Today's episode is brought to you by Automatic in Motion Closing and on NewSend. Three companies that genuinely back the people building on the web. Hey, this is Adam Weeks from OpenChannels.fm and today I am joined by my good friend, Nate Stewart. We are going to be diving into AI and how that relates to the open source community. There are a lot of things that because AI is here, it's changing how we think about open source. And I couldn't think of anybody better than to talk to Nate because Nate, you have been around for a long time in the open source community, but you kind of like, you've jumped in and then you've jumped out and jumped in and jumped out and all of those were for different reasons. Before I get too far into that, I'd love for you just quick introduction. Who is Nate Stewart? Yes, so I'm Nate. I would call myself like a big open web fan, you know, for pretty much the whole life of how I got into it as I think maybe a 13 years old when I built my first site and all of that was off the back of people putting everything out in the open of HTML JavaScript. Like how do you just go in the web and figure it out? And from there when it e-com built a startup had that acquired by a company that went public, spent a lot of time in the open ecosystem including WordPress and building also other from the ground up open source projects that became pretty key to the company. That company that acquired us was a big commerce and their whole strategy was at the open SaaS platform. You know, so it was something I built over many, many years with the team there to illustrate the best of open source and of that SaaS platform we built. And now I'm back into open source land with AI and trying to blend the best there as well. Doing it again. Let's give people kind of a time frame or reference. When did you sell your company like to big commerce? How long ago was that? 2015, 2015. And then when did you move to Costa Rica? Puerto Rico. Everyone. It's okay. That's a normal thing. Yeah. There's one similar. Puerto Rico is a different place. Yeah. That was 2020. So in the middle of COVID and the backstory there is my wife was deep in. I think all of us were in the podcast that were like how to be a better partner, how to live your best life in middle of COVID and all of the stuff happening. Because she came out one day and is like I'm going to ask you a question and whatever you say, I will support you. What do you want to do next? And I'm like I want to live in an island. And she's like what is that? What does that mean? What? I don't know. It just came out. I want to live in an island. I think a lot of that was, all of us were feeling that we needed to get out of our house or apartment or wherever we were. And what's going to happen there in the future? So let's do it now. I made the move. We rented for a couple months and then we liked it and we ended up staying there almost five years. And then moved to San Francisco late last year. Yeah. Yeah. What was the impetus for that? So you were in Puerto Rico, beautiful place and you were leaving big commerce and starting your own thing at the same time. Like was there a spark of an idea that said, yeah, I'm ready to do something different? It was actually, again, my wife drives a lot of decisions there. She's like, okay, are you going to slow down for just a little bit? At that point, I'd been building through IPO and big commerce for almost 10 years. And it's like, yeah, yeah, let's have some us time. We're in Puerto Rico at the time. There's the beach there, rainforest, all this really cool stuff. And I told her, yeah, I'll slow down. It actually was not go straight into let's build a company. It was let's go just be together and do things. And I did that for a few, I want to see a few months, okay, for almost two months. It wasn't even a few months, but I did try. I did try. And a lot of that was just trying to say, okay, yeah, I was like, what's next? Now, did a lot of things I like doing with sailing, with the theme parks, travel around. I came to SF a couple of times and met with my previous co-founder. He'd already moved back to SF while it was going through just bouncing around the States. And I kind of just drew me back in. There was a lot of innovation is like the wrong word for it. It's an energy. Keep in mind, this is before, I would say in January or over the Christmas break where a lot of the models just started really speeding up in terms of what people could do with it, agentically and coding and otherwise. So then it was a lot more experimental energy and people just really trying to figure out, like what's the best way to do it? Leading up to that, I was just seeing patterns in what people were doing. I'm more of a pattern guy. I'm like ecosystem. I like to like seeing how people work. And that's really what drew me to what I'm working on now is that, confidence of factors of having some head space, seeing how people are working, seeing the energy. And I've been five years in Puerto Rico. I mean, at that point, you kind of. You did it. Yeah. Now, there are days where, you know, I'm like, I'm in a bean. You're gonna go back, but I can always go back and experience that. Yeah. Now, we were on a rooftop at Word Camp, U.S. I want to say it was last year, was that last year? Man, time is blending. And you were really excited about this idea and you'd already started on it. Just for the people to understand, you've left big commerce and you're ready to solve another problem. I'd like to know, in this age of AI, brains are spinning. The models are changing quickly. And you saw that there was a problem and then a solution. So let's start with the problem that you observed that caused, you know, part of even like you're moving from your home in Puerto Rico on an island that took you to San Francisco. What's the problem? So the problem was a lot of what was popping up then to the train with a vibe coding. So now that's normal. But about a year ago, that was a lot of like loveable, both.new, all these startups booting up. And the marketing was you can do anything. And I mean, it's kind of still the marketing. But the capabilities definitely weren't there for long running, you know, builds. And there were certain types I was seeing that would almost always fail and cause people to do one of many things that are frustrating. Like one set of stories that really just hit me hard is people taking their savings or, you know, kind of what was not meant to build an app money and put it towards finishing their loveable project or something else. And those were normally AI native projects, meaning somebody one had an idea for an AI product that could help out somebody or an industry they knew about and they prompted it. And it just couldn't help them finish that. Now web apps, web apps were going fine. And web apps today are super easy to build. And you can build a lot of things. You can build three D apps and whatnot. But when you think about it, the reasoning is like all of that is in the training data. Like building AI with AI is not the training. You have to kind of use your own intuition and understand what people want to do. So I think that's where it came from. I've seen people hit the building AI apps with AI was difficult. And I just started tracking that. I built a bunch of apps myself to try to figure it out. I joined the bolt.new hackathon that was like for a month and just was in the community seeing what people were building. Just a lot of just research finding, okay, well, this is still happening and not only for them, but also for founders and for like really like experienced people that they could build web apps and e-commerce stores and all of that all day. But then you get them to build an AI experience and it was difficult. Yeah. So it's using AI to build other stuff was working pretty well. It's using AI to build better AI tools. That's where the struggle was. Yeah. You look at like you have to control kind of two things. One is the experience on the app or dashboard or whatever you're building. And that's when I say web app, it's that. But then you get to, okay, now you have to tune the agent. You have to have, give it a personality. You have to tell it what it can. It can't do. You can prevent it from going off the rails and not like hallucinating was what we dealt with a lot, right? For most of the time early on. Now a lot of it is using the right tools. Are you asking me for permission in the right areas? And all those, there's a stack of things. There's like a hundred different variables that you have to get right as you scale. And probably like 10 early on that you have to get right or won't even work. And people were just kind of struggling with that. You know, we're getting a little bit better there. But it's not accessible for everyone like building a website is accessible for everyone. Yeah. Inmotion has been in the game for over 25 years. They burn their reputation. It's MV me, powered hosting 99.99% up time. And when something goes wrong, you get a real human on the line. Not a bot. They cover everything from share.
hosting all the way up to dedicated servers and their team will migrate your site for free. Over 170,000 professionals and agencies trust them and we're proud to count them as a partner supporting what we do here. Check out your options at inmotionhosting.com. Automatically, back in the open red since 2005, they're the company behind WordPress.com, group commerce, jetback and other open source tools that power millions of websites. Automatic co-founder help create WordPress itself and the team continues to build platforms that give creators, businesses and developers real ownership and control. We're grateful to have a sponsor that supports the same open web principle we care about. Learn more at automatic.com. OmniCend dropped their SMS pricing to 7/10 of a cent per message. Not a gimmick, they did it because when their customers win, they win. And if you're ready to switch, their migration team moves your automations, templates and contacts in five days, free. You could be up and running and saving up to 35% by next week. We're glad to have them supporting the show and the community we're building here. Get 30% off your first three months at omniCend.com forward slash open channels. When I think of the agency space, agencies aren't just asked to build websites very often anymore. They want a digital experience. There's a whole lot more that comes with that and custom software and AI and the mentations and like, "Wow, we were just building websites, but now we got to do a lot more for going to stay in business." Yeah. There's a transition from an agency that typically is like, "Oh, we do web work, we do app work. We optimize now to AI agency and AI consultant." For sure, that's been a major trend over the last year. Yeah, definitely seeing that. All right, so we're on that rooftop. You share on that idea with people and it's fascinating and you decide to move to San Francisco. You started run type. That's the company that you're getting going. But you also, along with this because apparently, you have too much free time. You're like, "Hey, I'm going to do an open-source project inside of this." This is kind of the transition to the story we were saying, "A.I. Big deal. We see a problem. We're going to create run type to solve that problem." But then how does that come back to the open-source community? What's this project that you're doing there? There's actually an order there. Run type came more beginning this year. We started to be building run type as a company and what run type was actually a different company, a different thing where it was more of me prototyping. When I was researching, I was actively prototyping. It was almost more automation as you would think about it in this space. A lot of where NADN was hitting blockers and what my research saw. It's a "Can you build a more AI native NADN that works with agents better?" That was that. It was cool and I learned a lot there, but it wasn't actually the whole landscape. I learned a couple other things. The second thing I was working on was actually something called vanilla agent. It uses vanilla JS and worked at any site. The reason that came out is there's a ton of React and now even Angular, other framework-driven conversational UI libraries out there. But there weren't a lot for the rest of the web. Most of the web is HTML and JavaScript and not a framework. It's not only like in something like WordPress, most themes just to drop and react is going to kind of be hard, but every other platform really, when you get down to it, they're typically like even a Shopify theme that's going to be more like liquid and whatnot. They're not all React. I built that to try to work around the problem of, "Can you make it where this conversational UI is themeable, brandable, customizable, and can drop on any site?" That was step two of going around. That became persona later on after we had run type. But then it transitioned. The open source project. Yeah, the open source agent UI library, as we call it now. But yeah, there's really a three-step process to going through the research, prototyping, figuring out what's working, being honest with what the utility is there, building the UI library for the rest of the web, finding out what's working there, and then run type as kind of the combination of a lot of those factors that we saw, "Hey, that's what's needed to do this efficiently." Got it. How did it feel important to you that persona needed to be open source and not just a closed, what you were selling? I think I'm very important because my bias is there. I've done that a few times. If you're building something that's for the whole web wants to make to where most sites that aren't in one of these other frameworks can use it, you have to make it open. It would be weird to say it's for the open web, but not open. I think it was just default. There also was early research and even design partner as we were working with companies. That meant a lot to them around they could control it down to each individual bite. That's just the power of open source so you can see exactly how it's running. It also was a good forcing function to figure what's unique that should be where values derived from and you can build a company around it and what should be shared so that there's momentum around it. I think a lot of open source, you want to build momentum and eventually other people feeling like they can make it their own. I think that's the right split for how we think about runtime now. Persona now is you can do whatever you want to it. I love that. You could use not runtime and that's fine. We get a lot of joy and happiness around someone just putting it into their product and they don't really talk to us because that adds up over time. That compounds. We've seen that so many times in the past. I think it was a natural fit there. I wish you had a better answer from I'm biased and that's the right thing to do. I think there's an important thing because it's the right thing to do only takes businesses so far. We don't litter because it's the right thing. I don't litter but there's a lot of things that it's the right thing to do but doesn't necessarily have a strong business case. What you are experiencing is that this thing that we've created doesn't make sense in the other way except for it to be open source. As you are thinking about open source and in this age of AI, does that calculus change at all? Or AI does not change the reason to have something be open source. There's a couple of factors in that. It does change things. I think the people having the hardest moments right now are thinking it doesn't, trying to basically work out in the open but protect themselves from how quickly people are moving and a lot of times people that would never ever contribute or never ever be able to use a open source project is too technical. Now they can. It's more of a matter of in my opinion, how do you split up your community and people you're working with in a human way to guide them in what they should contribute to, what they should try to utilize from it? People still need guidance on, sure, you have cloud. You're pointing it there. What are you trying to achieve? I saw that you submitted an issue. What's your outcome there? It might be that they don't need to customize anything. They just need a documentation to complete their use case. Well, that guide them how, hey, I would love a contribution and documentation. They could totally do that because they have the use case and they have the problem. So someone else might be hyper technical. They have an optimization and that's a different route of how do you contribute that back or plug-ins and all that. So I think it's waypoint, like, finding, path findings really important and communication, if you don't do it right, it's just the real world. People need proper communication to know where they can help or not. Oh, no, I love that. There are some things that have changed and some things that haven't. I don't know where, how do we go pre-AI post it? I don't know how or-- Well, I think it highlights a lot of what we were never good either personally or as an industry, whatever, in the first place. Communication, we talk about documentation, what I was, what I was talking about. A lot of products and open source projects don't have documentation that's actually helpful enough before, during, after AI. So that's more of, okay, what's the blocker there? And we were always trying to move faster. We just kind of always have that we really should have in terms of, like, enablement. You should go along the way, write down what you're doing and then, like, put in documentation. I experienced that building SaaS products and outside of open source, where you launch a product and, like, the documentation isn't there or the team doesn't know how to sell it. And again, I'm going to be back.
like we've just been given tools to move way faster, but let's be honest, we always wanted to move faster than we maybe could. So this is more of like an adjustment of, hey, what's the better way to work in general with and without AI? And I think a lot about that by the way of, what should be AI, what should not? The only way is to really learn yourself and then to apply kind of principle of that makes sense. And don't shy away from either side. Just get smarter, understand more about, what is this world and try to shape the world and what you want. Don't let it just kind of happen to you. - Yeah, just because we can, doesn't mean we should. - Yeah, and keep in mind like everybody is, almost the should button is duct taped down. - Just down. - Yeah, it's just there. They ripped off the switch or whatever. That's gonna happen. It's the knowing that that will happen that someone will put their foot in the gas and not stop 'cause they have these tools available to them. But that doesn't mean you can't be a voice and at over time maybe even influence around asking just good questions or I'm like, okay, so yeah, how would someone get started with this if they didn't have an agent? Okay, how someone, they have a bunch of tokens, they have the subscription, they wanna spend it on something cool, they wanna help. Where do they help? And also, how do you give other people chance to shine as well? When we launch Persona, I try to do a really good job of shouting out to all the other open source libraries that influenced us. Now, we're kind of in a different pocket, like a lot of them were React libraries and whatnot. So we're going after a different market of the rest of the web, but Persona still got a lot of value in me, like designing it and our team looking at it because we had other people that did great work before it. Let's give a little bit of credit, we're a credit to do. So I did all those things we're having to re-analyze, right? And it gets that always been in the world. We're now just accelerating every aspect of it. - For sure. - Yeah. Oh man, this is such a great topic. And we've just started to scratch the surface on this, AI and how it is changing open source. I want to touch a number of these different topics. And so I think we're going to do, we're going to go ahead and wrap it up here because we've started to kind of just scratch at a few of those topics. I mean, everything from, yeah, what if I want to contribute my tokens to a project? Well, how can I do that efficiently? There's all kinds of things around security and what does it mean for our AI agents to have access to all these huge repositories as that bad actors? I mean, I was on the thread this morning. There's another big, oh no, the sky is falling. Everybody update. Like, you know, stuff is happening quick. And so yeah, I'm excited to talk to you about does open source matter when code is cheap and then the human part of that. So I tell you what, let's have this be our first conversation, one of several where we are pulling at some of these different ideas. And yeah, I would invite people if they have any questions about this topic, anything that's interesting in open source and the age of AI, we would love to hear about it. Nate, as we're wrapping this up, if someone wants to get in touch with you and learn more about what you're working on, your open source project, your AI project, how would someone get in touch with you? I would say
[email protected] is the email and then on X, become vocal. Just to hit me up there. - That's a good, yeah, that's a good handle, become vocal. - I get it. - You know, it's an aspirational one. Like again, my wife would say, you need to follow your username because I always need to get out there more. Like we're doing right now. So I'm becoming more vocal. - Hi, you. - You watched that. - All right, well, I'm excited to help ask you questions and learn from you and help you become more vocal because Nate, you're one of the most brilliant guys I know in this space and you're doing some really cool stuff, moving from Puerto Rico to Silicon Valley to the Bay Area to launch this AI. We might even get into some of this stuff about, you know, like funding and funding an AI company. I mean, that's a pretty cool stuff and I can't wait to talk to you more. So please, if you're not already subscribed to OpenChannels and we'll see you guys here again very soon.
PodcastTranscript AI