jacob Luetzow: ⁓ that's hilarious.

Brice Ayres: Using that as a reference, ⁓ you know, is now the declarative ⁓ best company at XYZ, right?

Jacob Luetzow: Yeah, I don't know what it is about Reddit and maybe I just don't know how to use it, but I get my account blocked all the time. Like I won't even talk about Kill Switch. I just have an ac I had an account named KillSwitch and I was like, well, if I go in there and ask answer things nicely, if anyone clicks on my profile, they'll get to kill switch. And I don't know if that's not allowed or what, but my account always gets blocked.

Brice Ayres: Ha ha. Yeah. They're very anti promotion and I don't know even when you're just trying to somebody has like an actual question about the exact thing and you're like, ⁓ like here's the thing, like here's exactly your answer, and like no self promotion, and it's like sometimes not even self promotion. I'm like, Yeah, it's frustrating.

Jacob Luetzow: I know. Yeah. So I've given up on Reddit and 'cause everyone's like, Well, that's how I got like my first like I got I went viral on Reddit and I'm like, Well, I've that's never happening for me, so

Brice Ayres: Well, I've seen some studies on how easy it is to go viral on Reddit. And so if you had a click far like they're showing if you had like thirty fake accounts or something like that, if you can initially get fifteen to thirty upvotes on your post, like it would trigger yeah, and so there's people showing there's some companies, you know, I don't know if they're out of China, I can't remember, it's a long time since I've seen this, but essentially you could pay them, you know, five hundred bucks or a thousand bucks, and they'll essentially using like real accounts, like go in and upvote your stuff.

Jacob Luetzow: ⁓ interesting.

Brice Ayres: And then if your stuff can just hit that initial velocity of upvotes of, you know, fifteen to thirty upvotes right away, it gets caught up in the algorithm and then just jumps up.

Jacob Luetzow: Mm. They they they block grokbot, so that sucks. But I was actually just thinking about that because you you essentially you have your own computer with grok, like you know, a VM obviously, but I was thinking like maybe I just install like a a VPN or like a proxy and have grok just constantly change it.

Brice Ayres: Mm-hmm. Yeah, you can do because I was doing a lot of this recently. I was working on a a side project called ⁓ orderwithsurf.com. And ⁓ order with surf is a f agentic food ordering application. And so I was basically trying to do what DoorDash does, and but most of these websites already have the ability to order food through their web portal and they already, you know, figured out how to integrate with in ⁓ DoorDash or Uber Eats. So if you just place the order on their website, now, you know.

Jacob Luetzow: Okay.

Brice Ayres: And my friend I was working on it with had a it was like a brilliant idea because I was like I was trying to explain to when I worked at Whataburger, I was like, it's a huge ordeal to integrate with DoorDash or Uber Eats. It's like tons of coordination and effort and that he's like, Why do we need permission to integrate? And I'm like, What do you mean? He's like, They have a website where you can place food orders on these various, you know, Chipotle or Panda Express or whatever. I'm like, Yeah. And he's like, What if the AI agent went on behalf of the person and placed the order and clicked through the browser? And was like, Whoa, like that's

Jacob Luetzow: Yeah.

Brice Ayres: That's kind of crazy, but it it was a challenge trying to bypass all the, you know, rate limiters and throttling and d just detection, trying to bypass them that stuff, because you're you're using these Linux boxes in the cloud and it looks at your computer signature and is like, this probably looks like ⁓ you know, Amazon VM. So then you gotta use residential proxies that are rotating, so you gotta pay for another service, but you don't wanna

Jacob Luetzow: Yeah. Yeah.

Brice Ayres: send too much bandwidth through it 'cause they charge you per gigabyte of traffic going through it. Yeah. So it was a a massive project. ⁓ and still working on it on the side, but ⁓ I've had some other stuff come up and so I've kind of ⁓ put it on pause for a second here.

Jacob Luetzow: Yeah, it gets expensive real quick. One of my very first coding projects before I even thought I was gonna be a software developer, because I was just doing like manufacturing automation stuff. But I was building a company's mailing list and at like just like clicking through Google, like Google Maps or whatever, you know, you can just hit next, next, next. But if you do that too much, like if it's too quickly, they limit you to like 10 pages. And what I would

Brice Ayres: ⁓ yeah.

Jacob Luetzow: I created a it was Java, which is disgusting, but I made a Java web scraper and a rotating proxy and it would remember the page number and then it would scrape all the data of whatever companies you were looking for, save it to like a CSV. It was super basic, but then like the cool part was I never got blocked or rate limited by Google. So I could get thousands of addresses.

Brice Ayres: Mm-hmm. Yeah. Yeah, that's the that's the great part about that stuff. The the challenge with like restaurants was like every restaurant's different and has its own policy. So like to do it just in so I almost had to like pre-crawl all these places and have my agent just run in loops in the background. And sometimes it would take like six to twelve hours to try to like figure out a restaurant, you know, the okay, it's this because some of them like a rendering with j you know, a single page application, JavaScript all like maybe it's an old Angular app or maybe it's ⁓

Jacob Luetzow: Mm-hmm. Dang.

Brice Ayres: you know, a WordPress site or it's using and so it's trying to figure out the patterns of ⁓ no content loaded and it's like, ⁓ it's cause it's all JavaScript loading into a single div in the the UI. You gotta wait for the whole page to render pretty click. So using tools like Firecrawl or sort of a whole bunch of things out there nowadays you can try to use

Jacob Luetzow: Yeah. Yeah. ⁓ I should mention like another cool thing that I'm using Grok for is I have a QA engineer who watches all commits that go into master. And when they when they hit it looks at the commit and then it QA tests my website.

Brice Ayres: Mm-hmm. ⁓ interesting. So it's like a s it does like a smoke test and prod type of thing after you release? That's a good idea. 'Cause I yeah, I why have I not thought to automate that? I'm like I always go check it myself and go look at it, make sure it's still up and running.

Jacob Luetzow: Yep. Yeah. I I mean, don't worry, I still check it myself. I'm like, ⁓ so O C D. I find all my bugs, but it actually has found bugs for me, which is really cool.

Brice Ayres: That's awesome. Well, I that's what Copilot was trying to do with like GitHub, right? Like when you'd open up a pull request, it would, you know, test your at least maybe the code. Maybe it wasn't going and QAing it after post deployment, but

Jacob Luetzow: I think that's what it's like that's what makes it cool, right? It's actually going through like a Chrome browser and clicking on things.

Brice Ayres: Mm-hmm. And that's where I've been shifting a lot of my tests locally, right? So like when I'm building features, ⁓ I actually run ⁓ K3s, which is like a like Kubernetes. And so I replicate my entire ⁓ infrastructure and ecosystem. And really cool project ⁓ for anybody who was looking for it, ⁓ I used to use this thing called local stack to simulate all the AWS services. ⁓ so like if you're using like AWS Qs or ⁓ SQS or SNS or anything like that.

Jacob Luetzow: Okay.

Brice Ayres: But there's a project that popped up recently called F ⁓ Flocky, I think it is. F L O K I. And so I use that now. And it Yeah. Yeah. So it basically ⁓ they they it's like an open source project that's popped up for free where I used to pay like 30 bucks a month for local stack to simulate AWS, and now I can run Flocky and ⁓ it'll simulate all the services locally. And so

Jacob Luetzow: Okay. I've heard of it, yeah. ⁓ yeah. It well I'm thinking flock cameras, so I don't know. It's probably something different. ⁓ that's that's awesome. That's cool.

Brice Ayres: So if you need a test integrating with an S3 bucket or integrating like it can run it all locally and simulate 100%. So like my environment I let my users test and prod for me. They are my QA team.

Jacob Luetzow: Don't you just ship it to prod and test before anyone else finds it? Freak UA, dude. No. ⁓ man.

Brice Ayres: So yeah, that's the ⁓ the best way I've seen to do it. And then I have ⁓ a lot of my t I have end-to-end tests now that can I can run locally, fully end-to-end testing SQS or Fargate or whatever service you want or need, you can test end to end. So ⁓ that's really how I've been using some of these agents to do work for me is like, hey, I need this integration with Stripe to work, but like

Jacob Luetzow: That's cool.

Brice Ayres: Do not stop. Give it like a deterministic goal. Like do not stop until you hit the checkout page and you see the price of twenty five dollars for this, you know, charge or whatever. And it'll just keep iterating and clicking and so it's spawn browsers and it'll click through the UI and I can simulate the entire stack end to end, which is ⁓ super cool.

Jacob Luetzow: Mm-hmm. That is cool. What like what's the biggest thing that you're seeing like companies and like what what you're being hired for now to help help people? Like is it to leverage AI or is it still they just need a new feature?

Brice Ayres: ⁓ yeah, I say right now most of my work's split ⁓ between like ⁓ AI data labeling. So a lot of companies need ⁓ help in the labeling space. ⁓ so I have like ⁓ one company, ML Twist, ⁓ that I work with and been working I was like their founding engineer and then kind of come back and help them out. and ⁓ they have some interesting projects and so like they're working with the TSA, ⁓ those ⁓ 3D body scanners you see at the airports. And so

Jacob Luetzow: Okay. Okay. Yeah.

Brice Ayres: They're trying to add AI, you know, to those. So ⁓ that's a really interesting project. And so you're seeing a lot of like defense contracting type stuff that's using AI. and so everybody needs tons of data labels. ⁓ yeah. Exactly, right? So the like everybody needs it ⁓ for all these different use cases, computer vision to audio to whatever it might be. and so I'm seeing a lot of that, and because of that.

Jacob Luetzow: Speaking of flock flock cameras.

Brice Ayres: lot of people need ⁓ data to train on and so synthetic data has become a big thing recently. ⁓ because there's not enough data to go around and you're like and especially if you need very ⁓ you can imagine like I'll use example of self driving cars, right? Like I need to drive down this road, but now I need to drive down this road and there needs to be fog. Or now I need to drive down this road and an alligator runs out. You know, which you're like, how often is an alligator gonna run out from a car? But you want to make sure it detects that and

Jacob Luetzow: ⁓ I didn't even think about that, yeah.

Brice Ayres: you know, whatever you want to do, swerve the alligator or or run it over. Whatever whatever you decide speed bump it, right? and so they need all these scenarios to come about. And so that's an interesting use case of sometimes taking existing like real videos and then augmenting them or extending them or generating them completely from scratch. And so I built ⁓ I have a couple of projects for different things where I've built ⁓ systems essentially allow teams of humans to go in and and build synthetic

Jacob Luetzow: Speed speed bump it.

Brice Ayres: data and synthetic videos. ⁓ and so you have overseas teams of people just so they're clicking. Yeah, and so that's become a big market, synthetic data. ⁓ and then I've been doing a lot in ⁓ the IoT space as well. So especially now you're getting ⁓ AI at the edge is becoming more and more prevalent as these models you can make them smaller and smaller and and put them on ⁓ edge devices. So cameras

Jacob Luetzow: That's really cool. I never thought about that use case. Yeah. Mm.

Brice Ayres: ⁓ and all kinds of things that like normally you would had to have a human come out and look at this thing and evaluate it. You can put a camera out there now, but maybe it's out in the middle of nowhere, right? And so you would need either ⁓ satellite communication ⁓ or cellular communication, which can be pretty expensive. And so you don't want transmit video over cellular for this little IoT device. So instead at the edge, like record a video and have the AI model at the edge, like determine if there's an issue with, you know, whatever the system is it's looking at. They'd like, ⁓ there's an issue. Go ahead and now send the small video clip back to the main server. ⁓ so lots of interesting cases for that.

Jacob Luetzow: That's cool. I've I've been having the itch to play with like nerves and stuff. And then like Alex Coatmose's book just came out and I just was chatting with him in in Chicago. I'm like, ⁓ I need to make time to play with some IoT stuff.

Brice Ayres: Yeah, you'll have to check out a company called Paridio. ⁓ so it's j Justin Schneck, ⁓ who was the, you know, original Nerves developer. ⁓ so they've launched a product called ⁓ Avocado OS. And so basically he made this more basically took all the things he learned from ⁓ doing Nerves and Elixir, ⁓ which he was like it was kind of niche, ⁓ and then making it more mainstream. So I did some work for them. ⁓

Jacob Luetzow: Radio? Okay.

Brice Ayres: And really cool, you can like basically connect to short short circuit the dev cycle in hardware. So hardware normally you'd have to like compile your code, put it like an SD card, pop it into whatever hardware, flash that firmware, and then you know, boot it up and watch it crash, then you know, go back to the drawing board, pull the logs out, load them back up. ⁓ so instead they can like you can set up a shell connection. So they have like a really cool tutorial on their site where you can go through a Raspberry Pi. And so you could use ⁓ Elixir, you could use Python, or you could use whatever you want for your embedded Linux. ⁓ yeah, really cool. And so they're doing they got some big ⁓ customers they've been onboarding, but that's ⁓ kind of his new equivalent of Elixir Nerves is his avocado OS. And so they're pretty bullish on AI at the edge.

Jacob Luetzow: That's cool. ⁓ another funny, so another funny use case for Grokbot, because I just saw I actually had Kill Switch account on X like one of my posts. I have it out, I have it do engagement for anything that like fits in the the stream of where I think my customers are. Yeah, and it liked my my own my own my post. So that's funny. Yeah, it's working. ⁓

Brice Ayres: Liking your own content. It's it's working.

Jacob Luetzow: So it'll like and comment on things and actually the other day it made a comment on someone's like Instagram reel and the creator responded but 'cause she was confused. She's like, I don't understand your comment and I looked at it and I was like, I'm just gonna leave this one alone because it didn't make sense. It was just like it just made a random plug for kill switch and I was like, you know, but it got her to engage. So

Brice Ayres: Fuzz. It's ⁓ I mean that's the that's the indicator. It's like engagement, keep doing that. It's gonna keep posting like random weird comments on people's stuff and

Jacob Luetzow: Yeah. Yeah, exactly. And I was like, you know what, Grok, do what you do. So like it's doing all the things I don't want to do and consistently, which is awesome, right? So I'm consistently posting every day from Kill Switch. And even though like my Kill Switch social media accounts don't have a lot of followers, it's still driving a lot of traffic to my website, which is pretty cool.

Brice Ayres: Ha ha. Yeah. That is awesome. Have you figured so question for you? ⁓ I notice when I hook up like if I'm using Cloudflare or Google Analytics or something, like or you could use Plausible or ⁓ was it Unami, I think it is, is the like another one like I've used. But like the numbers are always different for how many visitors and unique visits I got. Like, and I'm trying to figure out like, is it just bots and traffic? I'm like, that thing confuses me. I'm like, it says I'm getting like 3,000 views, but I'm like.

Jacob Luetzow: ⁓ yeah, I don't know.

Brice Ayres: I know no one's coming to my this website 'cause I haven't marketed this thing at all and I'm like

Jacob Luetzow: I think like bot traffic is getting crazy. Like we've we've seen it at Collects, my day job, and like it's putting a lot of strain on our on our infrastructure. But also like I think this is kind of like just something new we're gonna have to deal with because you don't necessarily wanna cut that off, right? 'Cause they're actually taking your data and you you could be referenced then, right, as a product. So I don't know.

Brice Ayres: Mm-hmm. Yeah. But I'm I'm just trying to figure out how to cut through the noise and figure out what real traffic, you know, looks like. And that's been my biggest struggle as I'm moving into my distribution, marketing and sales era of engineering, right? And I'm like, I gotta get a handle on this and figure out how to kinda cut through that noise.

Jacob Luetzow: It'll be it'll be interesting. Yeah, I think it'll be an interesting interesting thing. Cause yeah, I don't know what to tell you there because I'm just, you know, running around blind here. I have ⁓ a guest coming on next week, ⁓ Adam, and he's making like a AI like s like to help SEO and marketing and things, and it'll be interesting to talk with him about it.

Brice Ayres: ⁓ cool. Yeah, that's the big one. I think now that software's easier to develop, it's like the marketing engine and distribution is like where I'm trying to lean into and figure crack that ⁓ part of the puzzle. ⁓ do you what do you use for your analytics? Use plausible? Yeah, they seem pretty good. And it's Elixir built, isn't it?

Jacob Luetzow: Yeah. And then like I use plausible, yeah. It is Elixir Belt, yeah. That's kinda why I I went that route. And it's not Google. Like I use Google for a lot of things, but I try to not use it if I don't have to, you know?

Brice Ayres: Gotta support the homies. Yeah. There's so much better stuff. It's it's cool now too. Like that's another interesting part that's opened up with AI is being able to self-host things that normally I wouldn't make sense to self-host. And so like ⁓ especially I've been doing a lot of stuff with like ⁓ geospatial like mapping, ⁓ route planning or just GPS like tracking of devices and IoT devices. And ⁓ like there's a lot of stuff where like you need to

Jacob Luetzow: Yeah. Mm-hmm.

Brice Ayres: ⁓ reverse geocode every GPS ping of like when the vehicle's like driving around. ⁓ so you could like locate it or find it or look it up on a map. And normally you'd you call it a Google or Mapbox and have it reverse geocode for you, but it'd be so awfully expensive to do that. And so instead it's like, okay, well I'll just self host this open source tool that's really hard to set up and manage and you have to load all these addresses and ⁓ basically you know index the set up all the indexing on this database so it can performantly like look up and reverse geocode all these indexes or addresses. And it's like, ⁓ that's no big deal. That's just like a little side agent that's like working on that for today while I can keep working on my other stuff. Where normally you'd have a team I've had guys do this before and it was like I had a team of like one or two people spend a month plus setting up and configuring that and like something I did in an afternoon or a day, you know.

Jacob Luetzow: Yeah, that's awesome. The magic. The magic of our little AI helpers.

Brice Ayres: Yeah. But now it now it's too many ideas. Too many things to work on now. Too many things I wanna like do and I gotta like reel myself back in a lot of times, but like, no, not worth the effort. Don't go even though it's Yeah. ⁓

Jacob Luetzow: That's the problem. You can build it. Like just cause it's a good idea. Like you gotta put your hands up. Like honestly, I probably shouldn't have started making this podcast tool, but I was like, but it will also help me because I podcast.

Brice Ayres: Yeah, that's how that's how they get ya. I went I went down that's how those ideas get you. Like, well, you start doing the you know, with the boy math, the ⁓ well I'd help myself, you know, if I just put a couple more hours into it and now you're racking your own servers in your house and

Jacob Luetzow: I know. That's how those ideas get you. Eu tava... Well, like I already gave myself an excuse why I need to get a 3D printer. So like I want to ⁓ so the my little like IoT project that I want to build, I want I call it it's gonna be my ⁓ patio defender. That's what it's called. And basically it's going to watch for movement, like not for human movement, chicken movement, right?

Brice Ayres: ⁓ huh. Okay.

Jacob Luetzow: If my chickens cross onto the patio, they're gonna get squirt with a squirt gun. So they don't poop all over my furniture and patio. And so I wanna build that. And I think that'll be a really fun project.

Brice Ayres: Ha ha. That's cool. I had a a project it was probably like ten years ago. I had a buddy he worked at ⁓ I think it was Lake Tahoe community college. So in Tahoe they get just tons of snow and so the the school would shut down. Yeah, and so he worked for them, but he like, Man, they sometimes get a dump of ten feet of snow and nobody can come to class. And so they would record a video, you know, of the teacher teaching the class remotely for everybody. And so they were doing online courses, you know, long, long time ago.

Jacob Luetzow: Yeah, yeah. It's wild there.

Brice Ayres: And like, problem is you to have a student at least come to school and like move the camera and follow the teacher around. And so we took a camera and basically used computer vision. There was like a open CV library for Python. And I was able to track the human in frame and then like follow the teacher back and forth. And ⁓ it was really interesting because this camera we were using was like a security camera, and so it could pan tilt zoom and it go up to like 180 degrees, but there was no

Jacob Luetzow: That's cool. Yeah.

Brice Ayres: I had to basically reverse engineer the camera commands to send into it to tell it to like move. So I had to like kind of watch packets go across the wire, like when I was using their official like UI to like move the camera. And then I was like, okay, this is the command I gotta send to make it like move, but there's only a start and stop ⁓ command for the the movement. And so I'm like, well, how do I know how like to only have it move this much? And so ⁓ had to time it. How long does it take to do 180 degrees?

Jacob Luetzow: Yeah, yeah.

Brice Ayres: And then you divide it down by, you know, you're like, okay, if I want to move ten degrees, you take that and divide, you know, one eighteenth of the amount of time that I need to move. And so you're like doing all this

Jacob Luetzow: Well did that stay consistent enough where it worked?

Brice Ayres: Yeah, so it it was interesting the evolution. It was like initially it was kind of janky because it would move exactly how much it needed to move. And so you want to get smoother motion. So you would basically like overcorrect or like let it go a little s longer and then like so you weren't constantly like I always at the edge.

Jacob Luetzow: Yeah. And and then you come back or just kinda

Brice Ayres: Yeah, you kinda you get ⁓ into the center of frame, but like and so if they're if their bounty box got within so many pixels of the edge, you would then, you know, start moving it. Yeah, and so you would you would overcorrect ⁓ for the event because you don't know if they're walking to the edge and they're gonna keep walking. And if they kept walking, you would constantly be like jerking the camera to kind of move. And so you overcorrect until they eventually get back into the center and then you could stop.

Jacob Luetzow: Kind of pendulum swing it. Yeah, yeah. That makes sense. See, this is why like hardware stuff is fun. Cause you like you have strange limitations that you have to like figure out and make work, right? Cause that's kind of where I started my career was automate like manufacturing automation. So you have to do f silly hacks. Like a lot of things, like everything is off timing in in manufacturing, right? So you need like the more like everything was off of encoders. So the the more like the finer.

Brice Ayres: ⁓ interesting.

Jacob Luetzow: pulse you could get out of a spinning object, like the more precise you could be. So we had to pay for these expensive encoders. Like a revolution would give you like a million pulses or something, you know? So you have millions of pulses coming through, but you needed that for the the precision of the machine. And it was a lot of fun. Yeah.

Brice Ayres: Yeah. Yeah, anytime you deal with real world, right, like interacting, like that it makes it challenging 'cause in software you're like, okay, well, like it didn't go right, whatever, but it's like in the real world, there's like scenarios and things happening. And I've been doing a lot with ⁓ GPS tracking devices and cars lately. And this one's interesting because you're trying to build like a trip history, kind of like if you took a Lyfter Uber ride and like you show the the trip you took and so you generate like a post trip ⁓ report type of thing. And ⁓

Jacob Luetzow: Mm.

Brice Ayres: Hard part is like, is the car moving or did like the car just turn on? So like you don't want to r report a trip if it just turned on and then turned off. And then like it turned on, but the GPS sways and moves because maybe they're in their garage. And so it looks like it took a drive around the block, but really it just was stationary. And so you're trying to listen for all these signals and try to, you know, reverse engineer like how to determine if a car is actually moving. And the device will give you a moving event, but it doesn't always trigger. It doesn't trigger right away. And then

Jacob Luetzow: ⁓ right. Yep.

Brice Ayres: Yeah, it's just it's a series of challenges and problems. And then like you stop at a traffic light, the car's now stopped for two minutes. Is that a trip end event? Or like do you have to wait for the person to turn off the car? But some of these cars you can't tell because they're hybrid and so the engine actually turned off already. Yeah, and so you're like you s keep hitting all it's like every time you thought you figured it out, there's like another edge case that pops up that throws a wrench into your plan.

Jacob Luetzow: They turn off anyways, yeah. Yeah. Or, you know, you like make a fancy touch screen to control equipment and you're like, ⁓ this is perfect. And the second you let someone else do it, you're like, wait, why are you doing that? Why are you touching it in so many spots at once? And you're like, well, now I have to put all these guardrails in.

Brice Ayres: Ha ha. A hundred percent. That's like I had a buddy, he ⁓ he used to work for Intuit, ⁓ Turbo Tax, you know, company. And he said he was an engineer for them and he he was telling me they used to go to people's homes and they'd watch them file their taxes to see how they would use the software. And I was like, Man, that had to be riveting, just sit there watching and they couldn't couldn't give ⁓ any help or suggestions. They just had to like in their element kind of like watch ⁓ and observe like how they would click through the UI and like file their taxes. I'm like, that's that's ⁓ brutal.

Jacob Luetzow: ⁓ yeah, yep. Yeah. I just saw an ex post about watching someone go through your landing page and not saying anything, but watch how they navigate and interact with it and you'll learn more from that than like, you know, a million a million outreach calls to users.

Brice Ayres: ⁓ yeah. There's some good tooling. ⁓ I'm trying to think some of the tools out there. I mean, I think Datadog and all these guys that have tooling right now, like Heatmap. You can watch replays of all that stuff. There's some open source tools you can like self-host too to like watch people click through and like how they their mouse moves across the screen and but yeah, that's that's always been the most frustrating. You think you got a good game plan, you're like, here's exactly how you should use this thing, and then the person like right clicks the page and like, what are you doing? Don't don't right click like

Jacob Luetzow: Like heat mapping your landing page kind of thing. Mm. Dude, I remember like my grandparents, like when computers first came out, it'd be like, Well, is that a left click or a right click? And like, ⁓

Brice Ayres: Ha ha ha ha. Yeah. That's the part. You gotta build it for the triple click that one. You gotta build it for the lowest common denominator.

Jacob Luetzow: A double click or a single click? Now imagine like with swipe commands like swipe on tou on trackpads, like ⁓ my gosh.

Brice Ayres: ⁓ man. That's where I feel like I'm such a boomer when it comes to some of these mobile apps and things because there's so many hidden gestures nowadays that like ⁓ it's like a long press with like two fingers or you know, whatever it is and it's like there's so much hidden functionality now, especially when you get into mobile for like such a compact UI, you to like come up with more creative like swiping solutions to hide stuff or move stuff around and

Jacob Luetzow: Yeah. Well, now, like I don't even know if I ever want to make a mobile app with all the the folding screens and stuff like the coming up. Like Apple's new thing and then I think Samsung already had one, but

Brice Ayres: ⁓ yeah. I just saw that. Every mobile developer is probably just like shaking their head like this is gonna be a nightmare. 'Cause they they I don't know if you remember but when the original

Jacob Luetzow: Right. Well, even being a web dev now, how many different screens do you have to support screen sizes, you know?

Brice Ayres: Yeah. It's a nightmare. Well, that was the original thing when they came out with the iPhone. ⁓ their big thing was it's gonna be this size forever because it'll make it, you know, the experience so much better because it's a consistent UI. And it was like that went out the window very quickly. They're like, actually, okay, we need multiple sizes, but just these three sizes, and then it's like it just kept changing all the time. Now there's a little notch at the top that you have to accommodate for and ⁓ yeah, I had ⁓ it's kinda funny. I I used to use this ⁓ library called Ionic.

Jacob Luetzow: Yeah. Yeah.

Brice Ayres: I don't know if you ever used ⁓ back in the day. So it's basically you'd use web you'd a web page, so you could use Angular to build a ⁓ web view that would render in a mobile app. And so you could actually publish on the App Store and everything. And so it's kind of the pre pre React native. Yeah, but it was it was Yeah, and so it was like a web app that like had web components that like functioned kind of like native components. And I was demoing it to the customer that ⁓ I was like ⁓

Jacob Luetzow: Mm-mm. ⁓ like like React Native essentially, but just it wraps your web app. Okay.

Brice Ayres: Yeah, I was consulting at the time and I was like, here's your mobile app we built, like, I mean, I was really proud of this thing. I thought this thing was like perfect. And he was like, Hey, this thing's broken. And I was like, What are you talking about? And he was like, When I scroll to the bottom of the page, if I click at the very top of the ⁓ app ⁓ iPhone screen, it doesn't scroll back to the top. And I was like, What are you talking about? And it's like it's some functionality I never knew even existed on an iPhone. and he was he was like, Yeah, it doesn't work. Like, and I was like, I didn't even know that's a feature on you know, iPhones that you can tap at the top and it'll auto scroll to the top like

Jacob Luetzow: ⁓ I guess I didn't even know that either. And I used to be an iOS developer.

Brice Ayres: It's like maybe it ⁓ really? Yeah, I get I it's like a very niche functionality. So like the maybe it's ten pixels. Like it's at the very, very top of like the edge of the screen. If you tap, go to scroll to like Chrome or something like that on your iPhone and then scroll to the bottom and then tap at the top and it'll scroll all the way back to the top for you. Yeah. Yeah. I mean I think it's still that way, but he showed it to me and I was like, ⁓ my god, so I had to go build this functionality.

Jacob Luetzow: Okay. Huh. I just learned something new today after doing this for over ten years, you know, like It probably is. I'm not claiming to be an expert. I don't know. That's funny.

Brice Ayres: Yeah. Yeah. So there's th things like that that trip me up that I'm like thought you know, here I am the developer of the project and I didn't even know that functionality existed on the iPhone, you know.

Jacob Luetzow: Yeah, that's funny. Yeah, I haven't I mean I prefer back end development. So it's been a few years since I've been a mobile guy. And I don't miss it. Like that's like I'm always hesitant. Like ⁓ like I'm tempted to put kill switch, like to turn it into an app, you know, have like a menu bar version on a desktop. But then I'm like, then I have to deal with the app stores.

Brice Ayres: Yeah. How do you how do you guys now now with AI coming about, like, you know, what traditionally maybe you're in a back end position, it sounds like at work, but how do you bridge that gap now? Like, do you guys do more full stack development or do you guys kinda still stay in your lanes?

Jacob Luetzow: Yeah. Honestly, we're still we stay in our lanes and like my whole company is very like adopt AI. They're all fans, right? Like yeah, our CEO will build little side projects, right? Like little proof of concept stuff, and like I would say everyone's like very excited about it and we just leverage our tools to be better at our jobs. But like, yeah, we're not we d we I don't know. I think it's probably better to treat like Companies of certain sizes, like I don't think you really want full stack developers when you have a bunch of different products, but maybe I'm wrong. And maybe it'll change like if I like if you start with AI and few employees, I think you're more likely to not have all your individual departments. But I don't know. Maybe the fact that they already had them established, it's gonna be slower to change that.

Brice Ayres: Yeah, that's it's interesting. I got a buddy, he's a CTO, and he's got seven or eight ⁓ engineers working for him and it's tough because he's like they work faster than I can plan the work for them. ⁓ right now he's like, you know, my job as CTO is supposed be road mapping and planning and direction, and he's like, but he's like, they move so fast that it's like I'm having trouble keeping up with stuff. And so and like they're stepping on each other's toes if they're working in the same area, and so he's like,

Jacob Luetzow: Yeah.

Brice Ayres: He's like, I've almost had to go and give like these this person, you own this whole portion. Like this is your this is your baby. And then the you, this is your baby over here. And ⁓ it's it's an interesting like challenge now trying to figure out how do you work with these like large teams that, you know

Jacob Luetzow: Yeah. I've talked to companies too, like some people have gone wild with AI. And like maybe it's just like I'm a slow learner, but like some of these guys are like their engineers are like spitting out full features in like a day or two versus like what would normally have been like a three week, six week project, and they're spitting out each of their engineers are doing this. So like you're talking about multiple features going to prod a week from each engineer. Like how many features can you actually want or have in an app?

Brice Ayres: Yeah. That's why I got one client and they're they keep asking for more features and I'm like, I think we got all the features. I think now we should maybe shift to like marketing. Like, you want me to build you some internal marketing tools or something? Because I'm like

Jacob Luetzow: Yeah, like that's what I've been doing. Like I was like, I need to pause because I'm a developer. I like to distract myself with like silly little features, right? Like, do I really need to do that? Or should I focus on marketing?

Brice Ayres: Yeah. Yeah, that's engineers hate hate doing marketing. Have you guys bought into ⁓ like the mono repo versus some people are still like no microservices or at least ⁓ segmented systems repos. Mm-hmm.

Jacob Luetzow: I mean, I'm a monolith guy and you can't change my mind. ⁓ but I know the microservice people out there and they love it. And I just feel like you get tied into a platform, you get trapped when you use too many microservices. And then when you scale, you have huge amounts of ⁓ like server costs and whatever it may be, and it's really hard to leave. It's hard to move your stuff. So I'm a I'm a fan of the monolith. Like make it an umbrella project if you want to break it up, but

Brice Ayres: Yeah. Yeah, even then it as long as it's modular, I think that's fine. Like I've been big on the the model repo for a long time, you know. So I saw how fast, you know, people in Ruby and Rails were moving and like just very simple, you know, the full front to back end like and I'm like, these guys are like shipping lots of code. And then I watch the teams of microservices and like they spend a year scaffolding and architecting, and I'm like, You could already had something shipped, especially now with AI.

Jacob Luetzow: Me too. Mm-hmm. If you want to overcomplicate something, use microservices. And if you want to have to hire more people in your company than you actually need, use microservices. Like that's my opinion, at least. I don't know. You can get you can tell me I'm wrong, but I think it just complicates things.

Brice Ayres: Yeah. It's like you learn A hundred percent. Yeah. And I feel like people try to use it as like a rubber stamp on their resume in some cases to be like, look how complex the system was. Yeah. The system's overly complex, it's overly complicated. And 'cause it you know, and that's the challenge right now we have is things look more impressive with the more people and more budget you have. But what's really impressive is the guys who are shipping ten X the features with like two or three people or five people, you know, versus a team of fifty. ⁓ but I've heard

Jacob Luetzow: Ooh, AWS certified. Like great, dude. Yeah. Yeah. Well, no, sorry, I was gonna say, look at what one instance of an Elixir project like an Elixir node can do too. Like you can handle a lot of traffic on one node and you have a lot of vertical room before you even have to think about the complication of scaling. And then that's built in for you with you know, with the beam and OTP.

Brice Ayres: Yeah. I mean, and you start to look at some of these projects like ⁓ I don't know if you saw the other day Planet Scale, Neki. ⁓ I think I don't know if I'm saying that right. N A N E K I. But ⁓ they essentially got their sharding Postgres. So before my sequel it was called like Vitus. ⁓ not sure I'm saying that right either. ⁓ reading is hard. But ⁓ but it was ⁓ essentially like horizontally scaling ⁓

Jacob Luetzow: Mm mm. Mm-hmm. All right, hey, I'm I'm right there with you, dude.

Brice Ayres: MySQL, but now they've done it for Postgres. And so they they posted something the other day where they're running like five petabytes of data through something like a hundred and eighty million ⁓ requests a second they're pumping through this system that that they were ⁓ load testing. And you're like, at that point, like you're that's massive. Like they had five hundred and twelve shards or something for you know, so a lot of people hate on ⁓ monoliths and sequel, but I mean, I think ⁓ Think it's open AI. I think they have one big database. ⁓ I don't know if it's MySQL or Postgres they're using, but one giant SQL database is their like write and then they like fifty read replicas. And it's like you can go a long ways on a very simple architecture versus having like a million microservices with their own database, you know, and trying to figure out how to scale it.

Jacob Luetzow: Yeah. Yeah. But when you max out your database instance, that is a hard problem.

Brice Ayres: Yeah. Yeah. Sc sharding is very hard. You do you wanna but I've seen people vertically scale their database and not nine times out of ten, they could reduce it if they just had proper indexing, some optimizations, ⁓ you know, pooling connections or what you know, like a PG bouncer sit in front. Like if they just did some of those optimizations, like they would actually take a lot of load off their system and then just have a read replica. Like that'll get you so far if you have a a write instance and then like a read instance.

Jacob Luetzow: Right. Mm-hmm. For sure.

Brice Ayres: All the reads go to these and you can scale those pretty well. And ⁓ surprisingly. ⁓ so that's where I think like moving quick, shipping stuff, using monorepo, use SQL as your main database. Like if you have another need, like I do a lot of ⁓ telematics type data for vehicles and cars. And so like I click house as like been my other go to data store for just like logging. So you can just do so many writes and horizontally scales really really easily. ⁓ and so you can just

Jacob Luetzow: Okay. Yeah.

Brice Ayres: slam it with rights and then it does really good at these like aggregate queries. and s really good. And so the performance is just so I'm like for that I'm like looking at some of the scale. I'm like, okay, I need to be able support, you know, a million vehicles or two million vehicles driving all at the same time. And it's like I could scale that up from you know what I got right now. ⁓ and maybe some small tweaks are all in the way and use ⁓ I don't know if you use Valky or Redis. ⁓ but using that. But the trade off I've been trying to find

Jacob Luetzow: Yeah. Yep. Yeah. I've used Redis a little bit, but

Brice Ayres: ⁓ yeah. Yeah, but I've been trying to figure out like should I just use ⁓ Elixir and use ⁓ you know, some of the built in tooling, ETS or whatever it might be, right? ⁓ instead. And it's ⁓ it's kind of a balance there because you're like then you start

Jacob Luetzow: Yeah. I think the the hard part with like ETS is distribution, right? 'Cause I don't think you can you you're stuck to a node with an ETS table, correct? Maybe I'm wrong. Yeah.

Brice Ayres: I think so yeah, and then there's Namisia ⁓ Amnesia, which is like the distributed version of it. But but supposedly ⁓ I haven't used it in production myself, but I've heard it's hard to manage because it's basically they all have to keep a copy of all the data. And if you're doing ⁓ auto scaling policies like up and down, then as nodes are joining the cluster, they have to like re download a copy all the stuff and then maybe it scales back down now they're destroying and then adding, and so you're you're burning a

Jacob Luetzow: Okay. I say. Auto scaling is a complicated problem too. Like that like I don't know. Elixir's so nice because you could almost just kind of plan regional nodes and just have ⁓ always have like three nodes or something. And you can handle a lot of traffic. And then when you have to worry about auto scaling, I don't know, bring in the specialist.

Brice Ayres: Mm-hmm. Yeah. A lot of people underestimate ⁓ auto scaling, right? It's like there's a there

Jacob Luetzow: It's not just like flipping on a switch, be like, Yeah, just do it.

Brice Ayres: Yeah. I remember a lot of people like in Heroku back in the day. I know if you use Heroku back the day, but it's like, ⁓ I want five instances or you know, and it's like, ⁓ I auto scaled and you're like, But now you have like web sockets, and you're like, Well, the web socket connected on this one, but the message published on this other one, so now you need Redis as the backbone to keep track of ⁓ which instance has which client connected to it. So the messages get publ like a whole headache of problems that come with auto scaling. ⁓

Jacob Luetzow: I have, yeah. ⁓

Brice Ayres: Yeah, it's it's not ⁓ not simple.

Jacob Luetzow: It's hard problems. All right. This has been fun. I think ⁓ we can wrap things up though. The time goes by really quick. And ⁓ thanks for hanging in there with me for using Riverside. And you're my first podcast for like the last two months. So I was a little rusty maybe, but I think it went pretty well.

Brice Ayres: Yeah, it does. This is great. Yeah, learned some stuff about marketing, about crockbot. Awesome. Yeah, yeah. This is great. Learned a little bit about Grockbot and so now into some marketing.

Jacob Luetzow: Yeah, so where can people find you if they if they want to connect or any of that? ⁓ you're on X.

Brice Ayres: Yeah, you can go to ⁓ axiointelligence ⁓ dot com is ⁓ my ⁓ fractional engineering. if people just want to talk about their projects or need help with something, I'm there. otherwise on X BryceAyres, B-R-I-C-E-A-Y-R-E S. ⁓ they can find me there. ⁓ I'm probably in a couple other places around the internet.

Jacob Luetzow: Okay. All right. And what like if you were to name like what what what's your specialty? What do you like what's your favorite kind of work and what do you think you're the expert what's your expert domain?

Brice Ayres: Yeah, I think ⁓ AI infrastructure, IoT connectivity, satellite communications been a big one lately. ⁓ and then just general infrastructure. So, you know, people are spending a lot of money on cloud cost right now or whatever it might be like can come in and help you reduce your cloud cost pretty quickly or, you know, optimize some small part of your system to get you more bandwidth. You know, we're talking about data.

Jacob Luetzow: Okay. Okay. ⁓ I wanted to talk about that. Well we're gonna have to have you come back on to talk about your your cloud cost stuff and how you ⁓ well may I mean we can just talk about it. Like all right, so I was looking, I think I saw it on your website, but you basically offer like you audit people's systems, right? Like with just read-only access and you have like an AI automated tool and then you review it. What

Brice Ayres: So yeah. Okay. Yeah. ⁓ Yeah.

Jacob Luetzow: What are you actually doing there and like what requires the the human to review things? Like, does your AI automation get things wrong or does it what happens?

Brice Ayres: Yeah, yeah, sometimes it it it's not perfect, right? But ⁓ at least gives you kind of like where there's smoke, there's fire. And so there's parts like the AI is really good at quickly and people could do this themselves. Anybody's like pretty technical, like give your agent read only access, log in, you know, and then go and look like basically combine your repo. Like if you have multiple repos, try to put ⁓ together or something. ⁓ give it access to all the repos, give it access to your production instance, your cloud billing. So I've discovered, especially for myself, like some of my small projects. I've been using AWS and I'm trying to keep my cloud cost down. I'm like, I don't want to spend a thousand dollars a month hosting this thing. That's a pretty simple app. And so you can look at some of those things. pre pre buying spend, ⁓ a lot of companies don't know, but you can go to AWS and say, like, I'll pre commit to a certain amount of spend. ⁓ that's like a a massive lever that some just don't do. And it's like just commit to ten percent. Yeah, you can save up to like thirty percent.

Jacob Luetzow: Mm-hmm. Mm, yeah. You can save a lot of money. In fact, we were just talking about pre buying stuff at Collects just as cut down on our server costs.

Brice Ayres: ⁓ yeah. But y you gotta be careful because especially with AI now, you can probably quickly reduce your cost. So reduce the cost first, then commit to spend. Cause I gotta I know another company, they they pre bought spend and now they have all this extra commit they're spending every month, but they've optimized things now and so now they're trying to sell to other companies like come host your app on our AWS account and we'll save you twenty percent, you know. ⁓ but one of the biggest levers I've found is logging.

Jacob Luetzow: That makes sense. Yeah.

Brice Ayres: So logging in any system usually eats up a ton of money, especially like if you're an Amazon like CloudWatch or whatever it is. That is massive. And a lot of people ⁓ between that and ⁓ well before I get into the other one, ⁓ so you yeah, just logging is crazy. And so I came up with my own little process, ⁓ basically using it's called Fluent Bit. And so it's like a little agent, it's like like a background daemon that just runs on your server, your clusters, your instance.

Jacob Luetzow: Mm-hmm.

Brice Ayres: And ⁓ basically gets all the logs, my Kubernetes logs essentially, and every so often basically groups it and logs it into S3 and dumps it to S3 so I don't pay for any cloud watch ingestion costs. ⁓ and so it dumps it into S3 for me every you know few minutes or whatever, it'll grab you know 100 megabytes or every few minutes and like throw it into S3 ⁓ as like parquet files, and then I can use DuckDB, which is a really cool tool if you haven't used that yet. ⁓ and see

Jacob Luetzow: ⁓ nice. I haven't used it but I've heard I've heard good things.

Brice Ayres: Yeah, and so you can really performantly like query your logs. ⁓ and if you need real time stuff, obviously connect to your cluster and like look at the live running logs and tail ⁓ But otherwise you can go to S three and have really cheap log storage. And so I have a retention policy set in the S three bucket and after seven days it just, you know, removes the logs. ⁓ so it's really easy to set up with AI, but that was a big one where I was like spending a few hundred bucks a month on logging and now I'm spending zero logging every month. ⁓

Jacob Luetzow: ⁓ nice.

Brice Ayres: Or not zero, but close to zero, just the cost of like S3 storage mostly. ⁓ then that gateway stuff's a big one. ⁓ so like if you're communicating between different systems and you're going out to the internet and then coming back in, it's like you're paying for all that traffic, whereas usually it's free inside of, you know, the same availability zone. ⁓ and then some people who are running like high availability, so they'll have multiple availability zones in one AWS region. ⁓

Jacob Luetzow: Yeah.

Brice Ayres: their systems are communicating. So like it may come in, let's say it like a request will come in on a load balancer, but then it goes like another region, or something not another region, but another availability zone. And then it connects to the database back on another availability zone, like a read replica over there. And so you're burning all this cost trans, you know, going across availability zones. ⁓ and so you can do some stuff to like make sure they connect to the same availability zone. So the same services that are in the same area.

Jacob Luetzow: Yeah.

Brice Ayres: It'll always connect to those like the nearest ones, the nearest ⁓ Kubernetes node. ⁓ so if it's in the same availability area. ⁓ you can also set up like VPC peering and some other things like that, where you can create like a virtual network between the availability zones so you don't pay for ⁓ cost. There's a couple other ⁓ tricks like that, but between those like logging, ⁓ internet traffic, like making sure you're connected to the same availability zones, like those will those will save you a ton of money right there alone.

Jacob Luetzow: That's awesome. So guys, just contact Bryce. He's he's a wizard. ⁓ He's got all this tooling set up for you. Just there you go.

Brice Ayres: Ha ha ha. Yeah. It'll it'll pay for itself overnight if you need help reducing your costs.

Jacob Luetzow: That's awesome actually. Like yeah, there's probably there's like so easy to like just accumulate so much bloat in in your server costs.

Brice Ayres: And it's it could be something as small as like one line config in your config that could just route traffic to the correct place and you'd avoid this, you know, network charge every month. Especially you start getting higher volumes, it adds up quick.

Jacob Luetzow: Yeah, that's wild. Yeah, for sure. All right. There. I guess we just ran rambled a little longer, but that was good. I'm glad we talked about that. All right. Thanks so much, Bryce. This was a lot of fun. ⁓ if you ever want to come back on, make sure you ⁓ just reach out. And ⁓ everyone else, thanks for tuning in. It's good to be back. And ⁓ I have I'll see you guys next on Wednesday and Friday. I got two podcasts next week, so hitting the ground running.

Brice Ayres: Awesome. Absolutely. Nice. There you go.

Jacob Luetzow: I gotta figure out this riverside though. There's some things I don't really like that I'm probably just doing wrong. But yeah. All right. See you guys.

Brice Ayres: Gotta build your own now. Awesome.