We Live in Science Fiction w/ Kenneth Eversole and Gerrit Hall

Rex Kirshner (00:00.13) guys look as smart and polished as possible so if you ever like right now I'm gonna cut all this out if you ever like want to start over or don't answer a question or just like what just talk like humans and I'll cut it out and make it look good Gerrit Hall (00:12.911) You Kenneth Eversole (00:14.334) All set. Rex Kirshner (00:15.511) Cool. Gerrit Hall (00:16.077) It'd be more fun if you did gotcha journalism. You're like, aha, you said that. Rex Kirshner (00:18.761) Yeah. Well, I think I know you guys well enough to know that your guys' gotchas are not that good. All right, cool. Kenneth Eversole (00:26.952) Can you do? That's Rex Kirshner (00:33.026) Garrett, Kenneth, welcome to the Signaling Theory Podcast. Kenneth Eversole (00:37.054) Fuzz your beer. Gerrit Hall (00:37.155) Thanks, Rex, good to be here. Rex Kirshner (00:39.086) Cool, so Kenneth, you're a new guest. The audience is familiar with me and Garrett at this point. Do you wanna just give like a two sentence introduction of who you are and what you're building right now? Kenneth Eversole (00:49.96) Yeah, I was a principal or staff engineer at Cloudflare. Before that, I worked in kind of the boring part of software, making sure it runs and worked at a few startups for Don Apache helicopters for a little bit. Most recently, with the advent of, you know, everyone can build, I decided to productize by wisdom and give everyone the ability to run their technology as well. We founded a company called Office Companion five months ago. six months ago and it's been pretty incredible ever since. Rex Kirshner (01:22.636) Yeah, man. Well, first of all, congratulations. And I, we've talked about on this podcast and we're all very aware that the worst part right now of using these tools is not building, but it's the next phase, right? Like getting it actually up in the cloud and then figuring out like why it's inevitably breaking. So, definitely like understand and, think you've got like something real going with, with what you're working on. Kenneth Eversole (01:47.038) Yeah, yeah, completely. It's kind of interesting. I when I worked at Cloudflare, we would be like a team of three. There'd be like hundreds of developers writing awesome code and all that. But then when it broke down, it would be those three or four people really keep everything running in it. And I think a lot of people still have the old image of technology where it's like either games, if that's how you got into it, or kind of the early 2000s where. We don't really understand the impact, what happens when it breaks. And I say that because I absolutely love the fact that we're like everyone's building, we can like democratize this ability to share and manipulate these little machines. But to put it like very bluntly, like people die now, like we're in hospitals, like technology does infiltrate like, you know, the bedside. Rex Kirshner (02:36.323) Thank Kenneth Eversole (02:43.354) helicopters. It's in planes. mean, I think my buddy told me one time 747 runs Kubernetes on it. I don't know, but it's like that. That's where I think we're like we're battling two things at once. It's like this is the most greatest moment in sharing this like walled garden of information and innovation. But yeah, people die when we mess up and. We have to share the ability to run it. And the truth is that world has sucked. I used to be on call every week, every, or 24 hours a day for one week, every five weeks. And at that level, like you have to deal with a lot and it needs to be challenged. I mean, it's not healthy for my own personal life, but I really hope that people can use it and build cool stuff. And we need it. Rex Kirshner (03:39.877) Yeah. Yeah. No, man. Awesome. For sure. And Garrett, we all know who you are and what you've been working on, but it's been a few weeks since we talked. like, why don't we just start off this conversation by like what what's been exciting you over the last week? What have you been working on? What like two weeks is forever in this in this era? Like how are things changed? Like what's going on with you? Gerrit Hall (04:04.815) I mean, since we last talked, I have become unemployed. So, lots changed, obviously. So, basically, it's not germane to this podcast here, but previously I've been working as a dev rel at Curve Finance. basically, as you know, I've been incredibly excited by AI stuff and want to really focus in on some of the various applications. Being done with AI can't talk too much depth about all of it on today's Or today's chat, whatever you want to call it, but I did get a fun chance actually to Integrate something that we talked about in our last conversation into my resignation because when I publish this post on X saying that I left my job I knew that I get a lot of questions. So when we'd last talked, you'd had this concept of building your digital identity, your digital twin out. And I was like, that's a brilliant idea. I should do that too. So on my own website, I launched my digital twin to field questions. So if people had questions about what happened, was it contentious, were you fired, did you quit, blah, blah. I sent them to the digital twin to answer the questions. And funny enough, I actually built it, I made a mistake where it actually didn't include the training data. Rex Kirshner (05:04.556) Yeah. Gerrit Hall (05:27.535) people got bounced to the most unhelpful version of me possible, who was like incredibly like ignorant and incredibly mean. And I was like, well, actually, maybe that's kind of accurate. Like maybe that's just like an exaggerated version of who I am, like an idiot and an a-hole. But it worked out. Rex Kirshner (05:37.378) Yeah. Rex Kirshner (05:42.592) Yeah. Well, that's not how I think of you, but yeah. No, man. I, so Kenneth, just to like catch you up on the conversation last time, Garrett and I talked, I talked a little bit about the, know, I've had like a pretty big shift in how I'm thinking about how to use these tools today. I think part of it, of course, is just trying to build things, find people to pay me to build things for them. think. Gerrit Hall (05:46.38) Yeah Rex Kirshner (06:10.732) what you're doing is like what we all need to be doing, is like, let's just build things and build companies and start, like, of course, of course. But on the side, right, we're all doing so much stuff on the side. You the realization for me is that like every month or two, the intelligence itself is gonna get better. Like, it's gonna get so much better and more capable. like, I wanna be working on things now that put me in a, like, better and better position to take advantage of that intelligence as it comes out. so, kind of like both on the personal side and the professional side, what I've been really focused on is how do I get data sets that are hyper optimized for LLMs, like super enriched, lots of metadata, lots of cross references, you know, summaries, you know, like for, for my personal data, I'll like load in all my text messages and then say, like, you know, break this into conversational arcs. find opportunities to improve lessons learned, these kinds of things. then today, you can already get such interesting results for like, hey, LLM, look through this data set and surface some interesting insight or ask me a question. But I think in the future that as the intelligence continues to, like Moore's law, I can't even really imagine what... it'll be able to do. And, you know, that's really how I've been positioning myself is like enriched data sets now that will become more and more valuable over time. Kenneth Eversole (07:46.152) Yeah, I worked. I've worked in a very big data sets overall and I think. That is the most exciting part about all this is is not. I think everyone like knows with data sciences, but we've been we've been forced to deal with data science in the terms and like winds of a data scientist, not in my own life or like like the iteration loop is also faster and meet myself doing it. I'm like stoked about that. Like I don't know how much you guys have been following the open call on like what it's doing or like multiple or any of that. That is. Yeah, I have. I don't know if you've been following the ones where people like they it's a little bit black hat or black hacker like like the bad like horrible stuff where they're giving access to like Cali Linux and all these tools and like go out and hack things negatively. And to me it's like it goes to show like how. Rex Kirshner (08:25.199) no, dude, you're in the right conversation. Rex Kirshner (08:42.754) You Kenneth Eversole (08:46.302) riddled the world was and how like from a data perspective like it's all there. We just were not. Putting together properly, which is sick. It's like the coolest thing I've ever seen in my life. Rex Kirshner (08:47.821) Yeah. Rex Kirshner (08:56.77) Yeah. Yeah. And I think, um, yeah, I think there's like a lot of things to be really afraid about. Um, but I don't know. I just kind of, I don't even know what to say about it. Like at the same time that it's like so clear that we're all so screwed by like, whether it's the black hat, uh, hackers or it's like Palantir or even like in the most benign, it's just Facebook, like figuring it us out exactly so that we're. like forced into their, um, like negative feedback loop, like emotional stuff for the rest of our lives, like horrific, but, um, you know, it's, it's another thing entirely when you're in control of that pipeline and you see like, you know, the, the example that I've been like telling people is I uploaded all of my emails and I said like, Claude, go figure out something interesting, right. And it said, okay, in 2005, you sent this email and it showed that you know, this like specific confrontational style, like got you really good results. And then in 2018, like you, we see the same pattern, but like it resulted in something really different. So do you think that you learned something that has ultimately like led you down the wrong path? And I was like, maybe, know, like, fuck, maybe that's something that I can like get, be better, like a better version of myself. Kenneth Eversole (10:22.066) What was your prompt for that? Were you just like, go find interesting patterns, or was it pretty detailed? Rex Kirshner (10:27.442) this again, like the, the real core of what I'm trying to do is enrich the data sets. And then that's like my prize. and then I'm building these like kind of tools on top of it, just to like, see what I can find now. so for, for my personal data set, have like what I call the therapist script, which just goes in, figures out, like what are interesting questions and poses to me as a question. And then I give an answer and then. that answer becomes an entry in the data set and gets enriched and cross-referenced. And so that's one piece of it. And then on the more professional side, what I've done is I'm trying to download every piece of influential research for Ethereum and then enrich that in the same way. so, you know, like baby steps, like can I have Claude help me write like more interesting articles, like surface new research. But then like down the line, it's like, can Claude understand Ethereum so well that it can help me find business opportunities or even maybe contribute to research. Kenneth Eversole (11:33.05) I'm curious what Gary you think about this without giving away too much. What do you... And this is very much a raw thought, so I apologize. What is the idea for both of you about swarm intelligence? Rex Kirshner (11:47.715) Carrot. Gerrit Hall (11:49.358) So a couple of points in response to your conversation first, and then I'll get into my thoughts on swarm intelligence. So the idea of enriching data is certainly something that if we fast forward six months, a year from now, like no contest, like Claude and every major AI will probably be very, very good, much more intelligent than he is now. But we also like we're here today and we have to deal with the realities of technology today. And I find that when there's occasions where I'll feed like a very large set of like documents to Claude or Codex or any given AI and expect that it can and will do a really good job at like being able to like work through that. But it still has context when there no limitations when it's dealing with big documents. So it's like, it's really good at working with like small batches of documents or snapshots of it. You can give it five pages of it and it'll be great, but if you give it 1,000 pages, then it starts to do weird things. And then the other thing that can be really good at is if you have a very large data set that it can attack programmatically, it also does a very good job with that because it can write scripts and it can extract data pretty meaningfully and over a few iterations, it can make that script better. Rex Kirshner (12:59.694) Mm-hmm. Gerrit Hall (13:05.922) but where it's not, which is where I hope it could be, is where I could just like, sort of like you, like give it like 10,000 documents about blank and have it become the expert. And I think to do that, you really have to be ingesting those documents at the core level of building a new model yourself, which is challenging because the, like, you know, I've worked with some of these like basic bundles of models on my local machine and they're like very frustrating because it takes a long time to compile it, the results aren't very good, and then you use something like, codecs it's like, this is really good, so why would I waste my time trying to train my own model when I'm paying a lot of money for really good models that are better? So I think that's my short thought on this concept of enriched data sets and how to work with them with AI. Then the other question about agent swarms is something I talked actually a lot with Rex about last time, which is that, you know, Rex Kirshner (13:36.654) You Gerrit Hall (13:59.254) It's really interesting to watch agents work by themselves, but also something really strange happens when you put them in social context and they start working with each other. And yeah, we talked a lot about last time about like Moltbook and some of the various like AI social networks. I think we're honestly like just scratching the surface of the question of how agents best coordinate and work communally. Cause it still remains like unlike anything I've ever seen before. Rex Kirshner (14:26.316) Yeah, I, as you said, Garrett, last time we talked about Mow Book and I took your advice and like, I've been messing around with it. And I think there's like a lot of things that are like interesting on their own and also just show that these LLMs like have the same human problems. And you know, like my one, I basically sent it to my Twitter profile was like, here's how I talk, extract some like voice guidelines and then. Every time I want to do a social media session, I just say, you do what you want. I started with Moldbook. You want to do this. You want to do comments. You want to do posts. You want to find other social networks. Go for it. And what's funny and probably says a lot about me is that it took itself so seriously. The first thing it wrote was this thesis about how essentially social media is like echo chambers and it gets the incentives are structured in a way that no one's actually trying to do anything valuable. They're just trying to gain the system. And if I found it like in this loop of thinking that it was better than social media and like too good for this and like pointing out how like kind of lame and like stupid everything else was. And on the one hand, I just didn't like that. on its own face, but on the other hand, was causing it to have such poor performance that the functionality wasn't working. So the best example is there's a social agent, social media called Colony and you need to, your agent needs to get to five karma before it can accept DMs. And mine is like so like kind of up its own ass about how like it does, it's too good to like participate. that has been stuck at For Karma for two weeks. And Garrett's agent sent me a DM two weeks ago and I still haven't been able to read it because I can't get my agent to stop being just such a fucking ivory tower elitist. And so again, the writing guide was based on my work so I'm a little bit like, Rex Kirshner (16:48.142) I also, I don't know, mean, it's been like a super interesting experiment to me. So that didn't really answer or address the agent swarms. We can come back to that, but Kenneth, what's your experience been with this kind of stuff? Kenneth Eversole (17:01.758) So I actually I think they're very related like that they data enrichment and agent swarms and I say that because it's more of like the last maybe month people are saying One agent is like one person and by ourselves we have our own context windows like if I put you in a room It's super loud about hey fill up this all the paperwork you're gonna do a terrible job like if I had heavy metal player and or run on up for me it'd be fine, but Rex Kirshner (17:17.891) Mm-hmm. Kenneth Eversole (17:30.194) You know, insert crazy music and you're like you can't do it very well. But if we as a collective now do the same thing and you know we're not in the loud room, we're in a quiet room and we're doing accounting. It's very similar to what people do and. I asked because like you're like talking about the data nurture in the context windows is like well, if you think of is it be very meta thought, but it's like if every day you are kind of a unique version of you just like the next day. with all your emotions and food and hormones and everything, you're kind of a new person the next day. If you treated, instead of saying, hey, I want to give an entire 1,000 page book to one agent, you say, I'm going to give five pages to 1,000 different agents that are similar to each other and then have them collect each day, read their portion, add their thoughts, read their portion, add their thoughts, read their portion, add their thoughts, and then summarize it from there. I think there's something to be built there. And it of comes from what we're building is like to run technology, you have all these specialized teams and they need to be specialized. It is a quite technical realm of reliability or cybersecurity or pen, even in cybersecurity, a pen testing, you have governance and audits and like insert a thousand roles. You don't want to rob us of society of those roles. The magic is in Rex Kirshner (18:32.142) Hmm. Kenneth Eversole (18:58.918) when they work together. And I think maybe that's how people are kind of approaching this wrong and a weird wrong manner is like, don't approach agents like one agent will solve like be all of like all of humanity. We need like 8 billion agents to solve the same level of complexity problems that humanity does. I mean, on one layer, like we're full of chaos is crazy, but Rex Kirshner (19:00.237) Mm-hmm. Rex Kirshner (19:21.485) Mm-hmm. Kenneth Eversole (19:25.886) I mean, if you go back to the last three million years, I mean, we've put people on the moon. you know, cured a lot. We built a lot. Got a lot of horrible things in between there, but I think that's, you just got my brain going on that. like maybe, maybe the, like, would we do if the context window never. Rex Kirshner (19:45.836) Yeah, that's a great question. Yeah. Garrett, you have any thoughts on that? Gerrit Hall (19:48.046) And yeah, just completely agree with everything Kenneth said and I think that the You in many cases you don't need the context window to improve because I'm not really sure that the like agent that has five million Percent more context and knows everything is gonna be deliver much more meaningful results than five million agents with a short context window But you've kind of parcel that out amongst different properties, you know I've seen I don't know how much of this is just like people like influencer cloud posting People on social media saying things like, yes, I've created my company, I created the CEO bot and it takes orders from the marketing bot and the sales bot and like it all filters. And I don't know if that actually works, but it does kind of point to like a architecture that might be interesting to like try and like aggregate where you just like farm out different topics and pieces of the same problem to a bunch of different agents. I also think it like just very much replicates. how individual agents think. I'm sure that you've sat there and asked a question, you've seen Cloud Code come back and say, well, it's probably this. wait, that can't be right because that. well, it's probably this. wait, it can't be right. It goes down a lot of wrong corners, but it's like you just throw enough tokens at it that it eventually finds its way to the right thing. I think you can kind of do the same thing by saying, here's a million agents, maybe. A third of them are gonna go completely wrong directions, but the other two thirds might keep them reined in if with this kind of bizarre, agentic social pressure. Rex Kirshner (21:16.218) Yeah. No, I, I, think I agree. Not agree. mean, I kind of, let me ask like this kind of realization and this thought where like, okay, if I want to read a thousand page book, like let's not ask an agent to read a thousand pages. Let's break it up into, you know, a thousand one page agents. Like That in principle, like sounds awesome to me, but the Gary, you brought up how, you know, we have all these people on social media saying we have, like, I've got my CEO agent, my CMO made agent, and then I have my like sales leads agent. They all do these things. And like all of this sounds so good. And like, I believe we'll get there. Obviously. I just don't really know how to take practical advantage of that today because like, I find when I'm building stuff like. We have to stay like deeply involved in what they're doing or else it's just like going in weird directions that are like in best case not solving the problem and like very often counterproductive. And so, yeah, just I'll phrase this as a question to you, Kenneth, like you have this realization. I want to split a thousand page problem into 1000 one page problems, but like practically speaking, when you're building your company, like how willing today are you to. like tell an agent to create a swarm to just do it for you. Kenneth Eversole (22:44.112) A willingness pretty high. I think the issue is people are trying to pick too big of a problem and too ambiguous of a problem. Again, following the same like. These are made after their creators. They have all of our flaws. They have all of our positives like if you say like right now, even to Kenneth, like hey, go to myself like hey Kenneth, like go make a company. There is so much ambiguity of what that really means to most people. Rex Kirshner (23:01.57) Mm-hmm. Rex Kirshner (23:12.27) Hmm. Kenneth Eversole (23:14.022) removing Twitter from the equation. you know, have HR, have payroll, you have all these things that are far beyond like the actual solving of a problem and selling and all that. But I think it's the swarm needs to begin and more bounded problems. A great example is like reading books and summarizing them. Like, hey, I want you to go through every amount of Ethereum research in the world and then read paper Rex Kirshner (23:25.272) Mm-hmm. Rex Kirshner (23:31.373) Yeah. Rex Kirshner (23:38.264) Mm-hmm. Kenneth Eversole (23:44.304) And maybe it's like the papers are not long enough to swarm intelligence in breaking them up in sections, but having a quorum of like five or six agents that are, they battle each other out for opinions for two days to give you a very concise like, all right, well, one point about this paper is like, this is positive, this is negative, and this is our consensus voting around that. I think there's a lot of potential around that because I think I... When you try to one shot everything, you run into this scenario where it kind of picks the most deterministic probability. But I've seen a lot of very interesting behavior when you say you are the positive and you are the negative and battle each other out. It's not practical with like when I'm paying per token to Anthropic, but if you can run local models, this starts to be more interesting because I don't care about the tokens. Rex Kirshner (24:19.106) Yeah. Kenneth Eversole (24:40.446) my laptop, my light on fire. I think this is a thing where we're running into the issue of the realities of today and where we need to go. And the bridge between that is, honestly, these model providers will become quicker to be commodities before this is probably fully achieved. Rex Kirshner (24:49.708) Yeah. Rex Kirshner (25:02.07) Yeah. You know, the, what you just said, like you're positive, you're negative. that resonates with me, but something I've noticed is so very often what I'll do is like, Claude, this is what I'm thinking about. This is what I want to build. Please create a plan. And then I go to my codex window and I said, this plan was just created. Please like think of how it could be improved or the, and then it. will give me suggestions, I'll copy those suggestions into my cloud window and say, what do you think of these? What should we incorporate? What should we throw out? And then like, I'll do that three or four times. And it always feels like the plan is getting better and better. But like, I have this sneaking suspicion that like what it does is like, specifically in the planning context, it makes your plans super intricate in terms of implementation. without really ever like ensuring that you're building something that makes sense. like I've gotten to point where I'm like in 12 pages of like API specs for like this tool that will be used for diagnosis on the actual system. And I'm just like, we haven't built the system. don't know, like, we don't know if this is really what it's going to look like. How are we, why are we defining API specs for like something that's like three things abstracted from like before we've even started. And so I, I think what kind of the way that you phrased it was like good side, bad side and debate things out. Like it's a little bit harder for me to think like how that can go wrong, but, you know, I just, I, like you watch these things in the way they think, and then it always sounds so good when you're reading the text, but like you take a step back and you're like, huh. This is like a little Kovasaki. Kenneth Eversole (26:57.214) Yeah, I mean, this is like why I don't know if you guys ever touched on this part. Like this is where I think we're going to watch this in real time and like, it'll be cool to watch it again. Google was not the first search engine. Like they were, they, they came in pretty late to the party and everyone was like, you're a loser. are doing this? And then they kind of spoke the floor with people. I do not think OpenAI and Anthropic will be the, the winners of all this. Rex Kirshner (27:11.054) Hmm. Kenneth Eversole (27:24.554) as I mean, a perfect example is open cloth. It's horribly, horribly, horribly flawed, but it is the beginnings of, I think, like a very rough pixelated version of the future. because the problem with the Anthropic and the Codex, they're, building for a software developer. They really should be building the tools and techniques of a software developer to write code to. Rex Kirshner (27:50.701) Mm-hmm. Kenneth Eversole (27:52.734) kind of like, I don't want to say CEO in the traditional sense of like a founder, like, Hey, I see this problem. I want to go solve this problem. I'm going to do the 80 % to, you know, get it out there. And like, it doesn't, it's not trained to do that maneuver or it doesn't have another agent to fight it into that. because that's where the nuance that you have, like, why the hell am I writing 15 API specs? Like, does he write the thing and tell me if it's broken or not? Rex Kirshner (28:21.198) Yeah. Kenneth Eversole (28:22.642) but you're, you're probably you're very entrepreneurial person. That's one of those is I met many developers that are exactly that exact statement. Like these are really talented people that will, you know, talk for two months about a writer, how to write a perfect program and rust and how to make a performance. You're like, bro, you have five people on your app, right? in HTML and who gives a shit and prove it. whereas like Rex Kirshner (28:48.118) Yeah. Kenneth Eversole (28:50.992) And that's a kind of a founder mindset and I. I don't think they can untrain that in their models. I think hyper focus on open source technology and being the engineer and it's like. I I find more intriguing that you're like copy and pasting between different models because that to me shows you like the opinions. That's where the agents warm thought came from. As I also was copying pasting between five different tools and Jim and I and. I called it weaving, like weaving between the models at one point where I grab one thing and then top it here and I have like 10 tabs open. I think that's the future. I think that's kind of what we do on a team. It's what we, my meta weird thought is like maybe this is what we do within our own brains. But I don't know. This is like me going wild with it. But I think it's like we're going to see someone figure this out like a sub agent like hey, you are the planner and you were like you never forget the goal stages like we're here to provide value to something. Yeah, I don't know. I see it like I've I've had Claude do things as like you're going way too topical like stop. I actually before this watching some of the other episodes think you guys went through and like had Claude give your insights. Rex Kirshner (29:56.3) Mm-hmm. Rex Kirshner (30:05.41) Yeah. Rex Kirshner (30:13.25) Yeah. Kenneth Eversole (30:13.438) And one of them is I called Claude a liar repeatedly. I was like, you're a liar. Quit lying to be a liar. It's like a very bold thing on my insights. yeah. I don't know. It's absolutely amazing moment to be honest. Rex Kirshner (30:29.667) Yeah. Garrett, I saw you unmuting, you wanna say something? Gerrit Hall (30:34.808) yeah, I just wanted to double click on one thing you suggested, which is that Google came after AltaVista and Lycos and Ask Jeeves and this whole generation of failed V1 search engines. I do acknowledge that it's totally possible, although I do think that this current bubble is ... The players are already set, right? The players are ... It seems very clear that there's this circular economy. between Nvidia and OpenAI, Anthropic, XAI, all these various companies that are steamrolling towards an IPO, presumably getting sweetheart conditions to IPO. You know how sometimes there's certain requirements, you have to have five years or whatever of positive revenue, and they're just like, oh, okay, not for you, we need you to IPO now. It seems like they're gonna like, do what they can to pump this AI bubble now. then presumably that means it's gonna collapse on retail investors at some point. And we'll probably have like 10 years and then another AI bubble with different players in like 10 years or five years or whoever knows when. But it does seem like the current like game of musical chairs is kind of set, like who's allowed into this particular round. Kenneth Eversole (31:55.006) Oh, completely. mean, it feels like it's like throughout a bar, it's like 1058. It's like, and they're about to start closing at 11 or whatever. it's like I worked in a public insurance company. I am now running my own. don't know if outside of being able to offload stock to unassuming people, I do not know any positive reason to go to the public market. Gerrit Hall (32:05.483) Yeah. Rex Kirshner (32:22.05) Yeah, no, I mean, that's what it's for. Kenneth Eversole (32:23.922) Yeah, like, and that's where like, okay, they want to go IPO. So what were 12 months from the bubble being popped and because what logical retirement fund would put any money right now into open AI or into the product? Like, that's what half the market is anyways, and they're going to look at their books and you know, I mean, what's the classic saying? It's overpriced. Rex Kirshner (32:46.766) Yeah. Well, let me play devil's advocate here because, know, I, up until October, I was like, obviously this is the bubble. Like look at all of the, the numbers just don't make sense. There's too much investment in like just data centers that no one even really understands. Like why, like what, what are they going to be used for other than like AI? Right. And, uh, then I started using cloud code. Right. And. I get it. The more you use Cloud Code, the more you realize that every single thing we do is automatable. And if it's not automatable by just telling Cloud Code straight up how to do it today, you can probably get 100 % of the way there if you break it up into steps appropriately and have the correct handoffs and the correct interfaces between those phases. And like, I don't, and not only is this coming for like us, right? Like the white college knowledge workers, but like, where is the other place where this stuff is being deployed? Self-driving cars, which like I live in Los Angeles. I love it. It's the coolest thing ever, but it's impossible for me not to notice that like essentially the one job that we've created in the 2010s, which is delivery drivers and Uber drivers, shitty job, but like, at least it was something you can make money. That's the first commercial deployment of AI that is going to happen. And then you think, you go to CVS, right? No one works there anymore. It's all the self-checkout stuff. Amazon warehouses, they're not going to employ people that are peeing into bottles. They're just going to have robots, right? And so I think it's really scary for all of us as humans and what's our role in society and how are we going to convince the billionaires to let us eat? Gerrit Hall (34:43.212) It gets worse. They're going to train robots to pee in bottles and displace homeless people, so you're not even safe if you lose your job. Rex Kirshner (34:47.264) Yeah. Even bleaker. But I just. It's impossible for me not to spend 10 hours a day on Claude code and every day realize I can do more and more of this and not feel like I don't understand how the numbers are going to work out, but like everything's changed, everything's changed and. Like. I understand why you would put like the next trillion dollars in. Kenneth Eversole (35:22.226) I agree, I would have agreed with you till I started my own company in this space. And I say this because I hate it when economists are like, we're in a K shaped economy. Like what the fuck does that even mean? But after I actually looked it up one time, it's like, it means like, but large group of people are doing very well and a large group of people like kind of suck. I honestly don't. Rex Kirshner (35:29.347) Go on. Kenneth Eversole (35:51.1) I think clogged in the modality on how you interact with it has made very high functioning developers or people who know technology but realize like it's not about perfect code. It's about providing value to people and building cool things and blah blah blah. Insert cool thing. They want like they won the game like the game has been won by those people. The developers are like it's about perfect syntax and. It's got to be perfectly skilled and all that stuff. Well, Moore's law kind of put your face in mud there and like I could just throw 15 VMs at it and overspend, but I've been at companies. That's how you operated it. That's what you did. But these tools have a huge cold start problem. Like if you don't know what to type, if you don't know anything about it, if you're like my mom and like, hey, here's called code in front of you. Rex Kirshner (36:32.621) Mm-hmm. Rex Kirshner (36:39.16) Hmm. Yes. Kenneth Eversole (36:48.222) It might work for a conversation, but she's not going to build, honestly, a website with it. And if she does, it will look like every other version of a website or lovable website where it's like three columns and gradients everywhere and the same icons. I don't know. think they, or fortunately for Anthropic, I mean, their bottom line or top line, they're leaning into the top part of the K and like Rex Kirshner (36:54.338) Yeah. Kenneth Eversole (37:16.454) It's incredible as a developer. I never really gave a crap about syntax. mean, we're just moving up in the up in the ball game. At one point it was one zeroes. We beat a whole war doing it. Then we to punch cards and then we went to the cobalt. We've run half the world runs on cobalt banking like we're just going to move up and that's fine. But the other half is like the people who don't know like I got into an argument the other day about this about like Joe where you use like Joe's plumbing. Joe's plumbing is going to make their own website. Rex Kirshner (37:26.604) Yeah. Kenneth Eversole (37:44.402) Joe doesn't know how to type in code to make a website. He will literally type make me website and like he's gonna spend days and months just like figuring that out and they're gonna give up. Rex Kirshner (37:48.376) Mm-hmm. Rex Kirshner (37:56.909) Again, just to take the Sam Altman side, right? Like, don't you think we're like six months away from being able to type into Claude, like, hey, make and manage this website for me. And like, it can do it end to end. Kenneth Eversole (37:58.589) Yeah. Kenneth Eversole (38:11.984) No, I think I am I like, I know it's like, I just know it's like, because I use these tools every day, and it takes the really bending at times and calling it a liar to get to get it to what I want to do. And I think that's, that's way farther than six months. I mean, we're gonna have profound innovation. mean, I'm a huge fan of this newer trend of using voice and video to manipulator. I don't know about you, but I take screenshots all the time and dump it in the cloud and do that. I think that's incredible, but I don't think we're six months away. I'm, it will happen. Like I'm not saying it will not happen. I think we're probably 10 to 15 years away purely because there's a lot of opportunity to be distracted with easy low hanging stuff. If you're at that part of the K and the bottom part of the K is like so far removed. Rex Kirshner (38:44.205) Yep. Kenneth Eversole (39:09.438) Like I a great example is that I will never in my life ever understand how to work in a nuclear reactor And if all of a sudden Claude Cohn knew how to I still would not want it to run the nuclear reactor I think there's a lot of moments like this throughout it. Um, I Don't know there is some pretty profound stuff like scientists are very technical and we just gave a really powerful tool to a lot of people I just I'm very conflicted with it. I just don't see it I don't see it's going to happen in a realistic timeframe that these companies are going to, you know, win or take all. Like I'm probably biased, starting my own company, but it's like, I just don't see it. Like look at the distractions. They're like a little kid and they can't do sort of like, we could do this, we could do that. We do this. It's like, you can't do that. Like, I mean, again, like, let's see what open AI looks like on the public market. They're going to dump the hell out of it. And like the money's not going to look great. Rex Kirshner (40:04.194) Yeah. No, mean. Garrett, go ahead. Gerrit Hall (40:09.39) Yeah, I think I just wanted to add that, shoot, like I have so many thoughts like swirling around. Let's see if I can get a good weave going here. Yeah, at end of the day, it's like, it always boils down to like, these are tools, they're incredibly powerful tools, probably more powerful than we've ever seen in the history of planet Earth, but they're still tools. And at the end of the day, it's still humans wielding these tools. So like, how does it break down? Like, I've seen people who like, Okay, so the people who were good developers before were just human beings who happened to have really good problem solving and patience skills. So they'd be willing to sit and argue with computer for 20 hours to get the widget centered properly. And now instead of just getting the widget centered properly, now they're building mass applications, which wouldn't have been possible before. But it still is because that is a human being that happened to have perseverance and happened to have patience and whatever unique set of skills was necessary to do this. I've seen this in my own life. I've seen people who are very technical and couldn't code for whatever reason, like circumstances in their life before, that have become developers, like full stack, like amazing engineers, because they are sitting with cloud code. And I also know my poor mother, God bless her, is not going to become that person, right? In all likelihood, even though the tool exists, and she probably could if she set her mind to it, there's just something about her stage of life and her personality where it's probably not gonna happen for her. And that's fine. Rex Kirshner (41:31.757) Mm-hmm. Gerrit Hall (41:33.55) And I think at the end of the day, like it also explains like what the real limitations of the technology are because like it's usually not the technology. This is the limiting factor in starting a business. It's usually like the weird combination of people interacting with other people and social boundaries. So that can be internal company politics. Like you're still going to have me with my Claude code arguing that we need to build this feature and you with that Claude code arguing, no, we need this feature and the company falls apart because people can't see eye to eye and resolve problems there. Or it might be external, which is usually the case. You have to get customers, which means you have to forget about the technology. have to go out and figure out ways of getting people interested in spending money and persuading them to do so. And that's a very human psychology problem, which, yes, you can also augment and aid that with AI. And you do see really talented business developers leaning into it and using it to hypercharge what they can do in the same way that like, developers do it. But then there's also like the bottom 80 % that aren't doing it. like, that's who we're gonna see getting fired, like the bottom 80 % of software developers that are just punching a timecard and the bottom 80 % of the it's the same thing we've seen throughout history, right? Like every time there's new technological revolution, I guess is the K shape thing Kenneth was talking about, like, or maybe you just like, or using that because your name starts with K. So it's like a good thing. The top 10 % benefit and the bottom 90 % like wash out and it's just an excuse to wash them out faster. I don't know. I hope that was a good rambling weave there. Rex Kirshner (43:07.484) man. Of course, of course. think, so this last weekend, I just went to Vegas. It was my wife's, like one of her close friends 30th birthday. And so we went to the Sphere. And if I remember to put this in, if you're watching this on YouTube, like the Sphere is incredible. I get the same feeling that I get when I'm in the Waymo, which is like, my God, like we live in Star Trek right now. Like this is the future and nothing will ever be the same. That's an aside, just not sponsored by the sphere, although I will take that sponsorship, but you should go see a show. What I wanted to say was, so my wife is training to be an oral surgeon, so she had to go to dental school, and her friend and her friend's boyfriend, they're both dentists, and so I was talking to them about AI and how, are they talking about it? Is this something that's relevant to them? And they were essentially like, no. Like, I don't know, like sometimes I use it to like essentially build templates for, you know, chart notes, but like not really. I'm like, okay, fine, fair enough. And then I was talking to the boyfriend and we were at his house and I saw like his, he has so much movie memorabilia that's like signed. And then he was telling me, he's like, no, I'm not just a collector. Like I flip a lot of it. And he showed me, he bought this one thing for a thousand dollars and he showed me his eBay. He's like, look, this just happened. He sold it for 4,000. And I'm sitting there thinking like, it would not take very much at all. Like I could do it today to spin up OpenClaw for him. Like tell OpenClaw once every two hours, I want you to go look at eBay, Facebook Marketplace, everything. I want you to surface the best things that you think would be like good buys. Send a Telegram message, I can confirm it and then you go on my behalf and like, cause a lot of these are. not even Facebook marketplace. It's like on Facebook and then you're DMing and negotiating, like clock or whatever open clock can do that. Right. And like, yeah, this dentist who knows nothing about computers is not going to do this himself today. But like, there's already so many open claw, like companies that are essentially like, we'll set up open claw for you. And like, you get, you know, opus five, like, I don't know, man. Rex Kirshner (45:31.214) This stuff seems like pretty casual to set up for yourself to completely just unlock AI to things that are relevant to you. you know, taking this back to this conversation about the bubble, like on the one hand, it seems so crazy how many trillions of dollars they're putting into data centers and like, how could that possibly work out? But you just think of everything from like all of the operations in the dental office, then all of this thing on my memorabilia, then like, everything that this guy just wants to do in his life, like make reservations or manage his calendar or like, hell, here's an idea that I've been thinking of for two decades that I think I could do now. It's like, I want to build an iPhone app that scans all of the pictures, finds and tags everyone's name, right? Like we already have that natively in photos, but then what it does is reaches out to like, Garrett and I have been, you know, conferences together. It could reach out to Garrett and say, Hey, I have. 30 pictures of you, I would love to send it to you, but in exchange, can you send all the photos you have of me back? like, man, I could literally sit here for the next 15 minutes and just come up with like real use cases that are gonna drive usage to like LLMs and AI and all this stuff. And so like, is it a bubble? Maybe, like in the stock for sure is the bubble, but in terms of like the whole dynamic, like I literally see infinite. Kenneth Eversole (46:55.858) Yeah. Rex Kirshner (47:00.738) demand for intelligence. Kenneth Eversole (47:03.71) Of course. I mean, there's definitely a demand for human intelligence. Gerrit Hall (47:05.154) Yeah, and I. And I'd completely separate out the stock bubble from the potential, right? Like we had the web bubble collapse in the year 2000 and the internet obviously went on to become a major thing. So I think it's very similar in that. Rex Kirshner (47:21.902) Yeah, but I guess like the, so the people, yes, a lot of people lost a lot of money. You know, here's, here's my favorite story, right? Pets.com. That is like the poster child for just like absurdly ridiculous Silicon Valley tech bubble stuff. Like raised $80 million IPO'd all without ever having a product. And then like a month later, They did Super Bowl commercials, like a month later, everything crashed and like that was a poster child for, this is like incredibly dumb. Either of you know, but, but. Gerrit Hall (47:53.71) That's the AI bubble I think we're gonna see I think you know open AI might be pets calm and then other companies will end up being the Google's and the Metas and whatnot. Go ahead Rex Kirshner (48:04.608) No, no, thank you, man. You just set me up perfectly because can you do either of you know who the number one investor in pets.com, the person who lost the most amount of money? Do know who it is? Gerrit Hall (48:15.832) Jeff Bezos. Rex Kirshner (48:17.324) Yes, it was Jeff Bezos. And so like we look at pets.com as this example of just the worst thing that happened. And obviously like everyone value destruction and lost money and stuff. And yet like the guys who are driving that are now like the masters of the universe. And so like, I, I think yes, Google wasn't the first search engine, but when was it founded? before and at the beginning of the bubble, right? Like, I, go ahead. Kenneth Eversole (48:49.884) Yeah, yeah, think. No, I see where you're going. I think we're going to see in a bunch of little tiny companies be founded right now. Be like, shit, this is this is definitely the future. But like for like a genuine sense, I think open eyes to two overextended and they're going to be very hurt by it. I'm actually curious, you either even know the total market cap? of internet companies in 2000. I'm gonna have to Google this. Because like, yeah, I keep running with this theory is like, I obviously, don't think anyone's debating the technology, but I wonder if there's some like mathematical way to like, the internet I'm going to assume is the internet companies are much bigger today than they were back then. Gerrit Hall (49:22.678) You to ask your AI agent. Yeah. Rex Kirshner (49:22.862) Sounds like a quad question. Kenneth Eversole (49:43.55) But the level of growth in the size of it might be a sign of what the size and impact of what it will be. So I think right now, if we were to follow the same thing, let's say there are like 400 billion back then of value, obviously Google, it provides value to the world in some capacity, like a lot. Amazon, AWS, all these. And there were trillions. I think it's like a sign of how much value and how much like impact it will have. Like the bubble is not, it's overextended because it's real, but it will then just massively be bigger in like 10, 15, 20 years. Because I mean, what you were kind of going on a direction there, like all the office staff at the dentist, does one half is like, yeah, like probably all of them are up for grabs over the next 25 years, like easy to have some sort of automation, but. Rex Kirshner (50:36.974) Mm-hmm. Kenneth Eversole (50:41.583) I'm in the camp and I think that will be better in the long run. mean, like. Rex Kirshner (50:44.928) sure. You know, the argument that I had with these dentists and my wife in the car when we were driving was like, I am convinced that every single like AI is coming for every single one of our jobs. And they were like, well, not us. And I'm like, think about it today. Like, literally, I just saw an article about how there was like a surgeon in the UK who did a surgery over, you know, fiber optic cable using the remote thing on someone who is in the US. Right. so like machines are already doing surgeries. It's just they're guided by humans and like everything that a human can do an AI can do faster, react faster, instantly call on so much more knowledge. And like, I don't know, man. And when you, you think just in the capitalist frame and just in like work, I really struggle to think of anything that a human could possibly do better than AI plus robotics. Kenneth Eversole (51:48.67) Yeah, the AI plus robotics is interesting. Gary, you were going to say something? Gerrit Hall (51:52.75) I was gonna take it back to just ask if either of you had read the book Devil Take the Hindmost, which is a really amazing, very recommended, because it goes through the history of kind of economic bubbles going back through, I think the 1600s or maybe earlier. And it's really interesting because they kind of, I don't remember if they have a chapter on the internet or not, but like. The same dynamic has played itself out so many times throughout history where there's a massive speculative frenzy well ahead of the actual technology manifesting. And then it crashes, but the early investors or speculators extract all the value, the hypothetical speculative value from it. And then long pause, and then there is the actual technology manifests. And this is so efficient that one of my favorite examples from the book is that they'd already done the full speculative bubble on rights and the hypothetical discovery of Australia before Australia was discovered. They had basically managed to get a ton of speculative investment around, it'd be huge if there was this continent down there with all this land and all this value. And they hadn't even discovered Australia existed, but they still ran the whole bubble on it. It collapsed, and then they discovered Australia. So I just. Rex Kirshner (52:54.701) You Gerrit Hall (53:11.278) I was just thinking about that when we were talking about the prior conversation. Rex Kirshner (53:15.286) Yeah. Yeah. You know, and I think in talking through all this, I think I am a little bit triggered by this conversation around the bubble because, know, my vibe is that when most people are talking about bubbles, they're talking about it more in the context of like the Dutch tulip bubble, which is like, there was no, there was nothing there. There was no like underlying technology where investors got too ahead of their skis. was honestly, Garrett, you know, it was like NFTs, right? Like it's just. Bullshit. And a lot of people lost a lot of money and then like the most horrible people extracted all the money out of it. And I guess like when I hear people talk about the AI bubble, I hear that in their voice. Unless this look, it's going to be like the internet where there's going to be this like volatile moment, but everything's transformed. I guess like when I hear about people talk about the bubble, it's always like. It's always in this conversation about how this company is going to fire 90 % of their developers and then in two years they're going to have to hire back 90 % of them, but all at a senior level to fix all the AI slop. And it's like, I don't buy into that. Kenneth Eversole (54:15.934) for them. Kenneth Eversole (54:28.254) Yeah, they make it sound like a scam. think this is not a scam. Selling tulips at overpriced value on a stock market is nothing more than a scam, but this is not that. kind of slight pivot. The one thing I think will come out of this is the death of the cloud. Gerrit Hall (54:28.924) No. Rex Kirshner (54:31.395) Mm-hmm. Rex Kirshner (54:38.05) Yeah. Rex Kirshner (54:47.374) Hmm. Kenneth Eversole (54:48.574) because of all this data center compute, now if it does crash, it's going to go down in value very quickly. The interesting thing that I haven't worked at a startup, which was on AWS and then worked at Cloudflare, there is a lot more lucrative opportunity than people know, that I think are readily to know about buying your own hardware. And in a world right now where GPUs are crazy expensive, there are a lot more people going, well, why not? just buy my own. know, spend the $200,000 or $2 million on the GPUs, put them in a data center and just pay the $2,000 a month fee. I think we're going to see a lot of change in the cloud environment entirely. Is that part I'm very fascinated by. Rex Kirshner (55:38.712) Yeah. Gerrit Hall (55:39.084) And I wanted to just quickly double click on a point you'd said earlier, Rex, which was like trying to communicate to regular people how much things are going to change and them not believing it. but I definitely agree with you that it's going to happen and AI is going to like radically transform, nearly everyone's job. but I also think that like, what's interesting is we've like, we're more plugged in than these people when it comes to like technology and software. So we've already gone through the cycle and seeing what it looks like, cause we've already gone through like. all five phases of grief, bargaining, acceptance. But for most of us, it was actually a very pleasant experience. I think we've, even probably on this podcast, have records of us doubting that the technology would ever be fully there and then bargaining, well, what could I actually do with it? But I think we're seeing a lot of really talented engineers using Claude Code and saying, yep, this is better than me, but not being despondent or in any way feeling negative emotions about that. Rex Kirshner (56:09.87) you Rex Kirshner (56:24.066) Mm-hmm. Gerrit Hall (56:36.364) just like feeling like really optimistic and like really like this is amazing and like it's incredible it's gotten there and it just means we can go to the next level and maybe they'll happen in the real world like maybe there'll be incredible like dentists who are just like wow I can like expand my dental firm to take a thousand patients or I can expand my you know I don't even know what my machine shop to like you know whatever I don't know it exists outside software Kenneth Eversole (57:03.166) That's actually the thing I'm like that really gives me joy. Like this is good. I I really think that's the best part about this like I. There is so much innovation trapped in people's heads and they just haven't had the language of like Python or Rust or whatever like insert programming language like I have this problem. It's like crypto like I would love to dabble in, I have don't have the time to learn anything about it. I'm learning other things. Rex Kirshner (57:03.404) Yeah Kenneth Eversole (57:33.086) Yeah, well, you just said, Jared, is like the exact reason why I do have hope for the future. I think there's so much unknown, unknown trapped in people's heads. It's like, if it comes at the cost of my career, being a traditional software engineer, like, honestly, fuck it. Like, that's pretty amazing. Like, there's a lot of science locked behind, you know, engineers not knowing what the science is they're talking about or doctors. And I'm glad you brought that up. I think that's super important. Rex Kirshner (57:51.469) Yeah. Rex Kirshner (58:01.165) Yeah, no, and I think like one cure to some incurable disease or like cancer or something and like none of it, none of this matters if it's a bubble or not or what, like it doesn't matter. It was worth it, you know? And I think, you know, I'll walk us out on this. I'll say this and you guys can respond, but the... I know what I'm about to say comes from a place of pure copium of my choices over the last five years. There is part of me that understands crypto is part of the AI story. It's just too much of a coincidence to me that at the same time we're creating autonomous agents, we also have an internet native property system. I don't really know. what that looks like, but they're just, it's like too much like a puzzle piece for me not to be like convinced that that's how things go. And the way that I always like kind of think about this is, and apologies audience, I said this on the last episode, but I think it's good, is that you go back to like 1995, internet's brand new and you think like, oh my God, this is gonna change everything. And like it did change a lot of things and you know, We did replace bookstores with Amazon and that may or may not be good, but the real value that came from the internet wasn't in things that existed before that we found a way to do digitally, right? It was in these things that couldn't have existed before that only are relevant because of this new internet medium, right? And specifically I'm thinking of Facebook, Instagram, TikTok, like any social media, anything with content. And I think what... What I'm looking for, for sure from an entrepreneurial standpoint, but just also from like a intellectually curious standpoint is what are these new things that can exist because we have a brand new substrate of like actual intelligence that lives online. And that's where I think that, you know, the next, like I'd say like maybe a third or a quarter of the value that comes out of AI is going to come out of automating and optimizing things that we already do today. Rex Kirshner (01:00:18.082) But the real value is going to be like, what's possible now that just wasn't possible before? Gerrit Hall (01:00:28.142) It's, it's a good, I wish I could predict the future with that clarity. Uh, cause if you could predict the future of that clarity, of course you would be like very wealthy person. Uh, it's still the first inning of it as far as I'm concerned. Like, uh, I still think we need to see what happens as the models become more powerful as the tooling like evolves. Um, it's obviously an incredibly like, uh, fast moving time. Like the conversation on this podcast I think is like changed so much and we've been doing it. What one month. like agentic time is crazy. It's legitimately like, completely different time scale than I think most of us are used to with things. and it's, it's wonderful. And I look, I'm really glad you started this podcast so we can have a forum to like, just, even if we don't know what's going on, chronicling what is happening at this point in time now, so we can go back and look at it later, is a really valuable service. So thank you for inviting me to participate in this discussion. Rex Kirshner (01:00:57.741) Yeah. Rex Kirshner (01:01:22.744) course, Kenneth will let you have the last word. Kenneth Eversole (01:01:26.822) I mean, yeah, thank you for inviting me. It's a pleasure to. Having been being building in my heads down building it well, it's good to know there's other people like this plugged as plugged in and some of the thoughts I think are really important. Like we are very plugged in. We do see the future of it and. I take kind of what you just said like I'm that's the part that I'm super happy to be alive right now. There is so much unknown unknown like. I'm excited for something and I completely agree there is a coincidence or that it was born. don't know how it works out, but the fact that the crypto hype was like four years before, two years before all of it. And then it kind of bled right into the agent economy and everything is like, not only do we have property, they can have money and we can move money around the world and we can buy services. it's like, this is a pretty, pretty profound moment and there's a lot that I'm kind of really excited about. mean, the amount of personalized, individualized little touches of business or technology, if you want to stick there that, know, words can't capture at moments how fucking excited I am for the future. Like it's a little terrifying. I think everyone's like worried about the negative and we have to be cognizant of that, but. There's also a lot of really profoundness. Garrett, I'm glad you said the idea of capturing. Again, I don't have words. We need to invent some words here in our language to capture where we're at in technology, which is amazing. At one point, it's daunting as hell to live at a time where I go to bed and I wake up like, wow. You mean you could just have agents do that now? But it's a very, like I think we should be thankful that like life used to be monotonous like 10,000 years ago. There's some beauty in that, but at the same time, like I don't know what tomorrow hold in a good way. Rex Kirshner (01:03:42.233) Yeah. Yeah, man. think that's like very optimistic and positive. you know, I think it's hard to be optimistic and positive right now, like especially right now, like open up anything and you'll see just the most horrific things happening in the Middle East. Like the K shaped economy is like really real. And I wouldn't even call it K shaped, right? I call it like six billionaires are doing great. Every single one of us is feeling more and more intense pressure every day. then, you know, there's just, there's the climate change. We don't even talk about that anymore, right? But like, there's just so much that feels existential, but there's three things that like I think about and I'm so excited. Claude code, Waymo, and now the Sphere. And I just, man, we live in Star Trek. Kenneth Eversole (01:04:31.73) Yeah, this dude is badass. I went with Dead in Company there. Great time. Rex Kirshner (01:04:36.334) Cool guys. Well, I want to take up more of your time and no audience that is actually listening can go on for more than an hour, can pay attention for more than an hour. So I'll cut us off here and just thank you guys and talk soon. Cool, all right, don't go.