The irony of all this is plays to Apple's hands in a weird way. I love how little they're spending on CapEx. I'm smiling. Everyone else is spending hundreds of billions of dollars on CapEx. And Apple's like, well, we're spending like a few billion and less than we did last year. God bless. You guys have to go have fun burning money. We're going to sit here and everyone's going to buy Mac minis and run OpenClaw.
More or less. Dave and Greg plus Sam and Jess. Put it all right to the test. More or less.
Why, hello, friends. Long time no see. Welcome to More or Less. I saw you an hour ago and quite a bit over the last 72 hours. It's quite a bit. We have been hanging out in Jackson Hole. And let me just tell dear listeners of this podcast. We're doing serious work. Yeah. We've had a lot of really interesting conversation. I've done more deals in the last three days than like the last couple months. Yes. We've been wheeling and dealing. Sam and I host an annual little wheeling and dealing event that the More and Zerky part of that has been underway. But I just also want to say there are some days that, you know, I don't know about you,
Brett, or you, Dave, or Sam, but like, I kind of like brush my hair before the podcast. You know, like you, one thinks about their appearance before publicly producing a video product. And then there's other days when, you know, 20 minutes before you're about to tape, you find yourself at the top of Jackson Hole wondering how you're going to get in time to tape a podcast and whether you will have time to rehydrate after skiing so much epic powder this morning. So today will be one of those days. That's okay. We're here for it. But winter arrived. It is nice that there's, there's snow.
I was at the top of the mountain and I had nine minutes before the conference call had to be on. Yeah. I made it. It was good. Sam and I had an excellent meeting this morning. Yes, we had, Dave and I had an excellent meeting, an excellent meeting. We had, we had to disappear from the group to have a meeting and then the meeting was a success. But Sam, here's the question. Did you do, did you do the trick of wiping the goggle marks from your face before getting onto this conference call? Or did you just own it and like show up with a ski helmet on? Look, I'll admit that I canceled a few meetings this morning that I actually did want to do because the snow is too good.
And it's interesting question about whether it's a lie or not. Cause my first instinct is to be like, I'm sorry. And like, I'm like, it's true. We've been hosting a conference, Jess and I. So we're like, I'm so behind. But then I'm like, I can't, I just have to own it. It's like the snow is quite good and it hasn't been good all year. And you know, there's a reason that I'm a seed stage venture capitalist and not, you know, running a large fortune company. So I can do this. And it's like 12 to 16 inches overnight. That's basically the. I don't think it was that much. I don't think it was quite that much, but it was great. And the last 24, it was hours.
It was at least a foot, at least foot. Well, which is just, it's just because we've been starving. It's like, we've been, we're starving. We're starving. We've been given. Sorry, dear listeners. We've been, it's like, we've been starving in the desert looking for water. And now we're just like bathing in it. I just think that the whole, this whole community has breathed a sigh of relief. You can just, everyone's like a little less stressed out than we were two days ago. Okay. Well that I'm, I'm sure that, you know, that's a great concern to everyone. But anyway, we are here because technology still matters. What's happening still matters.
And there's a lot happening and a lot happening. I had to cancel Pilates for this. God, this isn't helping. Anything. This isn't helping. But yeah. No, Sam's become a Pilates convert. Okay. But as we were convening with people over the last few days. I'm going to let that marinate. My shock on this is I'm just going to let it marinate. Sam actually asked me if he could keep Pilates and not do the podcast. And I said, no. Well, I was thinking I could do Pilates and this. This coming from the guy who I feel like I remember many, many podcasts ago. Sam comparing the people who do stretches and the people who do strength training and the
people who do stretches don't win favorably in civilizations over the long term. Oh, wow. Sam being contradictory, Dave? What are you talking about? Pilates is not stretching. I believe they call it lengthening in Pilates. Toning. I do lengthening and toning in Pilates. Okay. So Dave and Sam, let's, you know, this conference is all those Chatham House rules off the record thing. But what themes are turbocharged in your brain at the moment, having hung out with tech people for three days? Let's get, we'll go, we'll go to headlines. We'll go to Open Claw, the Anthropic Fundraising, what's happening. But this is our edge.
Our edge is the stuff that not everyone else is talking about. So Sam, to you first, what's going through your head? Well, I feel like my real takeaway from our mini conference is, you know, there's a bunch of investors here, a bunch of media people, there's a bunch of founders. And my big, big picture takeaway is the founders are working really hard. Yeah. Right. So it's like kind of wild. It's like we get a bunch of founders, you know, stay in our houses that are, and they're just like going, you know? And so I think that's kind of the thing that I think is like of the time and of the moment is, you know, there's the open claw people in that cohort.
And, you know, there's just, that's moving so quickly that it's literally, I'm very happy. I have people who are spending all their time looking at it and can tell me what's going on. Because I think it's too hard for anyone who's not in it 24 seven to follow. I think the same thing is true on like AI front, et cetera. And so to me, I think that's really the, you know, we talk about where the world is going and, you know, people are like, oh, AI is going to make people work less or whatever. No, I think what's happening is especially right now, it's just created a work cycle that is truly 24 seven for the people who are in it. Yeah.
And then the people, if you're not in it, you can, you can hang around the hoop, you can learn, but it's an interesting thing. I actually think we're going to face incredible, like I'm very energized by it, but I think one of the stories might also be that people are just going to like burn out on this, right? Because the work truly. This is a good Brit topic. She's into this. She's having technical difficulties, but when she. But the work is like truly 24 seven, if you're in it right now, it's not like, which is kind of wild to watch, right? It's like, you know, we used to, we used to think like.
What changed, what changed everyone globally is doing it at such speed that you go to bed and you wake up and the world has changed. This is, there's so much energy in the system right now. And so many people lit up and candidly, even with AI now, you know, a ton of people that are like semi-technical are contributing in interesting ways. And so I just think it's, it's a relentless level of work for the entrepreneurs. So more bots means more orchestration of humans. Yeah. It's almost like there's a, there's like a flow state that you get into working with these new tools that it's like, as close as you can get to the stream of tokens that you're
generating, uh, and being in flow with it, the more productive you are. And so I think Sam's right. There's this thing going on now where people wake up and they're like, I don't have enough hours in the day to get everything done that I'm thinking about. I mean, I had a, I had coffee or I had lunch yesterday with one of our friends from, uh, that was here who runs another large seed fund. And he was so animated about everything that he's building. He's using open claw and he had built out. Like what he described to me was a truly impressive set of systems that he's been able to, you know, wrangle together.
And, you know, he's a guy who's built several large scale software operations before, but, you know, we were both joking that our wives have to tell us to go to bed every night because you're just like so excited about the speed at which you're able to output this stuff. And I think Sam's right that I think people need to like take care of themselves. Actually. I think it's, there might be a form of, I don't want to call it, you know, people call it AI psychosis or whatever, there's like a form of working. Like way too hard and not getting enough sleep and, you know, letting, letting the excitement
and the, this sort of creative flow state that you get into impact your health. And I'm hearing this from a lot of people. So I just be careful, everyone. It is a lot of fun, but you know, uh, you should also take care of yourself. Just can attest to my summer adventures where I didn't sleep for a while. Cause I was like too stimulated by building stuff. And it's like, it's interesting. I just think now like there's a lot more of that going on. Yeah. It's definitely happening. Are you with us or are you still technical? Okay. I'm here. I'm here. I'm here. And I have to say. We're talking about bot burnout. You're, you're worried about this.
I think I called, yeah, I called this a couple episodes ago. However, I'm now subscribed to, you know, I have, I have multiple bots too, or I don't know if we're calling them agents or bots or whatever, but I need to like name them. And so the best creative branding of agents so far that I've seen is this guy on Twitter. I don't follow, but was in my feed. He named his six bots after every one of the friends from friends. So like Rachel was his personal assistant. Monica does marketing. Ross was like his engineering agent. And like he, and he put their photo as like their profile photo. So it's fun to see people having fun with it in that way.
And I do think like, it'll be interesting to see where we all peter out with the number of agents that we have doing things for us over time and how many is the most sustainable number to have. But the interesting thing for me, so my experience is, I think about like a lot of the cursor, you know, vibe coding e type projects you'd over the summer and then transitioning into kind of this more consumerized version with open claw. Now, the funny thing for me is I know how all these things running around doing things for me that I don't even really care about, but it's like not even worth shutting them off. Right.
This is like, there's like, this is like, there's a heat of the internet is just heating up with all these things, doing things. And it's funny because a few things that I've been messing with over the summer, I like stop paying attention to them. They're fine. They do, but they weren't like useful enough. But it is funny because this rev around, I was like, oh, I'm going to like boot up a clod bot. And then I go see what those bots have been doing for the last several months. And I'm like, oh, good. They've done all the things I wanted. And now I can leverage it for the next thing. And so it's kind of this weird thing, you know, how many bots are in our lives.
It's like, I think about more is like, they're just daemons running around the internet on our behalf. And as long as you're willing to pay for them. I love calling them daemons that think. They're daemons. Yeah. That's what they are. How much are you spent just for the, for the point of, you know, the listeners, how much are you spending on these bots? I mean, I, I'm self-interested in this as well. I don't have any idea, but my, my bet, my bet is I'm spending, my bet is my full like AI tinker budget of things running around doing things for me is on the order of like $2,500 to $3,000 a month. Okay. So this is not a mainstream thing. No, no, but it'll get cheaper.
Well, but no, Jeff, you can get a lot done with the various Claude and open AI, you know, the, the pro subscriptions. So, I mean, you really, you really can't. But I have like all this digital ocean. Same. I have all this digital ocean infrastructure running. I have like all this other stuff. The thing about it is that it's like, it's just interesting because like, you're right. Any person on a budget who is thinking about this stuff, there'd be all these things you want to shut down. But I actually think there's a bunch of people like Sam and Dave running around and Brit who are just spinning things up and they're daemons that run around the internet
for you and you forget about them. And it's not a question about- You call them daemons? What is that word? No, a daemon is a technical term. It's a technical term that's been used in operating systems. Explain it to me. It's, it's an old term from early operating systems. Like it, in your Macintosh, there are a ton of daemons running around and doing various things. You know, anytime you like, if you open your activity monitor on your Mac- That, I know what that is. Oh, yeah. I know daemons now. There's a bunch of daemons that are like running in the background, making sure your files are synced with iCloud. They're things that are running in the background.
They're just doing things. Their background things. They're not like directly being controlled by you. Guys, guess what wait list I got off of this week? I don't know, but I'm glad it's shifting the topic. It's a Google lab. It's a Google labs product. It's another, maybe daemon, but it's called CC by Google. I think we talked about it in a certain pod. Oh, yeah. I got off that a couple weeks ago. Are you using it? Do you like it? I'm curious to get your reviews. Yeah, yeah. So if you guys remember, CC is just sort of, I mean, it's another, Google is winning the AI summarization battle in terms of like your existing products.
So they have this assistant that kind of emails you like summaries of things you might miss from your inbox or calendar. And it's sort of random what it gets, but I also have two different Gmail accounts that I'm not sure which it linked up to, but it's helpful. It reminded me of like forms I had to fill out. For school. Yeah. Yeah. This is like every mom's saving grace. But I am also a big fan of those extended summaries of your threads in Gmail. Are you guys getting those? Sam, I know you are. Those turn out to be useful. I got to say, this goes back to the whole like. They are very useful. It's the thing that's super useful is it'll be like, you'll be on the thread.
And like the one that was like, oh, that's like pretty good was, was when there's just like this deal we're working on. And it just actually nicely summarizes all the deal terms at the top of the email. Cause I wouldn't have remembered. And you're like, thank you. That's useful. Like it's not earth shattering, but it is darn useful. Once again, incumbents win. The incumbents are just going to crush it with AI. Yeah. So, but Britt, what was your CC reaction? Cause I, I, I'm a fan of it. Well, I was actually underwhelmed on CC. It got a lot of things wrong. It had my kids to do's and different orders of things for like school stuff.
And it was also processing a lot of like spam and marketing content in my inbox. It was like, don't forget you have a promotion for United. And I was like, no, I don't care. But I have a lower bar. Cause it's, that's what all, but it's, I don't know. It's still like, it's just another thing. Like if, if two of the things are useful, that's helpful to you. Right. I would say it was, it, it got more than 50% accurate and I can imagine it's going to get even better over the next year. So I do think this is something a lot of people will use and it connected stuff all like automatically into my calendar, which we've all been calling forever. Right.
Especially with Gmail and Google calendar. So, so I do think like moms are going to win big in this email and calendaring assistant world. Because this is like, and this is the moment, right? Of like summer camps. Oh yeah. It was like, don't forget summer camp sign up for this thing ends next week. And I was like, ah, so I have all day calendar reminders in the month of February to sign up for summer camps and the summer camp the kids went to last year emailed me because I had yet to sign up and did I want to go back? Oh good. Showing that my system is not working at all, but their system is working just next year. Your agent will just do it for you.
You won't even have to remember. Yeah. Well, I want to let any other, I think all these agents are an ongoing more or less theme. So we, we should focus if there's Delta, anything new that we think is worth sharing with this group based on our early adopterness. I think since we last taped this pod, I became a, an open claw agent user, thanks to my husband's technical prowess. And I'm using it very consistently, although it is buggy guys. I mean, the bugginess is, is real. I've seen multiple startups in our portfolio in the last week, starting to create skills and package skills and open source them for open claw, which is like really interesting
that the startup community is already on. So on top of this. Yeah. I mean, we also had the open claw. I think we talked about it last pod, but we had the meetup. Oh, we had the meetup. Yeah. Dave, give everyone the debrief. I was off in my timing. The meetup was one of the most extraordinary things I've experienced in Silicon Valley for, I don't know, maybe since 2007, we had a thousand people on the wait list. We ended up with 1500 people that showed up. We had a bunch of the OGs from the, you know, the original internet. I mean, we had Kevin Rose, Marissa Mayer, you know, Mariam Nafsi, who founder of Minted and current founder of Arcade was there. Um, Ashton Kutcher.
I mean, it was really extraordinary. And when Pete walked in, it was like a rock star walking in. Pete is the founder of open claw. Yeah. Pete Steinberger. And, you know, it was just really exciting. And the energy in the room felt really different than everything that I've experienced around town lately. And people kept coming up to me and saying, will you please do more of this? And I kept asking them, what is this? And they just said, developer meetups where people can hack and learn from each other. And I said, isn't that happening all the time? And they said, no, everything's so corporate.
I kept hearing that over and over again, which I thought was just a really interesting thing. And so I, it left me wondering, you know, those of us that grew up in the 2000s and the early 2010s in the internet, we used to do this kind of meetup all the time. And a big part of making sense of the big opportunity of the internet was meetups. You know, we're all going to get together. We're going to show what we've made and tinker together. And figure out how to make sense of all of this. And that kind of gave way to this like grind core culture of SaaS in the 2010s, where people were, you know, trying to get their bag and make their money and compete with everyone.
And I'm not going to tell anyone what I'm doing. And it's like this world of tokens and what we just talked about has created a whole new enormous opportunity set. And nobody knows how to make sense of it. And so it's like the, I think we're shifting back to the culture of let's all get together in a room and meet up somewhere and try to figure out like what to make with this stuff. And so I think that's also the cool thing that's going on here is it's really kind of brought that hacker tinkerer developer community back out into the culture. And that's, that's like exciting to me. So it was, it was really neat on that front. And where does it go from here, Dave?
What do you think the next three months bring? Clock on international guys, it's happening. We're spreading it around the world. We actually are. Yep. We're doing clock on Tokyo. Why, why'd you start Tokyo? Well, cause we happen to have a business trip there. So we're just going to kill two birds with one stone. There's two reasons. One, we immediately got a bunch of interest from Japan, um, coming into the sort of, there's a core set of us that are kind of involved at the, the center of the open clock world. Is that good? I feel like Japan always adopts the thing that doesn't matter. No. Robotics. Look, Sam, I think that's a reasonable. Japan is a weird country.
They do really cool stuff. But if you think about like globally, like Japan was like super deep Twitter when no one else was. Japan is like super deep, different mess. Like Japan always has a Japan world, which is different than what works globally. So I'm curious if you think the fact that it's so popular in Japan is a problem. Uh, I look, Sam, I think you're, you're right in some aspects, but I also think that if you remember in the 2000s, Japan was a very interesting place for the consumer internet. There was a lot of really cool projects. Yeah. Ringtones. They were, they were first on ringtones. The 2010s was not as interesting.
And look, they weren't the only ones that came out of the woodwork. I mean, we've got London, uh, clock on New York, uh, John Borthwick and Ikram are hosting that. We've got London. We've got, uh, Vienna happened two days ago. 500 people showed up in Vienna. Maybe Austin, South by Southwest. Austin at South by will happen. And, you know, to me what this looks like it, and this is why I keep referencing 2007. When I launched Facebook platform, we had this similar thing happen where we just had like meetups explode all over the world very, very quickly. And, you know, the same thing happened then when iOS went from being not available to developers
to being available to developers and then iOS and Android exploded similar thing. And so it's a very, I think, well-trodden pattern when there's a technology, developer technology that you can build on that feels very personal. Right. And I think that's kind of the big difference between the last 10 years and what we're experiencing now is that this just feels very personal. You know, you can put it on your computer, you can tweak it how you want. You can share the cool things you've made with other people. That's one of the other cool things that we're seeing a bunch of is people having their bot teach other bots the cool things that it's figured out.
I've been in this group chat where a bunch of this is going on where people have figured out their own workflows. They've figured out some really cool ways to use it. And they're like, oh, I really want to learn that. Can your bot like make a guide for my bot? And so it's, they're like self-improving each other, which is kind of interesting and neat. I've seen a bunch of that the last few days, which is kind of, which is cool. So, you know, I think, uh, Jess, the answer to your question is like, I think we're going to see a million developers on this thing within 12 months. Well, would there even be developers though, Dave, is like a different way to describe what it is.
Because I think like, to me, it's like, there's an interesting thing. I see two things going on specifically with OpenClaw in this moment. One is, look, there's a lot of people whose jobs require a bunch of repetitive work that they don't really want to do, but they have to do. And it's really funny because it's kind of like being in college again, where like the procrastination gene, where you're like, oh, there might be this thing that can do my thing for me. And it's way more fun to think with a thing than to do the thing that's repetitive and boring. So it's like, it's a really good cover. So like I have this, I'll make fun of, um, Jack who works, who works at Slow.
He's an associate on the team where like literally two hours ago, I was like, okay, so what are your priorities for the next week? And he's like, well, I'm going to spend like a bunch of time on OpenClaw and figure out, I think 90% of my job can be done by OpenClaw. And I was kind of making fun of him and laughing about it. Cause like on one hand, he totally should do that because I want him to go through the whole process of experiencing it. I need him to like understand the ecosystem. I also candidly want him to write guides for the rest of the team that might be less motivated, unfortunately to like, so I'm in, he should do it, but I'm also like, yeah, of course you'd
rather play with OpenClaw than do your job. It's like way more fun. Right. And like, so there's like, there's one spit of people like that. Right. And then there's another spit of like, look, you're not really technical. You're not even, you're just, it's kind of like playing on a computer. And so is it developers or is it just people like playing on their computers and building, you know, little like, and they're basically like, they're kind of like doing like, you know, you're using like MS Paint in 2026 for fun. Cause it is fun. Right. But they're not really developers as much as they are just like people, you know, I don't know about that.
I mean, Sam, I think I was telling you this yesterday, but one of the frames that I use for this is that there are around 3 million iOS developers in the world. And there's around another 4 million or so Android developers. And, you know, if you look at Meta's last earnings report, there's on the order of call it 4 billion users of software in the world. And so the gap between who can be a quote developer of software and the, and the users of the software is like really enormous. And so the idea that like new people are coming on board to be developers of software and you're right, it's like a type of tinkering that's different.
Or they just not, there's no such thing as developer anymore. Like, it's just like, they're just people doing things. Like I just, this is, I agree with your macro point. But I think that's great though. Right? Like they can write it in English. And nothing is bad. I'm just saying like, in some ways, like what is a developer anymore? Right. In this world. Right. There are obviously. I just feel like it's a person willing to improve the computer. Right? Like, are you, you know, not everybody in most people that get an open claw, they're like, I collected it to my, I connected it to my Gmail and my calendar. And now it's summarizing some stuff for me.
What's the point of this anyway? And then on the other hand, like the guy I sat down with yesterday, like, he like built an enormous software architecture. Right. And decided to try to improve the computer. And isn't that the difference? I don't know. I just don't even know. Like, I just think it's an interesting definitional question, which might matter. I'll give you an example. When iOS first kind of became important, it was like, oh shit, we need to build apps. There's a tiny number of iOS developers in the world and like, it was kind of like, they were the AI engineers of their day or the researchers of their day. It was like an incredibly valuable skillset. Right.
That then basically people were paying million dollars for trainings, right. To get, to take someone who's like a web, cause it was such a scarce and clearly important thing. But in some ways, like to be a developer is like, okay, you have, you're like, you're in it not for yourself or for your own use cases. Like you're built, you're intentionally building software for others and there's kind of a barrier to entry. I wonder if, if we're now in a world where everyone's just building shit for themselves and it just turns out that because it's so easy to like rip and replace and replicate
software that people just by building for yourself, you're also building for everyone else. I just wonder if it's like a very different paradigm. And I'm not saying that people aren't going to like mess, but it's like, what this distinction of like, are they use a, are you a user open claw or developer open claw? If it collapses one way to say, well, one way to say is, well, all the consumers are developers. Then it says like, there are no developers except for like a tiny number of maintainers maybe. Right. And really what it is, is like just the pattern of being a user includes packaging things in a way or things end up naturally like flowing in for everyone else.
I guess to me, this is the optimistic take on the future, right? There's a lot of doomerism going around like, oh my God, we're going to have a social unrest and everyone's losing their jobs and nobody can, you know, all like, I don't know. I saw another quote this morning. It's like all like 100% of white collar work will be automated in 18 months. Right. And I just don't see that happening right now. What I see is thousands of people showing up to meetups and people writing their own software for the first time. They don't have to use the thing that, you know, is the big SaaS app. Right.
And it's not like everyone's going to do that, but like if we have an incremental, like tens of millions of people writing software for themselves and writing software for their small teams and writing software, even for their big teams. Like I view that as like a net positive for society that like more people are influencing the future of software. And the way that all of us, all the 4 billion software users in the world have more customization, like more sovereignty over their software experience. Like that seems like a net positive to me and that more people can participate in that process. Like seems pretty good. Okay. We got this very high level. It's very early.
What should the listeners do? They should just try it. Yeah. I think it's like get in the game. Yeah. Just build guys. You're doing open AIs marketing for them. Just build. But it's not open AIs. This is actually important. It's everyone's marketing. This is like the whole thing. This is actually quite, I think this is, you hit on an interesting topic. Oh, interesting. Okay, great. So here's what I say. This is not good for open AI. Say more. No, you mean quad open claw. No, this whole direction. I mean, like if you think about it, like the direction, like if you're trying to make like there's kind of the layer of like, okay, token count, number of tokens used in the
world goes up, right? We can have arguments about intelligence packages, tokens, and like we can talk about how it goes up. There's definitely this monolith strategy, right?
Like I said, I wrote tons of tons of tons of tons of tons of tons of tons of tons of tons of that plays out from like a fundamental model. But here's the thing, it almost doesn't matter because if like the heavy brute force of LLMs is not a winner take all, it's not going to be an race to AGI, there's going to be a bunch of back ends that can do it. It's just like multiple cloud providers. The interesting game becomes on the edge. And things like OpenClaw are the decentralized version of it. So it becomes kind of irrelevant what's in the back of the stack, right? It's very anti-open AI is what I would say. I don't know. I mean, you're still going to be selling tokens,
right? Like these things, even OpenClaw consumes a lot of token, right? Totally. But here's the thing. That's the thing. That's like the bare metal version, right? So if OpenClaw, like I can point OpenClaw at six things or one thing or whatever, you get abstracted. It's like a top of wallet thing. So if my, if all of, if the future is a decentralized OpenClaw of people running daemons on whatever, and they're all interacting with each other in a cloud, right? That isn't vertically integrated, like OpenAI stack, then there's, yes, OpenAI can still sell 10 cents on the dollar of value tokens and like they can compete in that
market, but it's a very low margin, bad market, honestly. And their whole story has to be for it to be worth what it is, is like this very monolithic, we're so much better that we subsume all like all this noise and open that doesn't work. We're just so much better that you come to us and we are the top of wallet and we are your acute consumer experience and we are the advertising interface and we were all this stuff, right? So I actually think like what we're seeing now is kind of, you know, we always know there's like kind of closed vertical versus open stories. And for the last two years, the story was like, okay, we know what the closed version of AI is going to
look like. And those seem like they're getting powerful. You know, who's going to be the dealer of open source AI? And I think what we're actually seeing is like, there is going to be actually a lot of the activity is going to be open source on AI and distributed, but it just might not be at the bare metal level, right? It won't be integrated that way. It will be just like all these daemons running around on people's machines everywhere, right? Interacting with each other in a cloud. Someone described it and what you're saying someone put really concisely and it was something like- Better than I did. Yeah, fine. Just kidding. Podcasts aren't for concise.
They're for going long. That's what our editors are for. But it was basically like taking the AI and running it on your computer for yourself instead of accessing these sort of omniscient, all-knowing models. What you guys heard the same thing. A local model? Do you mean a local model? But the idea of like the power of it is just like taking it for being, it's what you were saying, Sam, the old vision of it and AGI being the pinnacle of it to like basically everyone on their own hardware is running their own AI, open source AI for them. And that is- Well, but it's not exactly that. They're running a AI orchestrator locally, right?
They've got a very powerful like computer that's like- We're getting very technical people and, and, and, you know, the billions of people on the planet, you're going to have to explain it to people who aren't like you. So try. Well, but I guess that's what I'm saying, Jess, is that to have a, some software that's on your Mac that can orchestrate across all of the different AI models to do jobs that are specific to what you're trying to get done. It's like, as if somebody assembled an orchestra, you know, before open clock came on the scene, it was like somebody assembled an orchestra, but nobody had written a symphony.
Nobody had enabled you to work across all of the stuff, uh, in a really simple way and get all the power out of all of it and make decisions, right? Like you do need the big models like open AI or anthropics models to get done things like making software, like planning really complex things. You don't need them for some things, you know, you can do it locally. I I'm trying to think of the person who for two years has been hearing that chat GPT is the cat's pajamas. They started using it and now they hear about this new thing and they don't understand how it's different or they don't understand how it's exciting or they don't like, that's where
I think the sort of running this yourself locally, even if you're not running the full models is a powerful way to understand it. Yes. And then we're gonna talk about x.ai. So look, I think the way to think about this is what the reason open clause actually interesting from my perspective actually comes down to trust and identity like most things. Right. And the pro there's a future world where in theory, this is like the monolith world where you say, okay, I'm going to give up all sovereignty to chat GPT. I'm going to tell it everything about me. I'm going to give it every password in my life. I'm going to like, let it truly own everything about me in a central way.
And I will trust it to that. And like with all of that centrally, that's like a very dystopian view in my world, but it's possible. Like it's a monolith. And by the way, that would be very valuable for chat GPT. There's another version of the world that says, look, one, that's not going to happen. It's not gonna happen for two reasons. One, you're not going to trust them with it. And two, they're not going to want it. They'll be able to take the liability of having it all. Right. So if you, right now, if you said to someone, Hey, do you want to give all your stuff to chat GPT? They'd be like, no.
And if, if, if, if chat GP said, Hey, do you want to have everything on a person? Like, and be responsible for all that. And if they make a request of you, like you have to do it regardless of what it is. There's like no scenario where that's okay. Right. Cause like these bots, by the way, some will do things that are like kind of malicious or semi malicious and you just don't know what's going to play out. So this, this kind of decentralized sovereignty thing where you're like, okay, I'm just gonna like run my own computer in the cloud or locally. It's my responsibility. I give it what I want to give it, but I'm a lot of people willing to give it a lot, but
no central big board company took it all effectively. I think that's a world that's going to move much faster because people will be way more experimental and do weirder stuff with it. And no central entity has all the liability of it. Right. I think this actually is a liability thing and story as much as anything else. And so as a result, it's like open AI, people talk about like tokens as like this form of intelligence and like what happens with liquid intelligence and things like that. I think the problem is, is like, if you don't own the sovereignty and the data and the interface, and it ends up being decentralized everywhere, you can still deal heavy metal.
You can still deal raw intelligence. It's just, unless you are 10 times smarter than everyone else, it's super hard to sell that at a high for a lot of money. You sell for a little money, right? Yeah. Like, but not a lot. Right. And so I think that's really what's happening is like, you're seeing things like open clause away. It's going to basically commodify the entire heavy industry of AI that's developing. Um, and that just massively changes its value and prospects and how the future looks. The irony of all this is kind of plays to Apple's hands in a weird way. Right? Like I have to say, I am a perennial Apple hater. I love how little they're spending on CapEx.
I'm smiling. I think it's the greatest thing ever. It's so awesome. It's like everyone else is spending hundreds of billions of dollars on CapEx and Apple's like, well, we're spending like a few billion and less than we did last year. I mean, that just the fact that they went down, you know that they were just like, let's- I mean, it's just awesome. It's like, it's everything I believe in, in life where they're just like, yeah, you guys go have fun burning money. We're going to sit here and guess what? Everyone's going to buy Mac minis and run open claw and you're going to be in bad industries. And it's just good for us. Like that's like a wildly fun story in my mind.
I mean, they make personal computers, right? Right. Well, it turns out if your bot also needs a personal computer and it's your sovereignty and your, I mean, that's great. But I mean, I think like, yeah, Jess, we are now in the era of personal AI. And that's what this comes down to, which is that people are going to have their own AIs now and you're going to be able to, you know, have the data, keep it safe, do what you want with it. Or to keep it not safe, because it's your problem. Yeah. It's just, you can define the AI how you want. And I think that's why, I think that's why people are so excited. I think that's why people are showing up.
The developers are showing up to things. It's just like, this is exciting. I have control over it. I have full sovereignty over this thing and I can do whatever I want. I can have, it can have memory. I mean, one of my favorite things that we've done recently is we, we, I call a call it the hive mind version of an open claw. Like we put a open claw at the center of our firm in Slack. And so it like becomes this like brain that's like everyone's interacting with. And that's really interesting use case. That's like totally different than using it just for yourself. Right. And it can be really powerful, but you know, if you're all talking to the same bot, that means
that the bot will tell everyone what everyone's talking about. Right. And so it's like a totally different type of almost like a different species. Um, and it's super fun and it works really, really well. Britt's nodding. Well, it gets smarter because everyone's talking to it more. And I think some snowball effect with these agents where, you know, you have to persistently train them and give them more memory and context. But if like a dozen people are doing it every day, then it's just getting smarter and smarter, much more. It's interesting. There's, there's an old Kevin Kelly thing.
I haven't read that much of Kevin Kelly's writing, but I think I'm pretty sure this is him. And I think it was a very smart way to think about it. Maybe it wasn't Kevin Kelly, someone of that ilk, someone like Kevin Kelly wrote this thing about how a lot of what actually happens in technology is you centralize the value of the technology and then you have to radiate or like push off all the liability. Right. And like that, this is kind of a very cynical view of technology, but it's like kind of the view of like the way you make a lot of money in technology is you centralize and control the value.
But then you basically socialize or like throw off all the risk associated with it to the edge and let other people deal with it. And like, that's the martin is a very cynical view of technology, but that is, it was an interesting framework. And I just, it's interesting because I think the big monoliths like open AI have this problem where like they're, they, they told a story where they were going to centralize all this intelligence and race to AGI and be first initially. And they'd have this like unassailable technology thing. But then they also were kind of like implicitly soaking up all the liability along with it, if that makes sense. Right.
And all the, and that creates all these weird issues, but what it can't do and can't, can do and can't do and who gets sued and whatever. And open clause is this interesting flip where it's like, well, wait a minute. First, you're not going to centralize all the intelligence. Like you might have a good intelligence foundation and other people will too. And like the race to AGI only one person wins isn't, isn't the right way to think about it. That was marketing. But then too, with open clause, all of a sudden, like everyone can own their own liability for the first time,
which means you can do way more interesting, weird stuff and like move faster and whatever else, because you just can't have a monolith take that level of liability. Yeah. And it's interesting if you look at the New York Times op-ed that happened this week, right? Somebody, I think quit open AI and basically said, I quit open AI. And the reason I quit is because open AI has the greatest database of human thought ever assembled. And they're not going to be able to keep themselves away from it when they're building their advertising system. Right. Which is like the most predictable, I guess, op-ed you could possibly imagine as, as ChatGPT starts to put ads into the game.
It sounds like, it sounds like it was written by a bot. I think it was placed by Anthropic, this op-ed. I know. I think it was probably Brit. I think it was probably placed by open AI because they're like, this is the brown. This is like more of the 201 strategy of marketing. Totally. Okay. You have to leave and write this op-ed to fear monger how incredibly special and powerful our advertising model is. Totally. No one else can possibly have. Totally. There is some truth to that, but it's also the most New York Times op-ed. Let's just say that as well. So it really just checks all the passwords. That is, I will say this.
It's clearly an op-ed that end to end could have been produced by AI. Right? Like it literally is the AI op-ed. Like, oh, the New York Times is like using their Claudebot. The New York Times is using Claudebot being like, write for us the op-ed we would write about this. Guys, I will say like, I'm talking to more and more journalists now who come to me and they're like, so I'm not allowed to do this. But like, I should totally be feeding my interview notes into a chatbot to give me a first draft of a story. Right? Like, and they come to me because they think I'm going to be like, you go girl, you do that. You screw your organization and their roles.
But I, I actually not sure that that exact workflow is going to yield like great journalism at the moment. But the principle is like, yeah, you should be trying to focus on getting the original information and maximizing everything else you can do with that via AI. Absolutely. Like there's no, but what's interesting is more- I mean, engineers are doing it, right? Yeah. And I'm also hearing that Nashville songwriters are doing this. Like we're hearing from our songwriting friends in Nashville that Suno is like- Your songwriting friends in Nashville. The more it's half songwriting friends. Guys, we know people in Nashville. I'm dead serious. Like I, I've had- I believe it.
I just think it's a funny line. I just think it's a funny line. Walk me through their Suno based, and Jess, your, the workflow you just said is literally what songwriters are doing now. Like they use Suno to get the thing going. And then, you know, it's like a core part of the workflow now. There's just a lot. I think there are some bridges we need to cross slash lawsuits. The publishing industry needs to win first to really unleash what would be helpful to newsrooms here. But it's interesting to me that the journalists now are, the smart ones are kind of figuring it out, which isn't shocking. Okay.
Speaking of people leaving companies, XAI, which actually is not a company we talk that much about on this pod an hour later, is experiencing an exodus. It's not even a company anymore, right? It's all SpaceX. Right. Correct. But despite all being pre IPO SpaceX employees, there is still an exodus. Is it an exodus? It read, it read more like they are reorging to me. Wait, what's the data? Can you explain? So they lost two co-founders. So I think they're down to not a large number. Aren't there like 20 co-founders? No, I mean, the people who follow it closely are very concerned. Really? Yes.
And I think it's especially the, as you guys know, not all AI engineers are created equal. So. But Jess, isn't this also like what AI company, it's like, there's so much drama. Like all the AI companies, they're just like, we're like, we're like mass quitting and we're all going over here and then we're all going to go over there. And it's just like constant. Right. Well, it's because it's because at least in theory, like, like you think about like, it's all just like power. Right. And in theory, if the researchers are so valuable and capital is not valuable in the AI context, which is what we've been led to believe. Right.
Like then, then obviously the drama is with the people who have the quote unquote leverage. But are they still valuable, Sam? I mean, Anthropics, Anthropics said in the last, in their last announcement that Claude now builds Claude. So they do, do they even need the researchers anymore? You have like the three people that, you have like the three people that attract the other people and like, but it is interesting. It's all the shitload of engineers and researchers, but wait, Jess, why, why wouldn't you believe that these people just got acceleration after the deal? I mean, I think. Elon would never let that happen. Yeah, exactly. That's not his style.
And all it, by all accounts, the IPO. Elon, the guy who sued the guy out of his Twitter comp package is not letting a bunch of XAI co-founders walk with all their equity after he sacrificed 20% of his value of his SpaceX to save them. I mean, it's just like, it's a real, we've talked about this on the pod. It's like a real jujitsu move, right? Like you, you sort of sneak this acquisition and then, I, I've actually been surprised by how nervous sort of collective investors in the space are about this, because I sort of agree with you guys. Like it's very personality driven, a lot of personality clashes, a lot of other great options for the, these truly talented people.
And Elon is objectively a tough person to work for. And that's saying, you know, that's being kind about it. Shocker. I think, um, but, but it's really shaking people. And, and I think the information of my reporter, um, or my colleague, Theo, wait, has a great, he, he broke the reorg that ensued from it and has a deep dive into the new structure there. But I guess my point is things seem pretty shaky over there. They've got infrastructure problems, they've got model problems and now they're losing people. So there might be some more twists to this. Here's the thing I thought was most interesting about this announcement was that they made a
point to talk about that this, I think there were four, there were like four points in the reorg and the fourth point was about macro harder. And they made a specific point that this is an agentic engineering, like organization where only agents write software. And back to your first question, Jess, about what, what are we seeing this week at this conference? And, uh, and then even last week I went to some interesting private meetups about agentic engineering and this concept of the people are calling it the dark factory or the dark software factory. I was going to ask you about that. I heard you use that term. What is a dark software factory?
So a dark software factory is basically you're running a agentic engineering organization where you, uh, the code must not be written by humans. The code must not be reviewed by humans. And you write tools to understand what's being built and to keep the throughput and the, uh, bottlenecks in the speed of your development at the highest possible rate. And so it's all being generated by agents and all the humans are doing is seeding it with the initial, uh, seed and then watching the software grow and validate against the tests that they write on the other end. And this is like kind of in the back channels of all the CTO conversations in the valley,
those who have kind of bought into this, there's like a divide between the people who are, you know, have bought into this ideology. Who's gone full dark factory? Like have any companies that people have heard of? Yeah. One of the guys that's most interesting talking about this is a guy named Justin McCarthy from strong DM. It's a, one of the bigger cyber security companies. Um, he's out there talking about this a lot. Um, they've done a lot of really interesting work. You know, I think that anthropic saying in their most recent posts, like Claude builds Claude, I don't think they're joking.
I think that, uh, you know, I think opening eyes like really headed down this from other conversations that I've been having around the open class stuff. They're really like headed down this road. And I think that this is actually one of the more interesting things going on ideologically, um, in how these teams are being built. But I mean, that's what also XAI said this morning, right? Or yesterday morning. So Dave, you think that basically they're going to be way, way, despite what the, the party line is not that tech companies are going to employ far fewer engineers, but that they're going to get way more productivity out of their engineers.
But you think there'll be fewer too. Yeah. I mean, some of these operations like are three people that are producing not just 10 X, we're talking a thousand X, the mouse amount of software. That is also the open claw thing, which is like, if you know, it's, it's a crazy wild system, but like it is actually incredible how quickly they move. Right. Like, and like the way they process, you know, diffs, et cetera, as I understand it is like, they're basically like operating as a several thousand person company with almost no one because of the leverage. Now I will say one thing. But saying the other thing on the diff thing I'm hearing is that nobody's using Git anymore.
They have these like merge bosses that, uh, you know, basically deal with the semantic, uh, merging. Well, Git is clearly designed for at least GitHub and Git kind of broadly as a human centric concept. Right. Like, and it's an interesting question of everything about legacy. Does that get upgraded in a more agentic way or does it get discarded for a much better platform? I think it's already being discarded. Right. I know. And that's really interesting. Right. Being short GitHub is an interesting place to be. Right. Like, and it, it makes sense. If GitHub, if GitHub was the winner in a world where human identity as a developer and what
you were, you know, you're contributing mattered, right? It's kind of, it is one of those places where I could totally buy it getting disrupted. I think in general incumbents win, but they might lose. Jess, I'm also hearing that people are, uh, there's people who are saying you must spend at least $1,000 a day in tokens, ie. Or spend as much as your salary. Like one of the teams that I talked to, they have the rule where you must spend your salary in tokens. And then you're into token economics. What, what does that equate to? A thousand dollars a day in tokens? That, that's. I'm not sure actually. It's always tokens. It could be crypto. It could be AI.
It's tokens all the way down. It's tokens all the way down. Oh God. One thing that is interesting. I don't actually believe, like, I think honestly the XAI stuff is deeper than this and like probably more problematic. Well, here's the interesting thing. I do think it is funny that like the engineers that are at the center of the AI stuff are like digging their own graves, right? To some degree from a career perspective. What do you mean by that? I actually believe that like you, you lay the foundations on these systems and then they're just, you know, anthropic is mostly just computers updating computers and it's all AI.
And like, you don't want humans reviewing the diffs anymore. It, you really are building like a doomsday career for a machine for your own career. Now I'm not saying you shouldn't do that. The reality is unless you think you were a one of one that will solve the problem, it's going to happen anyway. So you might as well race into the oblivion. But you know, if you believe, which I think is probably right that you'll have, this goes actually to our conversation about what's a developer, right? Like if the future is there's a absolutely minuscule number of people, right? Who are smart enough to point 0.001% and, and basically online 24 seven and like wired into the machine.
And they're super valuable and rare, but like, that's like the almost no one. Right. So no one is a developer there long beards, very long beards, like wizards, people. There's like 10 wizards in the world that interface. And then everything else is done by machines. And then consumers are now just like writing software for themselves that happened to be communally updating it, but they're not developers anymore. They're just like participants. It's kind of like, there was probably an era where like, if you were, if you knew how to drive a car, that was like a skill. And now it's like, well, everyone drives a car. It's not like a job anymore.
And then it becomes, oh, no, the machines take over. Like there's something going on where like the whole concept of what a developer is is splitting into the 0.001%. There's no more professional developers and everyone else is just a user of technology that happens to have technical leverage. That David, this is what we were debating last night and you were against me on this of like, how many jobs get taken and all the jobs. No, I don't know. And that like that companies will not need as many people to do different jobs, which I think is obvious. But you were saying that wasn't going to be the case.
I've already said it in this podcast, my case that like there are going to be millions, if not millions, tens of millions more people making software that is more custom. But that doesn't mean they're getting paid for it or that they're getting paid well for it. I think what's going to happen. Sam, we know a lot of people making a lot of money with this stuff. Or that Salesforce needs as many salespeople or analysts or, you know, like there's so many jobs that are going to get crushed. People really want the story to be that there'll be all these new jobs for obvious reasons, right? They don't want to think they're destroying jobs.
It's a really time but the reality is, is like, I personally think that all this stuff just gets washed away. Right? If you take out all the way and they don't come back because there is no P50 job anymore. Like there's a P99 job. If you're truly the best in the world at something, you'll be fine. Right? But everyone else is not going to be fine. Right? And they can't professionally do it because there's no friction involved. Right? There's nothing that, you know, so I think it's just gonna be a power law for everything. And it's not, Dave, to your point, everyone will write software, but that's just not a job. It's just like what you do. So here's the real question then.
Like, why aren't we seeing a Hollywood writers strike from engineers in Silicon Valley? Well, this is what I want to say. And I want to move us towards the conclusion here, but... Yeah. And I need to go because I have to do actual work. No, Sam, we have a, like a parent teacher conference thing. So you can't do actual work. No, that's in an hour. I have two, I have an hour to do these other two meetings. Okay. Um, but... It wouldn't be a more or less episode without Sam dunking on this not being actual work. It's not work, but it's fun. Again, this is actually, by the way, by the way, okay, I literally do have to do go, but I will say this. This is exactly the point.
We're doing this podcast for no money because it's fun. Right? There are millions of people doing fun podcasts. You think that's the answer as to why we don't have an engineer strike? There's going to be more podcasts. We don't need that many more podcasts in our life. Everyone's going to have a podcast and no one's going to make money on them except for like one person. My legs are just so tired. I can't even talk. We're doing this podcast because we have other ways to make for no money because we make money other ways. First point. Second point. I think as you're describing, what's really clear to me is that the incentives between the
leaders of companies and founders and their teams more broadly are about to take a big divergence. And I think if not already. Yeah, no, they are. I mean, that future is here. It's just not evenly distributed yet. Right? Correct. And I think that is going to have all sorts of second, third, fourth orders effects for sure that are more to that. And, and I sort of have been joking about this with my team a little bit. And I think it's kind of funny that founders like myself and other founders now think we have all these superpowers cause like we can build these things that, you know, are imperfect,
but all of a sudden like we can leverage ourselves more and get our hands in more places. Like I imagine that most teams aren't thrilled with that happening at the same time. I think it is important experimentation that needs to happen from the top to identify new opportunities for these companies. But I think that that split is going to become very, very real. And then the question will become, are the winners going to be the net new teams that can do it with the free three people from the get go versus ones that are going to have to really weather and manage disruption from their teams. Some of which are very, very sizable on this. To whittle down the team, you mean?
I mean, we might, we might have some Washington Post era style labels. Yeah, I agree. Or worse. Kind of across the board, but there's, but there's always more to that calculus. But I think that's a very, very important thing to spotlight in the future. Yeah, Jess, I think you're right. And that's the other topic we've heard a lot this week is like, is this SaaS sell-off? Yes. Is it like... People are obsessed with the SaaS sell-off. SaaSpocalypse is what they're calling it. And I get it. I get why, but like, it was so last...
I just think what you just said is right, Jess, which is that I think you're going to see like, whatever, some percentage, call it 50% of the SaaS startups that are still innovators that are still building great stuff, they're going to be fine. But then the bottom half that hasn't improved their software in, you know, 10 years, I think of the companies like MindBody and these kinds of things where like the software is just still just... They just got rolled up into a PE firm. So I'm sure that's going to go all of those plays are going to get eaten by the three person teams that you're talking about, right?
Like they're going to come at like, like a freight train using these new engineering methods. And they're going to be able to eat that part of the SaaS market. And there's still going to be a great SaaS market. It's just that some are not going to play very well in what's happening due to this new agentic engineering method. Okay, let's bring it. It's been a pop culture. I mean, we had Super Bowl, you know, what? Anything catching? Also, we, one of the fun things we do at this event is we get great recommendations from, from the crowd on, on media too. Yeah.
I mean, some of the, yeah, we, we had a whole panel about what's happening in pop culture and the highlights were one heated rivalry is the show that all single women must watch if you haven't yet. And I mean, he was, he was contending with the everyone needs to watch this. He claimed it was the most important pop culture thing of the year. I mean, I do think culturally it's relevant for men to at least watch one episode to know why this is intriguing women. Maybe like the fourth? Yeah. Wow, you're four episodes in on this thing. She's done, Dave. Four episodes in! I binged this like a month ago. Yeah.
Thank God Dave got COVID a couple of weeks ago because I was able to start binging. I'll try it. And then secondly, Nancy Guthrie, you know, we're all tracking his case. It's America's crime story. And what I think is so interesting now, people have started using AI to take the photos that the FBI has put out of this guy on the front porch, you know, trying like trying to cover the Nest Cam. And they've started like using AI to determine what he might look like under the mask. They, another person used AI to, to lighten the photo, to see the car more clearly behind him in the driveway.
Another person used AI because there was a different trespassing report in January of a different guy. And there was a photo of him and they used it to like match the same shoes to this intruder. So, um, it's like all the people are part of this investigation and leveraging AI tools to actually solve the crime. And I think it's really interesting for that reason. The decentralized use of AI. It's just so sad and tragic. All the claw bots are like finding Nancy Guthrie. Wait, what about the book from New York? But they aren't. But maybe they will. Strangers. So us West Coast people. Yeah, the book of the New York book.
I feel like all our non-Bay Area listeners have already read this book, Strangers, a sort of memoir about a wife whose husband very suddenly tells her he wants nothing to do with her or her family. After, I swear, like 20 people recommending this and even people who didn't recommending it know the intimate plot details from their wives. I read like 10 pages last night. So, so far I'm hooked. Okay. I need to get on it. Okay. Well, to the dear listeners, thank you for once again enduring an open claw deep dive, but I promise you it is important. And, um, for catching up with us. So we're grateful for your listening, your followership.
Don't forget to subscribe and we'll see you back here soon for another episode of More or Less. Bye. Bye guys. See you later. If you enjoyed this show, please leave us a virtual high five by rating it and reviewing it on Apple Podcasts, Spotify, YouTube, or wherever you get your podcasts. Find more information about each episode in the show notes and follow us on social media by searching for at more or less at Dave Morin at lesson at Jay lesson. And as for me, I'm at Brit. See you guys next time. Bye. Bye. Bye. Bye.