 Okay, welcome back everyone to day three of CUBE coverage here at open source suburb. I'm John Furrier. We're here covering all the action. My co-host Rob Streche analyzing all the data, conversations, our next guest is Matt Butcher, co-founder and CEO for me on great CUBE alumni. What's great to see you, welcome back. Thanks for having me. So we just chatted at CUBECon EU, CloudNativeCon in Europe, 10,000 people. Yeah, it's fantastic. 2,000 on the waiting list. This event is much smaller. It's more of the insights, the premier show where all the leaders are here, the white people. Not big numbers, it's not designed to get the big numbers but all the top brains are here, Linux Foundation, to look at across all the projects and assess the future, how things are going. What's your take on the summit so far? I mean, before we get into Fermion, what you guys are doing? Yeah, and I can, I mean, this is one of my favorite events. I've been many, many years now. In fact, Fermion's first ever conference was an open source summit. The open source summit in Austin was our very first one. And the reason why is because you get here and you've got lots of different tracks across everything from very low level kernel stuff up through the cloud and on into things that sit on top of the cloud or roll up on the operating system stack. And this is one of those rare opportunities to see, like you said, the leaders of these communities, some of the contributors that you never see out in front of anything but who are quietly pushing specifications forward and the kinds of code that's holding up this kind of massive edifice we build. And they're all in here, like cross-pollinating and sharing ideas. And you see things happen here that just start seeding changes that might even take two or three years to get to, right? As the idea is spread out. But in open source, you know, you get this diaspora effect normally where everybody's kind of going that way. This is one of those rare opportunities to sort of gather people back in together. It's the brain trust really of open source because there's so much change coming on. We've been reporting and we've been kind of pointing out that we turned down other events to be here because there was like, why are you going there? Like, because we can sniff out change. We know where the cue goes, where the action is. We'll do whatever it takes. But seriously, there are structural changes that comes out of this that will have a trajectory and ramifications and economies of scale. There will change the makeup of how the foundation looks. And let's face it, open source one. It is now the software industry. It's no longer staying on the shoulder of giants and, you know, beat the proprietary, they're done, they've been beaten. We have this proprietary differentiation out there in software, but it's generally accepted principles now that the software industry is open source. Okay, great, check. Now what happens next? With the wave of AI coming, for me on your project, your company, your team, you guys are solving a lot of the same kinds of problems. Make things easier for developers, but at the same time, get some stability under the covers. This is real big trend, this idea of making things simpler. Yeah, yeah, I mean, I think we see pendulum effects in a lot of different areas of, you know, science and industry. And this is one of those where, you know, there was the developer oriented movement in the cloud early on and developer oriented, you know, has platform as a service, right? Heroku comes along and really says, developers, we're going to make the cloud a thing that you can do without ever having to know anything about the cloud. And then the pendulum started swinging and we hit Kubernetes and then all the things layered on top of Kubernetes and, you know, the gigantic menu of services you get at AWS and at Azure and all of these hyperscalers. And it tilted very heavily toward ops and we started to lose developers and developers started saying, I don't want to be an operations person, simple by my life for me, right? And serverless is really the pendulum swinging back the other way, serverless functions, the kind of thing that Lambda really kind of pioneered but now we're seeing the sort of resurgence of serverless is a paradigm because developers say, hey, this is easy. So, you know, Fermion, we're all about making things easier and WebAssembly seems to us to be like the platform upon which to build sort of this next little motion, the next wave of cloud computing. Please explain the product. I know we had many times on theCUBE with folks who haven't heard about it because my first comment to you, I think the first thing is about time, someone did this like, what took you so long? What took everyone so long? It's a really simple concept but it took a while and it makes so much sense. It's the future. This is what it should look like. This is success, explain the model. Yeah, and the thing about the way we have built things as we kind of built up in that complexity chain was we started building servers and every software developer has to build kind of a big stack of stuff in order to handle the business logic that they care about. Serverless makes it possible for the developer to say, look, the only thing I actually really care about is this little chunk here. I want to get a request in, want to do a little bit of work on it and I want to send it back. Don't want to know anything about the infrastructure, don't want to know anything about how many systems my thing is running on. I'll build it to be scalable because it's a small bit of code and I'll push it out there and everything else is an operational issue. And that's a lot different from the Kubernetes world. And as a creator of Helm, which was designed, my original intent with that was to build a tool that would enable developers to express Kubernetes concepts in a very easy and reusable way. Not realizing that actually what I was doing was beginning to contribute to the complexity. You know, I was contributing to the problem and suddenly the developer is like, you're telling me I need to write 5,000 lines of YAML and I'm like, I don't know. This is really like an overcorrection for me personally in the other direction. We want to make it so that the developer just writes a chunk of code that they care about and then they push it out there. That's a funny story. That's the story behind the story. You're actually trying to make it better but you made it worse. Yeah, yeah. Which makes you zoom out of home. This is penance. You dig in the hole and you go deeper. So I go through China, do I go right through China or do I come out? And I mean, I'm going to say, home is a good tool that has good use cases but my original assumptions were wrong and I contributed to the complexity. Yeah, I know. Yeah, and I think that was actually one of the funny things coming out of KubeCon was the fact that all our demos are still using YAML on stage and how that is really a blockade to getting new people in in a lot of cases. And I think that's where, you know, wasm and that's how you chunk things up into smaller pieces. Are you seeing that it has particularly good use cases for certain applications? So we built it with the assumption that the primary use case would be, you know, building web applications, building the back ends for web applications, building microservices because, well, in part because we've all done those, right? That's just day to day bread and butter coding at this point. We were a little surprised to see that as generative AI took me by surprise, the sudden boom of that took me by surprise as much as it took anybody else, right? I played with AI for years and always thought, yeah, this is a nice little toy, it's not going to go anywhere and then suddenly boom, right? And now we're starting to see people use that because the serverless model and the open AI and generative AI models, that works really well in a serverless context. You've already got this big, beefy thing that's doing a lot of the really hard work. Now you're just writing small things on top of that that can take user input, do a little transformation, get, you know, query the AI and then be able to return some. So I think that's going to be. I think this is the dynamic of AI. We've been asking the question, this tornado of AI is going to impact open source. The large language models and the foundational models, large languages, text, you got audio, video and images and audio and video are really have changed things a lot, even though they're large and take a lot of energy and GPU power, sustainable issues we've been reporting. But the open source community is building more smaller tools. It's starting to get into this kind of self-efficiency configuration and it has that web assembly vibe to it. Yeah, they're very complimentary technologies, honestly. But I think a funny thing happened, like even from when we talked at theCUBE in Amsterdam, you know, when the universe was opening up, right? And we're like, oh, there's all these possibilities. And now we seem to be in that funny moment, happened with the web too. We're in that funny moment when it's like, okay, something big just happened. What do I do about it, right? How am I going to interact with it? And in one, you kind of see some people saying, okay, you know, the end of the world is nigh, right? This is going to destroy everything. But even that seems to be sort of tapering off and now we're going, okay, we understand realistically what this technology is. And it's going to bring some big possibilities. What are the possibilities, right? Is this a good idea? Is that a good idea? Good discovery. I mean, I think the web example, I love the example more than the iPhone because I think the web was crude at the beginning. We all know dial-up, the websites and HTML and HTTP with the foundational protocols for all that markup and then transport. But it was all about a new user experience and capability. I could do new things that might have been crude. So it was a lot of possibilities and how people did things. So a lot of conversations around reducing the time it takes to do something, reducing steps, making things simpler and easier. So those things were in play while advancing functionality. So it was like an entrepreneurial dream. A lot of action happening, opportunity recognition off the charts. Everyone saw a piece of the action, hence the bubble. But somehow we forgot the fear that people felt about that too, right? And we confront fear with large language models and suddenly it's like, oh no, the grand edifice of copyright is at risk now. But that, and in fact, I bring that one up specifically because that was one of the concerns with the web. If everybody can get access to all this text, then the grand edifice of copyright is over. Didn't happen, right? And so there's this, the fear here is starting to move out of the way a little bit. And particularly when you come to a place like Open Source Summit, right? Where you've got a bunch of people who are going, we get excited by the possibility of technology. What can we build upon this? Well, there's a lot of people, I just, on New York Art, I just had a rant on my podcast around, they were fiction writers. I thought he's hallucinating personally. I thought it was written by AI. His whole point was, will AI turn into McKenzie? And then he's, so he's got double, double trashing everything, McKenzie is evil, chopping the blades of capitalism to take away your job. When you look at augmentation of human behavior, creativity explodes, the web did that and then made things easier to find things, and search, self-service. And if you look at, like, say chess, when computers played against each other, they were good, but when you gave a human augmentation, they played better chess against the best people. You have an average person, chess player, can become a grandmaster with the augmentation of the computer that was well-documented in the chess days. And that's a perfect metaphor, not a metaphor, it's a perfect example that we can draw a perilous. But first, you know, now I really want a shirt that says, humans hallucinate too. That was, yeah. But chess is a good example because coding in a way, when you think about chess as a rule system, right? And it's a rule system where I compete against you and we're following the same set of rules and there's an objective. But coding is a rule system to the same extent, right? And I remember when Copilot came out from GitHub and my first reaction was, yeah, I mean, we don't want to use a tool like that. That's basically going to make me a lazy programmer, right? That's going to make me worse at my job. But I was talking to a friend and he's telling me, you know, we measured that, we mandated for our team, all but three people are going to use Copilot. This is Wade from Muvio and he said, what they discovered was that all of the developers who used Copilot to help them write code experienced a 30% boost in their productivity. Everything, you know, well, I think you said on average, but you know, that means every three developers are essentially acting like a four developer team. And that's remarkable. And these were not like... This is real results. Yeah, yeah. And these were not engineers that, you know, were inexperienced or were learning. These were engineers who were already relatively seasoned and it's just having some... XP, right? Extreme programming and pair programming, right? We looked at that and we said, this is going to be a great idea because when you got two people sitting side by side and we're writing away our code, we're going to produce better code than if I go out and write the code by myself and then try and explain to you, okay, you know, these 4,000 lines of code do something, but you know, in a way Copilot really is sort of the next iteration of pair programming. You got something there that's just, but it's a lot more flexible. You don't have to take lunch breaks at the same time and be in your seats at the same time and share the same keyboard, which is just gross. But I do think, you know, this is a, that initial reaction of fear, right? That, oh, this is going to make me a worse programmer or worse. It's going to make me an obsolete programmer. And every single inflection point, the argument was, you know, jobs will be lost, bad for the economy. Remember when big data came out, oh, bank, ATMs will kill the bank tellers. More bank tellers now than they were before ATMs. So services do shift, creativity shifts. I think the augmentation has been a big story here. And the question that we've been asking is, okay, open source went one, the generations of success and the shoulders of giants, will AI topple the infrastructure of open source? How the foundations are organized, the structure? Because they were built for different purposes, the same goal. They had very thin at the top management, they didn't want to overspend on overhead, put all the money towards the product, aka open, and communities. And that was just a very efficient model. And everyone bought into it. Does AI change that, the people equation, how people interact, ecosystems are developing? If it's an industry power dynamic, is that different than say, organizational kind of dynamic? What's your view on that? Well, when there's so many ways you could go with that, but you know, if we rewind back to the 90s and say, you know, when there was that open source versus the enterprise story, right? And open source was seen as a threat. The rallying cry to open source was, we're going to democratize the process of writing software. But really, we didn't totally mean we were going to democratize it at that point, right? Because not all of us knew how to write code and not all of us could participate in this ecosystem. And it was been able to dictate our model. Oh, yeah, well, yeah, well, yeah. By democratize, we mean approximately replicate other models but without anybody making money. Kill proprietary technology. Kill proprietary code. Yeah, that was the real reason. Yeah, and now, but if AI provides us a tool that will help people who have never written code before to start getting involved or people who are reasonably experienced in one language in one domain, suddenly pivot and be able to contribute to another project that's in a different language or a different domain than they would have and have the AI essentially act as that, you know, that other person that you're pairing with. And then we're suddenly going to get, we're going to get a multiplicative, it's going to multiply. Yeah, I think that's the interesting thing is like, again, like, hey, I built this in Python, now I want to go do something in Rust or something like that and trying to have that portability. And I think that's the interesting play with AI. And I think actually it levels, it actually levels the playing field versus unleveling the playing field. I mean, there will be people who can take advantage of it better, but it's a tool that can have that effect of, and I think part of it is, I'm interested in this, I think it will help people get to MVP, get to that first product a lot faster for a lot of these different companies. So if it results in more software being produced, we think that makes a better world, right? Yeah, well, to your point about peer pair programming, extreme programming, I think that's a great illustration that ties back to the chess example. Human plus AI is better than AI alone. That I think summarizes all the conversations we've had on theCUBE this week. Ed Warnacky, who pointed out that book from Tyler Cohen, Average is Over, points that out with chess, and we hit it up with coding. This is human and AI is better than AI alone. So I think, okay, it's a tool for, if you want to look at it as a tool, it's a tool. If you want to look at it as augmentation and an assistant, it's an assistant. If you want to look at it as automation, the heavy lifting, undifferentiated heavy lifting, but the human can move faster with the brain cycles. That's the thesis, you agree with that? Oh yeah, absolutely. And we have to acknowledge that there are scary ways that that could happen. As is the case with every single innovation we have made in the history of mankind from a wheel onward, right? But the opportunity to use those tools to achieve more, and you asked earlier about open source communities, to expand the community to include people who could not participate before. That's a tool we're having, right? The thing that's happening, that other thing we're observing, and I'd love to get your thoughts on this, is when you have these shifts like the web now with AI, some things went away, mechanisms, because the goal's still the same. The mechanisms change. What changes with AI in open source? Is there things that kind of were built for the old model that go away? If the goal is to have great seamless API interactions, systems thinking, I bring this up because the word platform engineering has been kicked around a lot. And so, and that's changed from defining it as S-R-E-G-I-T, as we're calling it. And so you're starting to see these new formations of structure, where the goal is to do what you're doing at Wasim and WebAssembly, which is make things easier, faster, and workable across all platforms. That's the goal. Is there things shifting in your mind that you see out in the landscape that will kind of be new structure that makes things faster, more agile, cleaner? And what, is there things that go away? Because that's something that we're trying to look at and identify with. What do you guys think? I'm going to go way out on a limb here, because I can. Okay. That's it. You know, if we think about how the generative AI works, and particularly in large language models, right? You know, I can express something in my preferred language, English in my case, right? Ask the system a question. The system can do some work and return back something that ideally in the same language. Most of, not most of, all of the edifice we have built called software is about me learning a different language and being able to express to the computer in this kind of intermediary language that ideally both I and the computer understand, right? Well, what if maybe the biggest risk is, or the biggest change is that programming languages in general will go away for the average developer? The coding will be more like sitting there going, okay, so what I want is a shopping cart, and the shopping cart is going to have these properties, and when a user comes, they're going to have a button to click, and I'm going to talk to the computer and it will be building out a lot of these things. Actually, when you think about it, if we learn to articulate ourselves clearly in English, this is going to be a far more concise process to use than sitting there going, okay, do I need curly braces here, or is this, you know, did I miss a semi colon somewhere in there? And probably there will always be room for the kind of low level programming and things like that, but I actually, what would excite me is if we started to see cases where you could express in plain English or plain language of your choice, right? This is what I want to do, and really we're training humans to be better talkers and the system to be a better compiler. Yeah, I think that's a great limb to go out on because that highlights voice, which we all know, hey Siri, which is weak compared to some of the AI coming here. Well, yesterday's chatbot is not today, just seeing that in the format. Yeah, and I think what's interesting is, and the way you approached it is also it's a CX back, it's a customer experience back type of view, and I think that even over the last five years, different companies that I've worked with from a product perspective, working backwards from the customer and understanding that and describing it has been a big piece of it, and I think even for like the last 10 years, getting the engineers to be involved in those conversations, sometimes pushing them out of their comfort zone, but I think it's a different skill set and I think to your point, I think that's where I think you're right. I don't think it's a huge limb to go out on, but you can get kind of a framework built out maybe with AI to begin with, and then you have the people who may not be a socially inclined can go and tune that from there, and I think it's almost like having the business requirements written down in CX, in UX, almost. Yeah, and to your point then, these things are iterative, right? It'll take us a while to get there, we will have some definite false starts. And then when you think about it, the low code, no code movement, I'm not going to call it a false start, but that was one where we were trying to get a similar CX, but we didn't at that time, our toolbox lacked this one new key technology that we have that might be able to realize a lot of that vision, which was let's just make this more broadly available to the people who just need to describe a business process, or to the people who want to build a pet food tracking database, you know? Ha ha ha. Don't you wish you were 25 again? Oh yeah. For so many reasons, but yeah, it's awesome. Ha ha ha. I mean, so many goodness from a coding standpoint. I mean, I think this democratization wave, call whatever you want, AI impact, the tornado, you got to get inside that tornado, otherwise it just spun out. I think this is what we are looking at is, there's definitely going to be a rattling of the open source way of life, just by default, it's the tailwind. That's a great tailwind, it's not a headwind. So, you know, what has to lift? That's the big question we're asking, and the work you're doing with Woz and WebAssembly is, to me, I think a North Star for all systems, which is make stuff that works for developers across multiple platforms. Yep, yep. And API's already set up. Yeah. This platform engineering, that's why Rob and I are really kind of honing in on, you know, what is a platform? Yeah. It's not a cliche. It shouldn't be a marketing architecture. It should be a legit platform. Yeah. And have tools and apps that sit on top of it. Medellin's got 4,000 apps. Teams, wow. That's how many teams they have. Yeah. I mean, and when you start to look at it, it's how, if you have to go and actually change those and did, you know, there have to be enhanced or patched or done. And I think you brought it up earlier, and I think it's always been one of those, don't SRE me. Right. So that I've always worked with, or like just don't SRE me. I don't want to have to build and run and get woken up every day at three in the morning cause something goes bump in the middle of the night. And I think that's where kind of breaking things down into a platform, on a platform into smaller pieces makes a lot of sense. And I think, I mean, and what I like about the direction of this conversation is, you know, we've talked a lot about the newest piece of this puzzle, which is, you know, the seemingly outrageously instant arrival of large language models of generative AI in general. But this technology is still blending with the technology that we don't want to be dismissive and say was yesterday's technology, right? What cloud and then what systems like Kubernetes and containerization and WebAssembly are bringing is this idea that we can distribute computing in new ways, right? And distributed computing, you know, as Brendan Burns, the guy who wrote Kubernetes used to say, you know, distributed computing 10, 15 years ago was the CS400, 500, the last class you took before you left school. Now it's the first class you take when you walk in the door, right? And that's a hugely enabling technology for how we're going to be able to use the compute power a little more efficiently, waste a little bit less, accomplish a little bit more and that took a lot of design in there. Now we can drop on a new layer on top of this and, you know, the gateways are opening now and we're still kind of looking as the door starts to open and you've got an eye up against there going, okay, what does the world look like? But we know we've built a foundation for something new and it's going to be very exciting. And what you just said is really important because that distributed computing architecture is a system, it's systems thinking. That's why it's the last class in computer science because it's the capstone operating system, systems management class that humans up to. You learn nine, 10 languages going in, you're doing compiler design, writing a link or a loader. All that OS stuff, that's the pinnacle. Now it's entry level. Then you've got cloud scale. Okay, so you've got complexity in the systems thinking and a system that has consequences in collateral damage. You can't just unplug something without having some damage so you can understand all those dependencies and then you've got scale. So I think the combination of that and then now with AI data, data scaling. So data scaling and that's going to create all kinds of glue layer impacts. I think that's why I think the Kubernetes container thing is going to be boom because microservices is going to fill that up with machines. Machine data and machines have to be doing that, heavy lifting, hence the human and machine angle. So that's why I think AI is really, really going to be impactful in the world because you can't ignore it. It's the only answer to handle security. Yep, it may turn out that we built an edifice big enough that we need AI to be able to help us operate it. Otherwise it will just be too, too cognitively overwhelming for us. The large code model. Yeah, yeah. I'll see you in. It's called open source. Matt, great to have you on. As always, a pleasure to commentate, opinionate on what's happening. Put the puzzle together on theCUBE in real time. Really appreciate you coming on. Give a quick plug for the company that you're working on right now. I know you had a lot of press interviews, analyst meetings, what's going on in your world. Take a minute to give a good plug. So for me on technology, we're building the next wave of cloud computing. We believe WebAssembly is a key enabling technology, but this is all about the distributed computing, right? And this is all about building kind of a foundation where developers and maybe machines and not very long will write small serverless functions that will be able to very quickly deliver value right away. Very excited about the direction that all of this is going. Really excited to kind of embrace this future that we're stepping into right now. But I really just want that human solucinate two T-shirt. Yeah, and that's a great punchline. We're never hallucinating on theCUBE. We're always keeping it real. We're riffing, we're going out there, pushing the envelope. Rob, great to see you, thanks for coming on. Great session, guys. Great, great analyzing and putting it all together. theCUBE, okay, live here. Day three, wall-to-wall coverage. Again, intimate event through the pioneers are, the leaders and the new leaders are emerging projects are here at Open Source Summit. This foundation is theCUBE, breaking it to you. All live three days. We'll be right back after this short break.