Welcome to the panel where founders talk about building a better business and a better life. I'm Justin Jackson, co founder of transistor.fm.
Brian Casel:And I'm Brian Casel, creator of Builder Methods.
Jordan:And I'm Jordan Gal, co founder at heyrosy.com.
Justin:And we have a special guest this week, Aaron d Francis of Laravel Holding Incorporated.
Aaron Francis:That's right. Honored to be here. Thanks for having me. Well, we've Welcome, Aaron.
Jordan:We've two two VC backed and two bootstrapped at this point.
Justin:That's right. That's right.
Jordan:That's right. The force the force is bounced.
Justin:Yeah. We had to bounce it out. And this week, Brian's hosting. So yeah. Take it There away,
Brian Casel:there are so you know, we we have our our our chat group where we sort of just like bounce around ideas. I feel like this week we've got a whole lot of them. I mean, Aaron's here. There's a lot of lot of things I feel like, Aaron, I wanna catch up with you on. We might as well do it on air.
Brian Casel:But, you know, we gotta talk about Jev. We're gonna talk about, you know, what's happening in in Laravel land. I I just wanna hear a lot more about Aaron's whole just career change, and and that'll I mean, I I know you've talked about it on mostly technical, but that I've always been fascinated with people making big career moves. That's that's really interesting to me.
Justin:What's that been like so far, Aaron, being a w two?
Jordan:W two and and finding out that DHH is a fanboy. That's that's
Justin:a lot
Jordan:all at once.
Aaron Francis:I mean, I don't know which one's better, honestly. They're both fantastic. I love it. Yeah. So I joined Laravel, like, maybe two months ago now.
Aaron Francis:And it's just been it's just been great, guys. It's so awesome.
Jordan:I mean, you had a big, like, coming out, like, opportunity at the at the the big conference. Yeah. Right? What Yeah.
Aaron Francis:Was that
Jordan:where was it? England or something?
Aaron Francis:It was in Boston. Oh, same thing. Same thing.
Justin:New England. New New England.
Aaron Francis:Yeah. Nailed it. Yeah. It was that part that part was a lot of fun because normally I've emceed Lyricon the past, I think three or four years. People like I was on the I was on the website as emcee.
Aaron Francis:And so Taylor comes out. I'm supposed to come out, but Taylor comes out first. He's like, hey, guys, you know, we're gonna change things up this year. Usually, we have a member of the community do this, but I just you know, we're gonna change it. This year, we're gonna have an employee do it.
Aaron Francis:Today is actually his first day as vice president of marketing. Please welcome Aaron Francis. And people were like, what the hell is going on? It was great. It was masterfully executed.
Aaron Francis:Nice. And everyone you know, part of the I think part of why it's so great is I've just been there for ten years or more. And so people were genuinely excited for me. I I knew, you know, 90% of the attendees there just because I've been around for so long. And, obviously, doing a job well is hard, but this has felt pretty easy because I'm not I'm not like trying to figure out what is the identity of Laravel.
Aaron Francis:How do I speak about it? It's like, oh, no. I've been there the whole time. And in some regards, I've helped set that as, like, member of the community, emcee, outspoken, you know, person. And so it just feels like slotting right in.
Aaron Francis:It's been it's been awesome. My wife is even more thrilled than I am cause I'm way less stressed.
Brian Casel:Yes. I mean, that's what
Justin:That's a big deal.
Brian Casel:That's what I think is so interesting when you when you see someone in our community make one one of these big moves. Like, when I heard about it, it was like, oh, I didn't even know that, like, a role like that was even something that you're would even, like, consider. But then it's at the same time, it's like, oh, obviously. Like, you're such an obvious perfect choice. Like, know, from Taylor's perspective, like, it it it makes perfect sense why he would go to you and and Laravel, and you're already in the community and and it makes perfect sense.
Brian Casel:I I mean, I wanna I wanna ask you all about more of the decision process on everything else that you've had going on in your in your life and work and everything and and then the transition. But where we sit today, you said that you're, like, two months in. Where are you at today with it? Are are is it, like, you're truly, like, in like, embedded, like, part of the team running running the show, like like, making a lot of decisions? Or is it still, like, sort of like the getting to know you phase, the honeymoon phase within Laravel, getting your like
Aaron Francis:Yeah.
Brian Casel:Are sinking into right now?
Aaron Francis:It's kinda interesting because, you know, I came into this marketing role with a a team full of pre existing people. And within Laravel Holdings Incorporated, the people that I knew the least and have the the least credibility with are the people on my team. Yeah. Because they're like, they're not the developers. Like, we didn't grow up together.
Aaron Francis:Like the, you know, Taylor, the VP of product, the VP of engineering, like, we all kinda came up together in this space. And then all of these professional marketers. And they've been inside the Laravel world for a little bit, but they're very, very, very good at being professional marketers. And I'm just like, I'm not one. I'm not a professional marketer.
Aaron Francis:And so I think a lot of the appeal for Taylor was we have this team full of people that are killer marketers, so we don't necessarily need a like person who's come up through the marketing ranks to execute on all these these playbooks. I refuse to call them motions because I'm trying to resist. I'm trying to resist marketing speech because I hate it.
Justin:Good luck.
Aaron Francis:So, yeah. Thank you.
Brian Casel:You're gonna find some synergies.
Aaron Francis:Oh, hey. It wording in there is just it's crazy. People will hit me with acronyms and I'm like, I'm sorry. I literally have no idea what that means.
Justin:So I think part of the
Jordan:appeal Okay.
Aaron Francis:I think part of the appeal from Taylor's perspective was we need somebody who has lived and breathed this whole thing to help, like, set direction and tone and, like, the soft feely parts of it. And we then have all these experts who can go run paid ads and write content and do, you know, marketing design, that kind of stuff. So it's been it's been awesome so far. It's almost like you're
Brian Casel:there to like bring it back down to earth. Like Totally. Bring it back down to the on the on the ground developer or end user, like make this feel real. And when and when it's coming from like the top, meaning like you at the top of the marketing arm of Laravel, that that brings a lot of weight to to whatever machinery they can they can put at it, know?
Aaron Francis:Yeah. Absolutely.
Justin:Yeah. I think this is like we're gonna see more of this. I mean, if we could all be so lucky to have an influencer like Aaron available to hire who knows the product, who knows the people, who has already built trust with the audience, and then be able to hire them as the key marketing person, I think this is this is what Lars Lof was talking about. Increasingly, you want your main person to be the person who knows the product the best, the best communicator embedded in the culture, all that stuff. You get so much as a part of that package.
Justin:And then you can hire a bunch of professionals to do the other stuff if it's whatever ads or whatever. So
Jordan:curious what marketing looks like in a product and a company and an environment that Laravel's in. Because I I assume it's not. Maybe there's a mixture between traditional and authentic marketing.
Brian Casel:Yeah. Like like what kind of motions do you have over there? Stop. Stop.
Justin:Get out
Brian Casel:of here.
Justin:What kind
Jordan:of playbooks are you putting into motion?
Aaron Francis:Yeah. So it's been it's been very interesting, at this point in the history of Laravel Holdings Incorporated. Right? Because we're going from we're going from world class in my opinion, application framework that's written in PHP to hosting infrastructure, big boy and girl engineering, like actual honest to goodness, hardcore engineering. And, you know, Laravel's history serves us well there because we have fifteen years of credibility and trust and Taylor taking things seriously and having a high bar for quality.
Aaron Francis:But there's also a delicate balance of, okay, we're going from like almost mostly feel good open source vibes of like, hey, we're all in this together to Laravel very publicly took $57,000,000 of VC funding and now we have a hosted SaaS application. And so that's that's like that's a big part of what I've been spending my time thinking about is how do we honor and respect the community as it exists, of which I am a part, which I think is very helpful. How do we honor and respect that community, casts like a vision for them that like things are going to get better. We're gonna get more people into the community. We're gonna have more resources.
Aaron Francis:Like everything about your life is gonna get better. By the way, the shape of things is gonna change a little bit. And that's like a pretty that's a pretty tough needle to thread. And so I have been really focused on besides, like, there's a lot of operational stuff I can tell you about. But besides that, I've been very focused on what is our story for the next ten years, And how do we how do we not abandon the people that got us here?
Aaron Francis:Because I was one of those people. And trying to figure out how can we welcome in new people that don't identify with the fifteen years of history that Laravel has without abandoning those folks, which has it's been been hard, but it's been a lot of fun.
Brian Casel:On that, if I like, I wonder how you think about that story in the yeah. Like, in the context of now and then going forward. So how much of it is Laravel and the cloud offerings and and everything is a product for people who, like, verbally choose Laravel as their stack? Or is it we're going more tech agnostic, like stack agnostic? Yeah.
Brian Casel:Because if I if I think about product builders today, it's it it's it seems like less and less about the stack and about our preferences of the type of code we want to be writing or the
Justin:Yep.
Brian Casel:Or it's it's so much more about, I don't know, like like even that it's really more about the choice in infrastructure than it is about coding tech language. Right? So
Aaron Francis:Well, that's a hard thing. Right? Because we have two we have two masters that we need to serve. And if anybody's up on their biblical history, not a great idea. You don't want to serve two masters.
Aaron Francis:That's always a bad idea. But we have to, right? We have to serve the open source framework as a discrete product. Like, we have to do that because, one, I think it's the right thing to do and Taylor believes it's the right thing to do. Taylor's also super passionate about the open source framework.
Aaron Francis:But also, that is a growth. That's a growth engine for us moving forward is more and more people choosing Laravel, the open source framework. On the other side, we have cloud, which makes the money that funds the open source. And so we have to grow that, obviously, because we've got people that need money to put food on their family. And so, like, we do have these two masters we have to serve.
Aaron Francis:But to your question specifically, I think going forward, that's part of what that's part of what the, like, the positioning of Laravel, not only the framework, but the company is is slightly trying to change is, hey, for fifteen years, all of us, all four of us, maybe Jordan, I don't know how much you handcraft code, but at least, you know, at least some of us have handcrafted code with like great care and great attention. And now out in the world, good or bad or at different, people are doing that less. And so do we do we want to win? Like, do we wanna capture those people? Or do we wanna hold on to the past and say like, no, Laravel's only for people who handcraft code.
Aaron Francis:Like, there's no world in which Taylor is like, well, let's slap something out and throw it into cloud. But if you look at the rest of the world, everybody's slopping things out. And in fact, it's getting less sloppy, but they're slapping things out and they're like, let's let her rip. And we to be the home for that.
Brian Casel:I I hear that, but I do I do think that like, just from a marketing positioning story perspective and like a person or a company who's who's building and and making buying decisions. Yep. Around what they're building and where they're to to me, the the actual buying decision or or choice is like which cloud provider. It really is all in the cloud. And it's not just servers.
Brian Casel:It's model providers, harness providers, which are now like a lot of them in the cloud. And then and then hosting provider. Like, if I think about like, alright, am I gonna choose Heroku for this app or or Digital Ocean or wouldn't choose a cloud.
Justin:Yeah. You
Brian Casel:know, or whatever. You're like like whatever like do it's it's it's actually like less about the the text. I mean,
Aaron Francis:as builders, we still come
Brian Casel:from that background, but I wonder if if, like, the next ten years anyway.
Aaron Francis:No. I agree. So on the framework side, our intent is to capture these these builders that are entering into the building space. Right? And our intent on the framework side is to capture them by saying, look at all of this foundation that we have.
Aaron Francis:If you go over to maybe like the JavaScript ecosystem, you're like, oh, let me grab this off package from some company and this queues package from some guy in Nebraska and like, let me piece it all together. So our builder story on the framework side is exactly that. Like, hey, if you're not gonna look at the code, then PH the fact that you don't like PHP doesn't super matter. Like, we've baked it all. It's done.
Aaron Francis:And then you look at the cloud side, and Taylor tweeted this yesterday. We host Rails. We host Django. We host Next. Js.
Aaron Francis:We host Java. And so like the dream on the cloud side is the love, care, and attention that Laravel is known for can be brought to these other ecosystems. And what we want is we want ambitious builders. Like we just want ambitious people who are trying to build and scale, and you need a database and a cache and a queue and emails and servers. We've got it all whether you're Rails, Laravel, Django, Phoenix, whatever.
Justin:Yeah.
Brian Casel:I'm I'm also I'm I I wanna take it back to Aaron Francis. Like, this the I'm curious about how do you feel right now about your day in the life or week in the life compared to being totally independent, which you were for, you know, a a while there before.
Aaron Francis:A while, yeah. So
Brian Casel:How how does it, like, logistically even look? Is is it very much different?
Justin:Is it Oh, it's
Aaron Francis:completely different. Yeah.
Justin:Yep.
Aaron Francis:I mean, I start I start my agents at night. I check on them in the morning, and then I check on them at night. So it's completely different because I've
Justin:done my job all day.
Aaron Francis:So which, by the way, I blew through four Codex accounts because I was alright. Just Y'all have fun. I gotta go to work. Just let her rip. And they boy, did they let her rip.
Aaron Francis:So I won't do that one again.
Justin:Do you just stack them up before you go? Like, you're like, okay. Codex one, codex two, codex three, codex four. I'm going to work.
Aaron Francis:Yep. Daddy's going to work.
Justin:Get busy.
Aaron Francis:I basically write this long spec document. And then I tell the lead orchestrator, load balance across these four accounts. Have fun. See you later. And it did.
Aaron Francis:My life looks completely different in that I'm in a lot of meetings. I'm taking a lot of Slack messages. And one thing that I really like is I'm doing a lot of operations. So, like, marketing, if you think about it, is like, oh, let's go out into the world and shout at people. What it's been what it's been for the first two months is where are all the processes?
Aaron Francis:Like,
Justin:where
Aaron Francis:do we keep stuff? How am I supposed to know what anyone is working on? When you say you're gonna do something, where do we track that? Did you get it done? And so coming in, there's been a little bit of turnover in this role.
Aaron Francis:Coming in, I'm seeing like half baked systems that have just been left behind. And I've been I've been COO somewhere before, and this feels very much like a COO thing of being like, oh, up. What's the system here? What is the root problem? You're like, we're talking up here.
Aaron Francis:We need to solve the root of it, and then we don't have that problem anymore.
Justin:Yeah.
Aaron Francis:So it's been a lot of like, let's get all of this stuff out of here and institute some actual good processes, which is fun. I love that.
Justin:Unpopular opinion. But I think that solo operators actually make great employees. Because when you're a solo operator, you have to be ruthless about, like, does this matter? Does this move the needle? Does this actually get me a paycheck at the end of this month?
Justin:And you deploy those people inside a company, and all of a sudden, you're just like you you have zero, like, appetite for bullshit. It's like, okay. Why are we doing this? Like, this doesn't move the needle at all.
Aaron Francis:Why does this take a month? Do it today. Yeah.
Jordan:Yeah. Exactly. Be the right company, or or it won't work?
Brian Casel:Yeah.
Aaron Francis:Yep.
Justin:While we're on the topic of Laravel, wonder if I can just share this Taylor Outwell post because I've been, like, referencing this in our own company and everything else. I think it'd be good fodder for right now. Look at this. Our goal for 2026 was to three x Laravel cloud. Well, already done.
Justin:We're on pace to five x. Love that. This is I loved this part. I'm like I'm trying to bring this energy to everything I do. Focus on making the product decisively the best in its category.
Justin:It's easy to be distracted from this, and there are no gimmicks or tricks to get along get around it. The product is either the best option in the space or it's not. And being the best is a blend of a lot of factors. I I think in the age of AI, this is the attitude that's gonna kill now. It's like there's sure, everybody's building stuff.
Justin:And what I'm seeing is an inattention to detail. I'm seeing people not building stuff that is the best in the category and figuring out a way to get there. So this whole, like, tweet is just awesome.
Brian Casel:What you got, Jordan?
Jordan:My response to this, it's not negative. It is unsure. Because when when I hear this, I don't know what to I don't know what to make of it. I don't know what to do with this. Because to be the best in the category sounds to me, and this is like where I want help or some explanation or something.
Jordan:Yeah. It sounds to me like I need to like absorb what's going on in the market. What the other options are. What their features are. What their strengths and weaknesses are.
Jordan:And almost react to that and make sure that I'm better in this way and in that way and that I got this covered and this other thing covered. And I I I don't like to do that. I I like to build a good product for the customers that are currently there. It might be because the space that we're in is less mature. So it's it's less of like, alright, people are our our customers are generally call it eighty twenty.
Jordan:They're not switching from an one provider to another. Mhmm. It's their first time onboarding into this type of a product in general. So is this a function of a more mature market where every single person's going through a decision process and you can kinda name what those options are and then you have to be the best there? I I don't know what what to do with how to be the best in the category.
Jordan:Category. Because I I like the idea, but how do you do that?
Brian Casel:I I read that or at least the meaning for me meant something a little bit different. I hear you, like, because I I I definitely relate to the like, I I try not to even pay much attention to competitors and, like, competing on features or competing on quality of our thing versus their thing. But I what I heard in that statement, which really resonated was really care ruthlessly about
Justin:Mhmm.
Brian Casel:Quality and and maybe putting your putting your own view of what the what like, defining the best in your view and going hard on that versus just shipping. Like, just ship or just ship the or or just ship or ship the perfect solution.
Jordan:Right. It's like the the quality bar and like that energy to go all the way up the absolute best you possibly can.
Brian Casel:And it's also like like really trust in your own conviction of like, we believe that this is the best way to do cloud infrastructure and this is what what we believe it should be. So we are gonna make it so rather than like, all we need to do to compete in this category is ship this, and if we can ship it by this date, then we can win. Like, that's a different thing that you're prioritizing for. Right?
Aaron Francis:I see a lot of, we'll just say people on Twitter. I don't know if there are indie hackers anymore or not, but I see a lot of people on Twitter that just like, that set their ambition bar too low. Like, I'm going to, I'm gonna build a rip off of Screen Studio and release it. And it's like, be a little have a little faith. You gotta dream a little bigger.
Aaron Francis:Like, set your ambition a lot higher. And I think Brian's right when he says care. You know, I don't speak for Taylor obviously, but when I read that, I internalize that as we have to make cloud the best that we can possibly make it. We have different opinions than Vercel, Heroku, Cloudflare, Render, Railway, whatever. And the market will decide who is the best.
Aaron Francis:And different parts of the market will decide Laravel Cloud is the best, Heroku is the best, whatever. But if we look at our own product and say, boy, this could be better, but it's good enough. I think that's where it falls apart. And I think that's where a lot of people don't take it that last 510% is they just don't have that care and attention or maybe even the ambition to like make this thing as good as it possibly could be.
Justin:Mhmm. Yeah.
Jordan:Maybe some discipline too. There's so much to do that it's it's a lot easier to get to 80% and move on to the next thing. Do that today. Easier.
Justin:Yeah. Yep. Can I give you guys a a personal example? I've been building for the first time, I said, I'm gonna try to build a podcast website theme. And so I've been working on this theme called Calm.
Justin:And built it, you know, designed it, built it, etcetera. And what I'm seeing is a lot of people doing stuff like this. And for example, this icon I noticed was just slightly higher than the search text beside it.
Brian Casel:Oh, that stuff
Justin:drives And me there's lot of details where they're just like, you're not you're not caring about the details. And for me now, I have I can, like, be the design director and say to the underling, which in this case is Astra, hey, like, let's clean that up. That's not that's not good enough. And just going through page by page. Here's another good example.
Justin:This page here is what got generated for the Patreon page. This is not good enough to ship. Like, there's a lot of things I want to fix on on this page. I don't want the avatars to look like this. I want them to be rounded.
Justin:This section up here doesn't make sense. And so I've been really enjoying using AI to kind of craft things, to really care about the quality, to test it out in all scenarios, and then say, okay, how can we make this better? But I think the the problem with AI is that a lot of people are just like, you know, one shotting it or whatever, or they feel like it makes it it's like a labor saving device, and it's gonna, you know, reduce the amount of hours you gotta work in a day. No. You gotta care.
Justin:Yeah.
Aaron Francis:And I
Brian Casel:I just went through the exact same process with BuilderMethod. Like, I just this this morning, I launched a new buildermethods.com. And that that process of of being the it's it's not even like the creative director. It's more like just dictating small, tiny change after, like rapid fire, like move that search icon one pixel up. We gotta change the the max width on on this section to compare to that section.
Brian Casel:And we gotta tweak this font size or, you know, it it at a certain point, gets like, I'm I'm being too much of a perfectionist on this, but but it it is like, that is the creative loop that I enjoy a lot too. It's like the the deep planning where I'll spend a whole day just writing up a a big spec for something, and then there's then it goes and builds that. It gets it to 95%, and then then I'll spend a few hours just rapid fire, like, tweak this, tweak that.
Justin:Yeah.
Jordan:Change that.
Aaron Francis:Yeah. I think Jordan, the question about Rosie, I think if you decide so like you can't compete on every vector at once, right? So like SavvyCal, Derek Rimer has decided I'm gonna focus on appointments and whatever his new one is for doctors. I guess that's appointments. Appointments and meetings, those are his two.
Aaron Francis:He can't compete on every vector at once, but he can decide, no, these are the two things where we're gonna be best in class. We're gonna be, you know, world class level of experience, design, functionality, whatever. And I think that comes back to like, as a founder or product owner, having a freaking point of view. Right? All these all these people that are one shotting stuff which I don't believe first of all, don't believe that at all.
Aaron Francis:But all these people that are one shotting stuff have no point of view. They just have a new hammer and they're like watch me bang some nails in. And I'm like, what's your point of view? Like, what what do you believe in? And like, I feel like Taylor has always had deeply held beliefs and the age of AI hasn't changed that at all.
Jordan:Mhmm. It feels like an accurate reflection of the population. Most people Yeah. Don't have the discipline, don't have the care. Yeah.
Jordan:Taylor from day one has and therefore it it just continues to perpetuate itself. It's also partly responsible for the success that got it there and
Aaron Francis:that's Yep.
Jordan:So it all it all kind of is shaking out
Justin:the right way regardless of Taylor
Aaron Francis:for the past fifteen years, Taylor was a filmmaker. Do you think he'd be good at doing AI movies? I think probably so. Mhmm. But I see all these other people that are slopping out AI movies and I'm like, this is literally the worst thing I've ever seen.
Aaron Francis:And I feel like it's the exact same thing but in software.
Justin:Yeah. I I think, Jordan, the other the other thing that intersects with this is that Bending Spoons episode we were talking about. I don't know if you've have you seen it yet?
Jordan:I read some of it. I saw a few clips, but
Justin:but please go go into it. Okay. Well, David Senra in that has this quote. And to me, this is just this is like the Taylor quote but distilled. If you have to pick the person that is the smartest or the person that has the most experience or the person that wants it more, you always pick the person that wants it more.
Justin:Extreme ownership. I think this is what it's about. It's like, who wants it more? And that that that's where I get kind of compelled to be like, I want to build the best product. What does that mean?
Justin:Well, I'm it means that we have people switching to us from other competitors saying, you guys are the best customer service. You guys are the best product. You guys are the best experience. And I want it. Like, I want it hopefully more than all my competitors.
Justin:And I think that makes a real difference that, like, who wants it more? And that's you, Jordan. You're like, I want it more.
Aaron Francis:Nobody wants nobody wants it more than Jordan.
Jordan:So can I give I I I don't know if this is a question for Aaron or or whatever? So I am the only one of the group here who has not handcrafted a single line of code in the last fifteen years.
Justin:Mhmm.
Jordan:However, in my defense, I have supported a lot of Laravel projects. Cardhook one was Laravel. Cardhook two, Rally, Rosie, billions of dollars processed. We've always just loved Laravel the whole time. As a non technical person viewing technical products, especially like frameworks, infrastructure, these big like Ruby, like these big like
Justin:Sure.
Jordan:It's like these clouds out on the Internet that people love and are attracted to and define themselves by.
Aaron Francis:Yep.
Jordan:What I wonder, what I would be thinking a lot about in your shoes is like this this arc. This narrative that seems to happen to these products And some go from this beloved niche thing into this bigger beloved thing.
Justin:Mhmm.
Jordan:And others go from niche and then something goes like wrong and they become the villain in some way. And it's a natural arc for for big companies but it seems to be very difficult to deal with for development products because your audience is a bunch of developers who are rabid in their opinions and sharing their opinions. So So it's like, I think right now, Laravel Laravel's going through this transition
Justin:Agree.
Jordan:Of this project that everybody loves and it just supports everybody. It's like there's nothing to not like. If you don't like an open source project, it's your problem. Not not the project. But as it's transitioning into this this other master that that comes into play, like, I I would be thinking a lot about that transition and how to, like, maintain the the the love, basically.
Aaron Francis:Yeah. I mean, a 100%. I think we have we have a few things we have a few things going for us and a few things we need to be cautious about. One thing that's going for us is the history of PHP itself. It's never been sexy.
Aaron Francis:It's never been cool. It has always been a blue collar working person's language. And therefore, it has attracted a lot of people that want to build things. Not a bunch of people that wanna talk about the newest frameworks, the newest tips, tricks, whatever. It's like when you get together at Alericon, it's like, oh, what's your side project?
Aaron Francis:What's your thing? What are you building? What are you doing? And it's less about like how do we destroy these other like, you know, how do Vercel and Cloudflare destroy each other? It's like, I don't really know, but I'm working on something on the side, you know?
Aaron Francis:And so I feel like that that legacy has served us very well. And I think that has I think that's coincided nicely with Taylor's history. Dude lives in Arkansas. Like and so we have, like, we have this, you know, in Taylor's realm, the man of the people feel. Right?
Aaron Francis:Yes. He he bought a Lambo from his bootstrap from his bootstrapped PHP software, and everyone in the community was like, yeah, we're winning. But he's still That's so so surprising. I can't
Justin:believe, like, that to yeah. That was surprising to me that people are just, like, on his
Aaron Francis:marketing we ever got. Cool. But, like, he doesn't he doesn't live in San Francisco. We don't have San Francisco offices. And in some regards, we're not drinking the Kool Aid.
Jordan:Right? Was about to say he doesn't sound like he's actively swimming in his own Kool Aid,
Brian Casel:which Yep. Correct.
Jordan:Is part of what people see through and are like, I don't like where this is going. Yep.
Aaron Francis:Yes. And you look at like, I won't name any names, but you look at a lot of like the Silicon Valley CEOs and you're like, dude, what the hell are you talking about? Why has no one pulled you aside and said stop doing this?
Jordan:Yes.
Justin:And you
Aaron Francis:know Taylor stays in his life he stays in his lane. He talks about craft. He talks about product. He talks about discipline. And so I think a lot of that comes, you know, from the top down.
Aaron Francis:But then the things that we have to be careful about is as Cloud hopefully, you know, he said publicly at five x, as hopefully at 50 x's, you know, in the next many years, how do we not become the evil mega corp? And I think a lot of that does fall on me, obviously Taylor as well. But I think a lot of that falls on me with my role as actually VP of marketing and community. And so I also have a dual mandate that I have to care for the community. And it's a good thing they got, you know, emotional little Aaron Francis in there because I don't want to leave people behind.
Aaron Francis:Right? Mhmm. I want to like I wanna set a vision for something they can aspire to and tell the story that the the salt of the earth, blue collar, working class PHP developers are going to be the center of this much larger community moving forward with the goal that all of these good characteristics imbue the newcomers, in the community. Right? So as we welcome in Rails and Django and JavaScript and all of that, the nugget around which we're building is this group of people that loves to build things.
Aaron Francis:This like 50,000, whatever it is, ambitious builders who have been downtrodden for so long. My goal is to say, listen, you are now the center of this big movement. Isn't this good for you? And if it's not good for them in reality, then I'm just lying. I think it is going to be good for them in reality.
Brian Casel:Mhmm. So Justin and Aaron, you guys just got back from a bit of a bit of a trip over to Vegas. I'm I'm always a fan of a of a retreat with a, you know, small group. Was this like a obviously, it's I'm sure it was like super fun, great time. But like, was there any element of like, let's talk about business or strategy or or think or maybe just personally for for either of you, like, was it a chance to refresh and step away?
Aaron Francis:Yeah. Just got one American flag tattoo. Then then we started then we started talking business.
Jordan:I'm sorry. Did someone get an American flag tattoo?
Aaron Francis:Just Justin. Yeah.
Justin:Pretty close. Pretty close. I came there and I'm like, I'm not big on America. I'm not big on Vegas. And by the end
Jordan:of the year the maple You leaf and just went over it with an Bunch American
Aaron Francis:bunch of frat ropes and carried it around,
Justin:you know.
Aaron Francis:Justin got invited to, like, four weddings.
Justin:So it's I was yeah. I I got very social at different times. This is my first time. And Taylor has invited me before, and I I just could never come or didn't make it a priority. And this year, I was like, I'm gonna skip the Podcast Industry Conference in New York and go to this instead.
Justin:And it was just like what the doctor ordered. It's kinda like the perfect thing. Like, Taylor gets a cabana every day, and then we just sit in the cabana most of the time and talk. And then every once in while, someone will be like, alright, here's a topic for the cabana. And then
Aaron Francis:Mhmm.
Justin:They tell something. They talk about some challenge and then get, like, basically the most distilled, honest, raw advice back.
Brian Casel:Drunk advice.
Justin:Yeah.
Aaron Francis:Yeah. And very, very dispersed advice. Right? Yeah. So each person there's, like, 10 or 11 or 12 of us.
Aaron Francis:And, you know, each person has this really, like, pretty straightforward take, but oftentimes, they're very different. And so then the participants start fighting with each other, which is great. That's the best part. Justin gave some advice, and I was like, that's insane. Don't do that.
Aaron Francis:Listen to me instead. And so
Brian Casel:Sounds like a big, snow, tiny comp.
Aaron Francis:It's exactly what it is. It is a give us a venue in which we can sit around with nothing else to do, and we'll just see what happens.
Justin:Yeah. But but and also, you I think everybody on Earth needs to experience a casino with Ian Landsman because that guy is a maniac. He's like Maniac. He's, like, complaining, like, none of us brought cash. And so, like, he's playing craps.
Justin:He's like, Jackson, what are you doing? And I'm like, I don't know. I don't got no cash. And he's like he takes out a big wad of cash. He's like he's like, come on.
Aaron Francis:Get in there. Big wad of hundreds. He's daddy Warbucks. Yes, man. He does not f around when it comes to casinos.
Justin:Which I proceeded to I've lost $400 in, like, five minutes. And, you know, it's like now okay. Now I owe I owe Ian $400.
Brian Casel:Oh, I bet that was a scene.
Justin:And he's like, else would go to bed. We're all, like, pretty boring, like, go to bed early people, and Ian would stay up till 03:30 in the morning.
Aaron Francis:Literally until three in the morning.
Justin:There's, a poker game at two. I gotta go make it. So yeah.
Aaron Francis:Justin and I played blackjack together. Yeah. We sat at a table playing blackjack together and
Justin:Then no fun table.
Aaron Francis:The no fun table. The dealer told me to be quiet because it was hurting her ear. And I'm like, this is I am in a casino in Vegas gambling, and you're telling me to be quiet? Justin came out even. I lost all my money.
Justin:So Yeah. Yeah. I I went a little bit back. Yeah.
Jordan:Your first time though, Justin, it's a pretty interesting weird place. Yeah. It's genuinely a strange place.
Brian Casel:Not your first time to Vegas, was it?
Justin:Not my first time to Vegas. No. No. No. No.
Justin:No. Because this was, like they started doing it after MicroConf left Vegas. Oh, yeah. So I've been to Vegas from I think the last time I was there was MicroConf, though.
Brian Casel:Was I
Justin:I think I I had a bunch of thoughts, like, out of it, like I mean, just being around that kind of group where people either are either ambitious or, like, really committed to their craft or both. And to be around that energy all the time, like, Yeah. I feel like this a
Brian Casel:good transition to to, like, an update from you, Justin. Like, did you did did the trip mark any change for you as you think about transistor the rest of the year or anything like that?
Justin:I mean, the other thing for me is basically everybody there was a transistor customer. So to get unfiltered I've done
Aaron Francis:almost twice. Yeah. Yeah. So,
Justin:I mean, I got a lot of good feedback there about what people care about. Unified analytics being a big one. There's even, like, some out of left field ideas. Like, Ian had this idea for Overcast has ads and Pocket Cast has ads. Could we be the unified dashboard for promoting your podcast across everywhere where you can promote a podcast?
Justin:So there's stuff like that. I I think for me, it's like I want to I mean, just okay. Here's another example. John O'Nolan's there with Ghost. I might be slightly incorrect about this, but I think Ghost and Transistor hit $2,000,000 in revenue about the same time.
Justin:Ghost is at 10,000,000 revenue now. And Transistor is not. And so just that gap of going, hold up, you know, sometimes comparison is the thief of joy. And sometimes comparison is like, it burns a fire on your, you know, it burns a fire on you. Like, gotta I want to I there's no reason we couldn't have been there.
Justin:So why? Why did we not get there? And John did. And I just kind of like that pressure. And then I mean, then I John and I were in the airport together for like six hours.
Justin:So it's just like back and forth. Like, let's just talk about, you know, what
Brian Casel:What do you think that they did right? I don't I mean, I I don't know that I necessarily agree with like, there there's so many, like, external factors.
Justin:Oh, there's so Yeah. When
Brian Casel:you I mean, what what John has done with with Ghost has been really awesome to to watch from afar. Like, what do you think?
Justin:I mean, what one thing is I think if we'd released video earlier, we would have we would have I think we we pursued too many tertiary features. There's a big this is something else we talked about. There's a big power law between the most important thing to build and then the second or third most important thing to build. And you just lose a lot of momentum that way.
Jordan:Are you able to say that directly to your co founder?
Justin:Yeah. Yeah. I think when I got back, I reported back and just said, like, to the whole team, like, I this was inspiring for me. I want to, like, I want to get more aggressive on this stuff and focus on stuff that actually moves the needle and kind of be more relentless about it. So, yeah, I I think it's actually nice to go to a a retreat where it's just me and not anyone else from the team.
Justin:Just like, okay, now I can speak freely. People can speak freely to me. I'll say one more thing about it from my side is it has that summer camp, that church summer camp vibe, which is like you bit just become best buddies with the people in your cabin and talk about, like, tons of personal stuff too. And I find that is almost exactly the balance you want. So, yes, business is important, and we're all doing this stuff.
Justin:And then you actually hear about what's really going on in people's lives and the real raw human struggle of being a human, I find that that balance is kind of perfect for me. And I don't I don't get a lot I mean, I get some of that at podcast industry conferences, but not like this. These are yeah. People have there for a
Aaron Francis:long time. Like, that that in person time lays the foundation for the rest of the year. Right? Yeah. So then then you're DMing, you're chatting, and it's just like, oh, yeah.
Aaron Francis:These are these are my summer camp friends. We get on AIM. We talk about stuff, and then we get back together at random conferences or at this specific thing. And it just brings that level of, like, trust or authenticity way up because you're like, yeah. I just spent four days with these dudes, and now I can remind them, you know, this part of my life sucks.
Aaron Francis:And they're like, oh, yeah. I remember that.
Brian Casel:Yeah. Totally. The you know, Big Snow for me has been a a big thing over the I think more than ten years now.
Justin:This really made me wanna go to Big Snow, actually. I I got an email from Brad, but it's not but, yeah, I it sounds so similar.
Brian Casel:It's great. What what's what ends up really being it it multiplies in value over time as you start to go back within and and the same group of folks that because because that's when those the the more personal side, like, you know, we we have a more, like, formalized like, everybody gets an hour to present a a topic or ask the group, you know, a big question. And there's a lot of business talk in there, but, like, over the years, some of those have veered into or completely changed over to, here's what's happening in my personal life that I can't talk about publicly, but we can in this room. And that that's when it's that I mean, it makes me wanna do another small like, the small retreat thing, I'll I'll never get enough of them.
Justin:Oh, yeah. It's it's the best.
Brian Casel:It's the best. Yeah. Yep. Yeah. What else we got today?
Justin:I I gotta talk about Jev because
Aaron Francis:Jev. Jevvy baby.
Justin:Jordan Jordan Gal and I are like the boomers in the room who I had an explainer. What is going on?
Jordan:Look, I'm
Brian Casel:just starting to, like, wrap my head around
Justin:it.
Brian Casel:I haven't been able to use it yet though. But
Jordan:If I jumped immediately onto everything I saw on the timeline, I would be like a stooge for the marketing budgets of these VC backed companies. Like, I I can't do that. What did I just see today? Just you wait a minute.
Justin:I saw someone post that says the only way to keep up with all of this AI stuff is to be unemployed. Correct.
Aaron Francis:I have. Perfect. However Perfect. As as an employed young man, I have tried JEV just very briefly.
Justin:Okay.
Jordan:What what you're doing? What
Aaron Francis:is JEV?
Jordan:Can we start with what it is? Fantastic.
Aaron Francis:Okay. We can start with what it is. So traditional LLMs spit out words, right? You say, why is the sky blue? And it's like blah blah blah blah blah.
Aaron Francis:Four four ten thirty paragraphs, right? JEV on the other hand is classification or decision engine. So you send it you send it a question and a structured set of possible responses. So like, is the sky blue? Yes.
Brian Casel:Yes, ma'am. That's it.
Aaron Francis:You can say like, is this email, should this email or which department should this email go to? Here are the seven departments you, Jev, are allowed to choose from. And it comes back with a confidence score of, you know, it should go to billing 98% confidence, support 2% confidence, sales 0% confidence. Right? Mhmm.
Aaron Francis:So that's already interesting because it cannot it literally cannot produce free form text. And so all of these hoops we've been jumping through to make these dictionaries produce JSON, we just don't have to do. It just produces JSON now. But the important thing or the interesting thing is it can do it in fifty milliseconds. And it can do it for like fractions of a cent.
Aaron Francis:And so you can run you can run 10,000 emails through it in maybe a couple of minutes and you'll get charged $23.04. And so that is where people are starting to look at it as like, holy crap. I could literally route I could route game time decisions through JEV, get a confidence score back and use that as like a branching workflow. Right? So Jordan for you, you could have somebody call in and they say, hey, I need to talk about x y z.
Aaron Francis:And before you send that to the LLM, you could send that in a hundred milliseconds to Jev and say, where should we route this? And they say, or you could say, is this person mad? Or is this person happy? And that Jeff would come back like that for $0 and then a hundred milliseconds and say, I'm 98% sure this person is mad and I'm 98% sure this call should be escalated to whatever the head plumber.
Jordan:Right. Right. That that can trigger, like, whatever we want it to from the agent to say Including did you wanna speak to a person?
Aaron Francis:Right. Yes. Okay.
Brian Casel:Yeah. Like like if you I'm just starting to catch up to this today as I'm reading up on it. And it's getting me more and more excited about possibilities of using it for builders. Like, if you think about, it's kind of like an unbundling, if you think about it. Like the stack of what we do with AI now.
Brian Casel:This is especially for like software applications, but I think the same could be run through just regular agent skills and stuff. Like Aaron was just describing that when you put in when you have a use case for AI, like like in my case, I have a bunch of really complex skills that help me with content development. But sometimes I'm developing a video, sometimes I'm developing a newsletter, sometimes I'm doing a tweet, and sometimes I'm doing and then, as I do all of that, I need it to go through a pretty complex process of voice training and removing slop and then sending me a review gate and like all this different stuff. And so a lot of the mechanics that go into that are really decisions. Yep.
Brian Casel:That should happen a lot. Like the way that it currently goes is like you send it to OpenAI or you send it to Anthropic and then the model's gonna receive like, is this a video or is this a newsletter? I'm gonna reason about that. And then I'm gonna spit back a response when really that should be like an instant decision. And then the next one is like, does this, align or or let's let's give this a score against Brian's training and and voice training.
Brian Casel:You know, let's give it a 75 out of a 100. Like, that should be an an instant thing, not a not a drawn out reasoned part of the process. Only the very end when it comes to actually drafting or or editing something. Like, that's something where a Claude or a ChatGeeBT can spit out the actual words. But all the logic, all the deterministic, you know, logic that happens before that.
Aaron Francis:So Jordan, try try this one. So somebody calls in, they're having a conversation with Rosie. They say, okay, great. I wanna book the Monday at 10AM. I imagine, I don't know your stack, I imagine you send that back to some LLM, whether it's low effort, cheap, or, you know, model frontier, Yep.
Aaron Francis:Regardless, you're taking, I wanna book that Monday at 10AM, and you're sending it to a thinking machine. What you could do is put Jav in the middle and say, alright. Here's the sentence I just got. Here are the available tools within Rosie. One of them is book appointment.
Aaron Francis:Mhmm. You tell and one of them is actually forward to LLM because, like, I don't I don't know what this is saying. So if you run it through this JEV classifier and it comes back with 99% book appointment, then you can skip the LLM altogether and immediately book the appointment. You could do another Jev call and say like, what time what what date and time did this person say? After they say after Jev says, that's a book appointment tool call.
Aaron Francis:I'm confident. Great. Classify it again. What's the date and time? Great.
Aaron Francis:Book it. And within maybe a hundred and fifty milliseconds, you can respond to the user. Great. We booked you Monday at 10AM. Without ever touching the the big model that's gonna take two, three, four seconds to respond.
Jordan:Right. Our our challenge is human conversation is unpredictable.
Justin:It is. Mhmm.
Jordan:Everyone like yeah. Can they
Brian Casel:can evaluate that and say, like, this is 98% certainty that that's what the person means.
Jordan:That's interesting because you can throw at it something that it still needs to figure out. Is it the response is more structured?
Aaron Francis:The response is completely structured.
Brian Casel:Okay.
Aaron Francis:Can't generate free form. It can't generate audio. But what
Brian Casel:it It's a can preliminary step. The final step that the end user sees is is what your clods and chat GPTs will will generate in terms of words. But all the decision making that goes up, that that happens before your system decides what to write to the user, there's a there's probably a 100 different decisions that need to get get made. Or like decisions or evaluations or, you know, yeah.
Aaron Francis:It can short circuit some of your flows. If Jeb says, I'm above 90% confidence that I need to call this tool, you can skip the LLM call and just call the tool based on Jeff's confidence and then tell the LLM, here is the result of the tool call and here's what the user said. We ran the tool automatically without involving you. Now I need you to generate the text to respond back.
Justin:You go
Jordan:from there.
Aaron Francis:So you've short circuited the decision layer by using this super cheap model that is very fast.
Jordan:Yeah. Interesting.
Aaron Francis:And then you told the big dog, hey, we already did it. What should we say now? Right. Right.
Jordan:Just figure out what to say. What you're good at is words. Yep. So come up with the right phrasing that it sounds the way we want it to based on the settings and Yep. And so on.
Jordan:Who who who made Jeff? Who who's this from?
Brian Casel:This this guy, like former OpenAI Former OpenAI, new
Jordan:new Individual developer? Not not like a company that released it as a product?
Brian Casel:He started a company after apparently, he was one of the integral contributors to the original ChatGPT. Okay. Left Open AI.
Jordan:Noticed this needs to exist.
Aaron Francis:Evolutionarily it's very strange because we jumped to LLMs, we jumped to like human level intelligence and so many workloads are solved by decision making. And so we went straight to the very top, whereas like classifier engines would have actually served our needs a long way. You imagine all the people in the world that are feeding customer support into fable or soul and being like, is this person mad? And it's like, well, that just cost you 8¢ to answer that question.
Justin:You could have done it with a
Aaron Francis:classifier engine for 0.008¢.
Brian Casel:And and that's what's really interesting because it's like all the, you know, your your chat GPTs and clods have been have been sort of striving toward like, let's just create this like, whatever, like this God. This this god like product. Right? When when really JEV is aimed at the product makers. Like, we Yes.
Brian Casel:As product builders can use this to make excellent products for our customers. And it's like the the big labs have just sort of skipped past all that. And so that that's what's really exciting to me about about the possibility of of really starting to use JEV in a big way for builders because it's like we now this is an actual tool for us to serve actual customers. We don't have to start to think about like, what can we make that's not just gonna be kinda eaten up by OpenAI next year.
Aaron Francis:So I'll tell you I'll tell you how I use it because this will help make it concrete. So Solo, which is my desktop application, soloterm.com, it's very good. Solo is basically a terminal. Right? More or less, boil it down, it's a terminal.
Aaron Francis:Inside that terminal, you could run Cloud Code, you could run AMP, Codex, Muse, whatever. Great website. So I don't know, and I don't super wanna know what the users are running. Right? I want to be a terminal that says run whatever you want.
Aaron Francis:That, that exposes one problem, which is when is that agent waiting on me? Right? I'm the human. I'm the captain now. When is the agent waiting on me?
Aaron Francis:When do I need to give it attention? And so what I have been doing is I've been training my own model, my own little classifier. This is pre JEV. I've been training my own model by capturing all of these terminal grids locally, feeding them through this model training pipeline, and getting a very small model out the other end that says this one's idle, this one needs permissions, this one is done, you don't have to attend to it. I've been training this model.
Aaron Francis:I have all of this labeled data of terminal grids and what the ground truth is, whether it needs attention or not. I have all this labeled data, I got Jev, I hooked it up, and I said, hey, grab actually told, Codex, I want you to grab 15 tricky grids that have already been labeled. Grab all of those grids that I had spent so much money on labeling Mhmm. And I want you to feed them blind to Jeff with our optional, like, labels that we have been applying. Feed them blind.
Aaron Francis:Say, is this idle? Is it waiting? Is it permissions? Is it whatever? Jeff got all 15 of them right for all 15 for less than 1 tenth of a penny and in less than like for 15, it probably took two seconds total.
Aaron Francis:And so I've been I've been busting my butt to train this new classifier engine and then Jeff comes along and one shots it and says, yep, that idle needs attention or that terminal is idle. That one needs attention. That one's stuck on permissions. And so it's just this like instant classifier of basically anything that you can throw at it.
Brian Casel:I think of my I
Justin:think your developers are gonna like it, Jordan, because you they've been dealing with nondeterministic models. So it's like so in one case, the caller gets this response. But in another case, the caller gets another response. Well, this gives you a new tool set to say, no, we're gonna make that way more deterministic. Yeah.
Justin:It's it's way more how programmers think. It's like, this happens and then this happens. You can create more rules. So I think once Rock looks at it, he he's gonna, like That's their
Brian Casel:love it.
Aaron Francis:Tell tell Rock it's in the Laravel AI SDK and he can use
Justin:it already. So just tell him he's good to go.
Aaron Francis:Oh, nailed
Jordan:that. That's the same. That's beautiful. We we are undergoing more review of how our system works over the last few weeks than we have in the last year and a half.
Justin:Mhmm.
Jordan:Because things are up until now, the stack has required three pieces. Transcriber, LLM Yep. Synthesizer.
Aaron Francis:Yep.
Jordan:And and ChatGPT Live just came out.
Justin:Mhmm.
Jordan:And it is the first glimpse of what it looks like when you put those three together. And we've gotten more into Luna as as as a model. And Luna can sit alongside Chat GPT live and it can still be used in that way. And it's it's like the first time we we really might move toward a new way things work. Not surprisingly, something new comes along like JEV and all of sudden that might kind of, you know, get incorporated in.
Jordan:And I don't know. I'm curious what you guys think of this. Yeah. I think a lot of a lot of founders, developers, people in companies like ours right now are when you see that sort of thing, you immediately get excited about, okay, this this is well, this can make our agent quality better. Right?
Jordan:Up until now, our entire business rests on the truth that a customer comes in, works with Rosie, and says to themselves, this is not amazing, but it's good enough to solve my problem. And soon, there will come a day when they listen and work with Rosie and say, oh my goodness. This is incredible. I can't wait to use it.
Aaron Francis:Yep.
Jordan:Mhmm. Obviously, we wanna get there. Yep. And then the second thought is like, well, isn't everyone else gonna get there?
Justin:Right? If you assume that
Jordan:your competitors yeah. If your competitors and all these other products are not stupid and they're ambitious, how do you think about that from, you know, not just Rosie point of view, but like, whatever you're working on.
Brian Casel:I just think about it from a product standpoint, end to end product. Who's who's delivering the best? Back to what Taylor was saying, the best possible solution to this problem that we know exists in this market. That's always gonna be the game.
Jordan:It's not just agent quality. We had this interesting experience that almost like drove some This forms
Brian Casel:is an opportunity to really speed up or improve or the improve the intelligence, the the overall cost. Like, there's so many levers within, like, if I think about Rosie, it's something like this.
Jordan:Yeah. There's you know, it's it's so easy to overemphasize that part of it. The for us, it's agent quality. Right? It might be like video output or it might be something else, but it's so exciting about how good that can get and what you can do there.
Jordan:It's very easy to kinda get pulled into that. But I I think you're right. The bigger picture is beyond that there there are other things. There's the full experience. Are you solving the problem?
Jordan:Is it easy to get to the solution? Do you have a match between who you're marketing to and who the product is actually best for? There's a perfect example. Like, as I'm just thinking about this this week, a few days ago, I think maybe yesterday, Eleven Labs, which is the biggest, most well known, best funded, independent, non lab AI voice company
Aaron Francis:Mhmm.
Jordan:Comes out with reception.ai. Right I hear
Justin:about this.
Jordan:Right at SMB. So, 100% direct competitor. So, of course, what do we do? We we go sign up. And it's amazing.
Jordan:The aid their Asian quality is out of control. It's so good. It's great. But the rest of the product doesn't, like, make sense.
Aaron Francis:Of course not.
Jordan:Like it's like overbuilt. Like they expect you to kinda use their like like mini CRM and their scheduling and their calendar and there's something else and there's some integrations but once you get in there, it all looks beautiful but you don't know what to do next. It it is very comforting
Justin:Mhmm.
Jordan:To to almost see that and be like, okay, unlimited budget and incredibly talented people. You you you still have to get it just right on all these other parameters. It's not good enough to just be like the product, the end product. In this case, the agent quality is so good. It's not enough.
Aaron Francis:Yes. I think it's a to make, like, to make the Taylor tweet more visceral, it's like who can eat the most pain on behalf of the customer? Who can chew the most glass so that the customer doesn't have to? Eleven Labs has put out something in their sweet spot that leaves a lot of pain for their end users to absorb. If you and Rock can just eat all of that pain so that it integrates with whatever system ServiceNow, I don't know what is in that world, but if you can eat the most sin for those people, then you win.
Aaron Francis:Like that is what we hire products to do. That's why in my opinion, Liravel Cloud is working because we've eaten so much pain. Like the fact that we can scale stuff down to zero and spin up databases and do preview environments, like that is all technically possible on any one of those other platforms that is basically just a trench coat on top of three dockers. Like, that is technically possible on other platforms, but we moved the pain line back to us and said, we're gonna eat that. That is
Justin:our problem. Yeah. And for you, Jordan, it could be, like, you make a call to Rosie, and you're like, Man, that first response took a second. Like, that's not good enough. And then you go back to the team and go, listen.
Justin:There there's ways for us to solve this problem. Maybe using something like Dev, we can get that first response to just instant and organize the user experience so that the first answer you get is just it's just instant. Like, we can classify what the person said, hello, or whatever, and give them an instant answer, and that already improves the experience. I think a lot of this stuff is gonna be about who cares about the customer experience more. I mean, there's other ways to win.
Justin:There's other ways to be the person who wants it more. Like, you could say, we're gonna just be the best at doing enterprise sales. That's also care less about it.
Jordan:Yeah. There's there's there's different approaches to it. I I think a developer type product is is very pain oriented. In in our universe, it is like who's gonna help make it the most feasible for the customer to succeed or or achieve what they're trying to. Mhmm.
Jordan:I saw an interesting conversation on Twitter after the Eleven Labs announcement. And one of these, like influencers that helps people like sell AI consulting to regular businesses was like, you know, now it's so easy to set this up for people and charge charge $5,000 because it's so easy for you to do and someone else came in and was like, that's not gonna work. It might work for a little while.
Brian Casel:Right.
Jordan:But you can't arbitrage that for very long.
Brian Casel:That's the thing with I think with that's a little bit different in the story of AI with Jev compared to the astros and the fables that that have come out. Like where you see everyone tweeting about like, oh, look, I I one shotted this this crazy video game with with Astra. Look how cool this little toy is.
Aaron Francis:You nailed it.
Brian Casel:But that but that's not a real pro like, you know, the the like we just Aaron said earlier, like, they're they're, these are not like real products that are actually reaching real customers. Where Jeff comes in, it's like, this is a tool that like real product people who are serious about building a cohesive end to end product for real customers to solve a problem, can actually take advantage of the benefits of something like this. Whereas like it's not something, because if I think about just your typical quick like agent skill, it's like you can't just like plug in Jeb, you sort of need like a software stack. Yep. And like some logic behind it and a product thinker to put the pieces together, know.
Brian Casel:Or a team to put the pieces together, you know.
Jordan:Jeff gonna make any money?
Justin:Presumably. We're talking about that in the chat. Some Right. Right. Some people are like, how long is it gonna be before the open source model comes
Jordan:out It's and a bit brutal in that way. Right? Like this this is like an advancement. This is well, okay. Maybe it's not an advancement the way the initial big huge advancements are but this is filling in these areas that we didn't know a few years ago even existed.
Jordan:But this thing is necessary and there are countless more inventions. It's not innovation. It's it's literally an invention. Yeah. There's countless more that are on the way that we can't even think about right now.
Jordan:But as this stuff develops, we start to encounter these issues, price issues, speed issues. It's it's it's amazing what's gonna happen. But like, are are are these companies gonna be able to to to make it work? It it needs a little response.
Brian Casel:Wonder like what the impact is gonna be on the major labs and the and the value proposition. Like, because like, will using if if JEV starts starts to to come come into into use in a major way in most applications, that's gonna take away, literally take away the the amount of tokens that we need to buy from Anthropic and OpenAI. And so then if if we are relying less on on the frontier models being as brilliantly smart as they are selling us on. If we can if we can outsource most of the intelligent decision making to a JEV or whatever next open source version of that is, and we and at the end of the day, we only need a a good word stringer together in model. We could use any Chinese open source for that.
Brian Casel:Then Mhmm. That's like real pressure against the value prop of paying for for OpenAI subscriptions.
Aaron Francis:You know? I think, Jev, in terms of in terms of, like, request workloads, like the amount of requests, I think Jev and whoever succeeds after Jev, they're gonna eat the lunch of the big labs in terms of requests. Obviously, it's like a 100 times cheaper, but you have to assume they're somewhat intelligent and now like their economics work for them. But this is like if there if there had been no OpenAI and no Claude and JEV had come out three years ago, people would have lost their minds much like they're doing today. Because you can imagine the power, like, the the way that the developer's wheels start turning when you show on the screen an if statement that says, if this customer is mad, then do this.
Aaron Francis:And you're like, woah. How does it know if the customer is mad? And it's like, well, it's a API call, a hundred milliseconds, and it costs less than less like less than half of a half of a half of a penny. And they're like, holy crap. This changes my entire life.
Justin:Yeah. And so
Aaron Francis:I think in terms of like pure number of requests, JEV and its successors are gonna just absolutely decimate the big labs. They're not gonna make as much money or I don't know, their profit margins may be fantastic, but net their dollars are gonna be lower because it is so much cheaper. But then again, I'm gonna send a thousand times the requests to Jev because it's so much cheaper. So I
Brian Casel:don't know. But this is what we But the reason why we we pay top dollar for the frontier models and multiple subscriptions to use them all day every day is because we need their intelligence. But if we can unbundle that intelligence to these other JEV like models, then we don't even need the end user model, the one that's stringing together the words. It doesn't matter how intelligent that is, as long as it can write some words based on the data we we feed it. Mhmm.
Justin:So then
Brian Casel:it's like the whole value proposition kinda falls apart for, like, you know, the fables and the astras. Like, we don't actually need to pay
Aaron Francis:for that. Stays it stays together for writing code. Like, JEV Jev cannot by its very nature write code. But if you're building and this is why I think it's interesting for Jordan specifically. If you're building intelligence into a product, I bet Jev is gonna eat like 6070% of that for a fraction of the cost.
Jordan:Like, I'm I'm almost worried about the labs now. I I feel like the labs were are
Justin:They'll be fine.
Jordan:Best friends. Well, I'm I'm worried they're definitely gonna be fine. But they're almost like too big to fail. Like, it is not it's not even an exaggeration to say that there's a large portion of the American economy Mhmm. That is riding on these on these companies being successful, their IPOs being successful.
Jordan:They're doing all this weird EA stuff talking about killing humanity like a month before an IPO. Like, this is weird. This is a weird situation.
Justin:So
Jordan:But isn't isn't, like, the the success of Open Router, doesn't that just, like, point to a reality of some commoditization among these labs?
Justin:Mhmm. You know you know what we call this, Jordan?
Aaron Francis:We call this capitalism, baby. Drive it down. Drive it to zero.
Jordan:I'm with you. The problem is where it crosses over into crony capitalism is when very, very successful capitalistic companies use their power and influence to to retard the capitalist process.
Aaron Francis:But the beauty of the self healing mechanism that is capitalism, which you and I live by Amen. The beauty the beauty of that mechanism is you get Zuckerberg coming out and saying, yeah, you guys should just make safe products. Like, we we released Muse and we just made sure it was safe. You get Satya Nadella coming out and saying, this is stupid. Like, We don't need to pace the frontier.
Aaron Francis:We need to win.
Jordan:Feel free to pace yourselves. Don't tell us what
Justin:to do. That's fine. I don't care.
Aaron Francis:You have Coke and Pepsi just pummeling each other. And Zuckerberg is no spring chicken. He's got many trillions of dollars or whatever. And he puts out this giant tweet that's like, Yeah, we're doing it too, but we don't need anybody's help. We just make sure our stuff is safe.
Aaron Francis:And it's like, Oh, wow. That rules. Beautiful. So there is a self correcting mechanism in there. Yeah.
Justin:Seth in the chat says we're in the brick phone minutes era of AI which I think is a good, it's like, I mean I remember when long distance minutes cost you money.
Jordan:Well, some that's right. Someone came out with unlimited and it was like, well, that's all over. That whole decade of how many minutes do I get? How many how much does it weekends. Yes.
Jordan:That's right. Nights and weekends. Oh my god. Then then you have like the five. Right?
Jordan:The five of your look, your family's friends that didn't count.
Justin:Yes. Yeah.
Aaron Francis:Yes. I should talk about Myspace next. I mean, we're
Justin:all old here, apparently. I mean, talking about old things. I think the reason this JEV thing is animating the discussion right now is up till now, we've just had LLMs. But now we're like getting into the incredible machine era of AI where you have multiple building blocks. So we just had LLM models before, but now it's like, here's a different type of AI that's very good at this.
Justin:So now we can start seeing these things as tools. And as we've said, they're becoming cheaper and cheaper and more commodified. It's like, if you're a builder, now you've got all these incredible tools and you're like, oh, wow, I've got JEV up here, you know, making the decisions. I can feed it down to the LLM here and then I can I could have multiple LLMs working on that part and then I can have it's gonna keep going, you know?
Brian Casel:And and again, like, it's it's something that empowers the builder in this in this economy at a time when like, oh, like, the general public is is still not adopting Claude and ChatGPT in any real way other than just a just a replacement for Google. Like, really, even like small businesses and individuals and teachers and schools and medical offices and whatever it else, lawyers, like, yeah, there are these companies that serve all those industries, but you're not seeing mass adoption. You're seeing like, yeah, maybe, like, we're sort of tinkering with it around the edges. Maybe some forward thinking companies are starting to adopt it. But like so I so I still think that there is that opportunity now for software makers to make products that actually solve real that can find their niche and solve that problem.
Brian Casel:And a tool like JEV makes that even more feasible when you think about the cost of running AI for a small business to solve their problem, you know? What what do
Jordan:you guys think about the the the consumer side of this? Right? What what Brian's talking about is like, I you know, from our all of our point of view, we we think b to b, at least when it comes to, like, offering things
Brian Casel:for Well, that's sort of what I mean is like consumer, like small, like prosumer, small business, like, not in the tech industry.
Jordan:Right. I mean, you see a muse and instinct are are that that's what they're going after. Right? I mean, instinct did you guys see the ramp up? I don't know if you can find this tweet, Justin.
Jordan:The the ramp up in valuation is like is hysterical. It's like a 100,000,000, a billion, 10,000,000,000, 50,000,000,000 in like twenty six days. It's like, you know, makes no sense, but you can almost see like, okay, something's going to break through into the consumer space and get 50,000,000 Americans on it in the span of, like, a year of of everyone just talking about it. And it's I guess, it's just worth the gamble, you know. It's just a bunch of Ian Landsmann's with someone else's money writing checks, like We're still Yes.
Jordan:We're in.
Brian Casel:I still think that we are in such a bubble. Like, all these things that appear on our on our Twitter feeds to be blowing up, like, the world does not know what these things are, you know? Okay. Like,
Justin:Gawk bot.
Jordan:You you mean like an attention bubble.
Brian Casel:Instinct. No nobody in the real world knows
Jordan:I agree.
Brian Casel:Those words. Like, they don't think about them, you know? Right.
Aaron Francis:Well I am
Justin:starting to see it I am starting to see it, like, you know, as part part I I did see Normies talk about this Astra three d modeling thing. I think that kind of burst out of the bubble a little bit. Okay. People were like, oh, that's interesting. Like everybody, you know, well, a lot of people have floor plans and then wanna like see three walk throughs of them.
Justin:I'm seeing some Pull up my of tweet
Aaron Francis:on that.
Justin:If you're
Aaron Francis:gonna show a tweet, show mine.
Justin:Yeah. You go.
Jordan:Yeah. That's right. Wait. What'd do For your office or something. Gotta find
Aaron Francis:your Aaron Francis office.
Justin:Do that. I was like, damn. Yes. Yes. I looked at
Jordan:that and I was like, Aaron Aaron had an exit, bro.
Aaron Francis:He's planning
Jordan:he's planning
Justin:like fantastic. He feels good, man. Hell yeah.
Jordan:You know what? Somewhere interesting that it's popping up that's unexpected is schoolwork. Like parents and kids are kind of ahead of most others. Yeah. So, you know, my 12 year old gets homework and then she has a math test coming up and then the teacher gives them whatever whatever is 20 questions as a review ahead of the test.
Jordan:And a lot of the moms, what they're talking about is you can take pictures of this review sheet and then ask Claude to make additional review questions that are in line with this level of difficulty and grade level and subject So, you know, that that's kind of an advanced use and it seems to be happening in like education and parenting more than other places. It's cool. Alright.
Aaron Francis:I have I have a walk through here, Justin. Can I share my screen? Is that possible? Alright. This is my, this is my astro coded three d.
Aaron Francis:I so I went to the physical location. I took a scan I took a scan with some iPhone app called Polycam that I downloaded, and then I took a bunch of pictures. And then I said, alright, Astra. Here's the folder. I gotta get on a call.
Aaron Francis:Make this real. And now I have a pulley, and so now I'm just starting to lay out, like, where's the furniture gonna go? Should I should I get a a desktop CNC? Probably. I got a sewing machine.
Aaron Francis:Was about that sewing machine. That's a sewing machine. That'll be a three d printer.
Justin:I love this I love this little Wolfenstein three d.
Brian Casel:Look at Wolfenstein three d.
Aaron Francis:And so here's the here's the actual that's the scan. So that came out of my that came out of my iPhone.
Justin:See a guy with a gun around the corner. That actual scan.
Aaron Francis:Then I I gave it the scan, and I said, alright. Turn this into, like redo it and turn it into this clean version, and that is the clean one that Astra just used Blender to do. I didn't I didn't do any of this. Wow. I just basically said, alright.
Aaron Francis:Move this over here. Let's put the couch on that side. Let's switch this up. And it's just freaking awesome. Cool.
Aaron Francis:I can walk around in it now.
Brian Casel:I just wanna stop thinking about business and just make like games and fun projects like this
Aaron Francis:all the time. Yep.
Justin:This
Aaron Francis:I mean This is the future.
Justin:Be the consumer side is very interesting. I think that's still I I think so much of that is gonna be when the UI becomes ubiquitous in the same way that hey, Siri or hey, Amazon or whatever became the Yeah.
Brian Casel:Like, I'm not I'm not really using it right now. But I but I saw Claude like, just the Claude web app launched docs. And I think maybe, like, slide decks or whatever. So like they're taking a shot at Google Docs. Like they're trying to I
Aaron Francis:think really like Jev is going to increase the rate of consumer adoption massively. Not because they will have ever heard of Jev, but because you can imagine your smart home instead of responding in six seconds to turn on the light, it responds in under a second. Mhmm. Like this the guys from syntax.fm just released a YouTube video where that was one of the things he had Jeff do is he said, hey, x y z, turn on, you know, turn on my office light. And instead of routing that to an LLM, it routed to Jeff.
Aaron Francis:It came back with a probabilistic answer of 95% sure he wants the office light on, let's just call the tool immediately and skip the LLM altogether. I think that's when we're gonna start to see When application builders start building in better, faster experiences, we're gonna start to see more consumer applications. They get adopted a lot faster. I'm still a little negative on the, I'm gonna have a bot run my whole life and then like, I don't, I've got 10 employees and I just named them VP of whatever. A little bit bearish on that, but I'm very bullish on stuff like this where it's like, hey, turn on my light.
Aaron Francis:Boom. Done. Oh, is actually magic. That is actually a good experience.
Justin:Yeah. Yeah. Yeah. It's gonna be an interesting time for sure. I think we're think we're at the end of time here.
Justin:12:30. Yeah. And I we got to well, we didn't get to the future of courses, which was that. Oh, yeah.
Aaron Francis:They're dead. They're dead.
Brian Casel:It's it's all done.
Aaron Francis:It's over.
Jordan:Online education in general?
Brian Casel:Yeah. Yeah. I I I made a list of topics we knew, you know, knowing Aaron was coming on, and you do have experience in dev ed, and I've been doing that with builder methods.
Justin:And
Brian Casel:and of course, we see, you know, some developments with like, Go Rails. Yep. Which I'm I'm I'm happy to see Chris, you know, really go hard on Hatchbox, which is a fantastic Yep. Service that I've been a customer of for a long time. Mhmm.
Brian Casel:And, yeah. So, I don't know.
Jordan:Here here's what I like. I like that this conversation is where we naturally end up at, like, this this this level of what's happening. But the 95% of my day is, like, what does this ad say in the headline? What what does what what are the buttons of this screen in in onboarding?
Brian Casel:I I tweeted yesterday about the the JEV thing. You know, I I was just I was just like looking at the website to be like, what is this thing? And then I saw the headline on the on the it was like, we give a we give a fact.
Justin:I like that. Yeah. The FAQ was a
Brian Casel:great yeah.
Justin:Yeah. It was cheeky. Yeah.
Aaron Francis:No. I think so more specifically on courses, I think what I used to do is a 100% dead. Teach developers how to be really good at, for example, databases. 100% dead. I've seen that.
Aaron Francis:Jeffrey Way has seen that. Chris Oliver has seen that. You just don't go to a twenty minute video or God forbid a four hour course to learn how to be a developer. I think Brian, what you're doing is slightly different and correct me if I'm wrong, but I would guess that your mix between education and inspiration is much heavier than mine was. Mine was education.
Aaron Francis:I'm going to teach you how to be very good at MySQL, Postgres, SQLite. Brian, I think you're almost casting a vision and giving people a community, casting but a vision for these are the things that you can do. Now the underlying methods and processes will change along the way, but you can dream a little bigger. Here's how I do it. It's less like here's the index that you need to add if your data is shaped like this.
Aaron Francis:That's a 100% dead.
Brian Casel:I think that's kinda accurate. The what I learned with Builder Method, I just did a big newsletter and post about this today, and and I launched the the new buildermethods.com today. The big learning over the past year, because it's now like, this is like the one year birthday of of Builder Methods essentially. I'm reverting back to where it started because I I did I looking back, I feel like I went a little bit off course and I could see it reflected in the sales and the engagement and all that. But as I went harder into courses and I went deep on creating these big flagship courses, like those actually resonated less than what I had in the first six months, which was live workshops and community and helping people adopt and helping people stay ahead.
Brian Casel:Like, literally, the tagline sort of is like helping people stay ahead of the curve and and like as we navigate all this rapid change that keeps happening in in our craft. And so now, this this new builder methods, the new design, the new strategy, all the new copy is aimed at like, this company is more about community and live workshops that are happening Let's on a monthly catch our breath. Let's talk about what's happening right now. And let's let's acknowledge that like, probably within three months from now, everything's gonna change again. Of course, we're just gonna keep changing and
Aaron Francis:that's People when need still wanna be a part of something. Yeah. People still want to belong somewhere. And I feel like that is most of what you're selling.
Brian Casel:And and I and I think where I went wrong in the offering and in the repositioning over in the last six months was trying to create things like courses and frameworks and tools that that fell into old habits, like pre AI habits where it was it was like, I need to create the method, the system that you can adopt and commit to for a long period of time into the future. But that's just not reality. Now, the reality is we need to be fluid and change. And the value that I'm bringing is not like, here's the system to learn and commit to. It's like, let's just like navigate this together.
Aaron Francis:Here are the mechanisms by which you can learn new things is way more valuable than here is what you should know. So if you're
Justin:teaching Yeah.
Brian Casel:Also just how you
Aaron Francis:learn. Fantastic.
Brian Casel:It's also just like, here's what's working for me right now. And I expect that this is gonna change within three months. Like right now Yeah. I'm deep into Cloud Agents. That's new in the past month.
Brian Casel:There's just
Justin:Well, look how fast that changed. It's like everybody was saying it's Orban time and now it's like, nope, gone, Jeff.
Jordan:It was with one week.
Justin:Sorry, Thorsten. You're done. You're you're getting the spotlight I want to
Brian Casel:know all about the orbs but yeah. But yeah, it's like, you know, the the and and then, like, you see these these patterns of, like, they're all, like, coming out with, like, like, the agent orchestrator pattern. Yep. Mhmm.
Jordan:Just think it's just the
Justin:I like, a workshop also just feels, like, way less like a a twenty hour course Yep. For I think most people is just, like, very heavy. But oh, Brian's doing a workshop on yeah. He's gonna cover stuff like building with orbs or building the cloud or whatever. And it's like, oh, yeah.
Justin:That's been on my mind. When is that? October 1, I'm gonna be there. It it it's just a a easier ask for people to to.
Brian Casel:And I'm returning to the business model that like, these are paid workshops, like low ticket workshops, and then you have the option to upsell into joining the membership. You get access to all the workshops as as a member. Mhmm. So I'm kinda going back to that model and that positioning and so more more an emphasis on these, like, live community, live live workshops. The courses are there for the for the like, I still get some segment that's, like, really beginner and they sort of just need to be on ramped into what it means to work with AI and I have courses available for them now.
Brian Casel:But I think the value that people had in the beginning is this type of thing. So, like, that's the that's the bet that I have. I mean, it's sort of an educated bet, but going forward, but like, it's been a lot of, like, I've had like a lot of ups and then downs and ups against, and it's been a weird first twelve months. Like, a lot of the ups and downs that would have taken years in other businesses sort of happened over the course of twelve months in this.
Jordan:Short cycles. It's Yeah. It's interesting that, you know, people say either gotta be you get the barbell. Either gonna be deep in AI or, like, buy a sports franchise because people are craving this in person connection, and it's kind of going in that direction for you also around the community and live more so than, you know, recordings and this, like, you know, knowledge that gets stale too quickly.
Brian Casel:Mhmm. Yeah. Yep. Yep. Cool.
Justin:Alright,
Aaron Francis:guys. Covered it all. We did it.
Justin:Covered it all. Nice to have you, Aaron. We will eventually have a mega mostly panel podcast where we will try to wrestle all five of us. Is that right? Yeah.
Justin:All five That would be tricky.
Aaron Francis:Nobody will be able to lick away
Justin:from I we gotta be drunk for that one. I I think I am I'm proposing a drunken episode. I like it. Combo episode. If you want a drunken episode, put that in the YouTube comments.
Justin:Just comment drunk episode. And yeah. But yeah. Nice to have here, Aaron.
Jordan:This is how I wanna sign off for for Ian.
Aaron Francis:Woah. Yeah. He's more of a this guy.
Justin:Oh, yes.
Jordan:Aaron, thanks for coming, man.
Justin:Appreciate it. Thanks, everybody. Nice to see everybody in the chat. Seth, Pascal, Ryan. Oh, we had a lot
Aaron Francis:of people show up
Justin:at the end here. Ian was there. Dave Jones. Keith, nice to see you. Who else did I miss?
Justin:I said Pascal already. Bo is here, which is great. Emmett, nice to see
Jordan:you all. Did Ian know we were talking about him, or he just, like, felt the emanations from the Penumbra?
Aaron Francis:Ian is everywhere all at once.
Justin:Yeah. He's just got he just heard it. He heard it. Awesome. Alright, everybody.
Justin:Talk to later.
Brian Casel:Later, folks.
Aaron Francis:Bye.
