Ty Pattison: Hello, hello. Mr. River. River Roberts: Hi Patterson, how's it going? Ty Pattison: guys, good. Goes very well. I have my ring light ready and I've got news. I've been speed running the news this morning and all there's some juiciness. River Roberts: Alright mate, well let's drop straight in. Ty Pattison: Let's do it. So couple of pieces that jumped out that I thought were kind of interesting. I've I've been thinking a lot more about applicability of AI. and River Roberts: Hmm. Ty Pattison: one of the things that jumped out was anthropic Building a digital watermark into all of their content and they're scheduled to release it next week, I believe. and what this looks like is like in text outputs, there is a digital signature that is not metadata that is supposed to survive being copied and pasted and also survive light editing. so it means that if like a research paper is done by an anthropic model, it will have this embedded and Anthropic watermark, so it can be attributed to anthropic. yeah. It's pretty interesting. I don't necessarily believe that it's because they want to claim ownership of all of this content, but that there is some legislation coming in that is about AI accountability and AI traceability, like being able to say whether or not something's AI generated. So I think this is their step in that direction. River Roberts: Yeah, well, you know, certainly off the back of last week's news with the EU, act and all of those policies in place. So yeah, we're starting to see some of the first moves. So Spotify has also been one of the first actors to kind of promote or have like badges on the songs. And they've actually started to take AI content out of the recommendations on the playlist, which I thought was quite an interesting decision as well. Ty Pattison: interesting. Becoming or like leaning more into the the yeah creator nature of content or understanding who the creator is. River Roberts: Yeah, but also just making some pretty big moves towards what is quality content and what is potentially not so quality. But that being said, I've been listening to some of the AI music that's been generated over the past months and some of it is just incredible. But again, there isn't a clear watermark and I'm not sure I've... How would you signify that it's AI generated? from an audio file other than listing it on each individual track title. Which if, for instance, if you're listening to it on radio, unless the announcer pops it up or there's this like really, know, notification sound, it kind of, yeah, I can see how these are good initial moves, but man, it feels like such a strange world that we might be heading into if, you know, 80 % of the content you're consuming is. AI generated and then every time you're faced with it, you have to be notified of it. yeah. Ty Pattison: Yeah, well that's the thing. I don't think that it happens in post. Yeah, I'm just fucking around with my my ring light. I don't think that it necess necessarily happens in real time. Like I think it happens if you ask. and I think that's probably what will be the case for a lot of content. It's like, if I'm listening to a song I don't care really if it's AI generated, but I want to be able to choose or I wanna know. I think that's kinda where the distinction is. It's like if if To use a weird example, if I'm texting you and I know that it's you and we're sending like sexy text messages back and forth, then I'm like fully aware of what I'm engaging with, right? But if I'm texting with you and you're an anonymous person and you could be like an eighty-five year old dude that I do not want to be sending saucy text messages River Roberts: you Ty Pattison: with, then if you were to mislabel yourself as you I would feel really uncomfortable about that, but if it was labelled as like I know exactly who I'm talking with, then it it yeah, it surfaces some of these things. So I think it's like removing asymmetry of information is the important part. River Roberts: Yeah, look, provenance is a really... Ty Pattison: Sort of. River Roberts: important issue for a lot of aspects of society and historically we haven't done it particularly well and it gets much harder at scale. So like one of the analogs of this is food. So understanding where food comes from. And so obviously you you can have the experience where you go and physically yourself go and you pick the something from nature, something from a tree. You could also then go to a local market and trust that the local farmer or the farmer's market as this kind of environment of, of truss around it. But then as soon as you go to a supermarket, that, that level of provenance is, is really obscured. And, you know, there's obviously certain labels and certifications that we've grown into over the years, but still it's a, it's it's a really hard thing to consistently be accurate about. But yeah, I think the provenance of. Ty Pattison: It's a good point to River Roberts: of AI output is a really important thing and also just like a really hard thing to keep track of. It actually kind of leads a little bit into one of the pieces of news that really jumped out to me this weekend. It was the announcement of auto mode. Did you come across auto mode? Ty Pattison: I'm aware I've ever I've seen it. Can you elaborate? River Roberts: Yeah, so automotive's anthropics moved to instead of seeking permission for certain actions, the model itself Ty Pattison: Yeah. River Roberts: is starting to do it. Because actually what they found was, I think, what were the figures? was... Yeah, anthropics system caught 80, 89 % of what they were called like dangerous commands and then the permissions and the prompt site on moving through those. then humans caught about 13, just about 13 % of them. So the insight here is really how much can humans actually accurately provide oversight. for different approvals as these models get more and more capable and shifting that control from individual to more of the system. And then the security thresholds are actually more compounded in auditing the system rather than each permission, which I certainly had early experience with in some of the Hermes harness pieces because it asks for a lot of permissions. Honestly, like the process of, of getting asked permission for access to a certain thing. It's, it's, it's hard to know exactly what it's asking access for in a lot of instances. And then by the time it, Ty Pattison: Yeah. River Roberts: by the time you research what it's trying to ask for it, it, makes things, you know, it exponentially increases the complexity and the time taken to actually get things done. And you get to a point, or at least I certainly got to a point where you're just like, all right, let's go just. you know, I approve everything, go for it all, which is like that decision is coming more out of frustration and confusion rather than, you know, deep, deep insight. And that seems to be pretty similar to a lot of people, as that particularly is that vibe coding, you know, a of this industry has gotten in on, on, on good vibes. And if you're vibing at one in the morning, like you're just going to, of course you're to push approve, you know. Ty Pattison: Yeah, well and it doesn't it it's like we've got these tools to go and climb a mountain and these things are really low stakes, but like you're up there on the mountain and it's like, Hey, would you like to use a pickaxe or a shovel? and you're like, Well, I don't fucking know. You are the one that is the expert here, so I'm just gonna go with whatever it is that you suggest. Like t we're like out in terrain that we're not equipped for or that we're not educated for and then our guide is this thing that is asking us for decisions and we're like, well this is what I think. Like luckily a lot of this is really low stakes. But if you are building it to me says that if we're building like security systems, you should know what you're doing before you go and try and build that. There are certain domains that you can really vibe code in there are certain domains that you should probably leave to people who know what permission is being asked for. Because I've had this conversation As well, back and forth with my agents quite a lot, like asking for permission to do something, or like explicitly me saying, Hey, it's really annoying that you ask for permission to do all of these things. I actually don't know the answer to most of this. Can you just stop asking for permission and do it? So the the auto mode that you're talking about? River Roberts: So you already gave it auto-mode. Yeah. Ty Pattison: absolutely. What what this news article is talking about is that they're setting auto mode as the default. River Roberts: default, yeah. Ty Pattison: Yeah, yeah, which is which is it's cool. It to me speaks to the anthropic have enough trust in their system or at least they're seeing that it's a better alternative than than letting humans do it and especially than letting humans who don't know what they're doing do it. River Roberts: Yeah, yeah. So definitely as a default piece, it definitely feels like a good step. So it'd be interesting to see how many other kind of providers start to move in that similar direction as well. But then still allow users to kind of audit the system in a meaningful way. You know, it's a little bit like, again, an analog would be like the social media age where you just sign on to these platforms without really knowing. what's happening with the content and how money's being made. And then over the years, like over the decades, you you start to get more ownership of, you can control this. You can turn this on, you can turn this off, but you know, a little bit too little, a little bit too late. So hopefully, you know, this industry can be a little bit more on the front foot rather than the back foot. And that's what this, this move tends to suggest. Any other news pieces from you? Ty Pattison: Yeah, I had a couple. so Cloudflare have this so Cloudflare the Flare, the big cybersecurity company, have this agent first browser that they've brought out or that I think has been out for a while, but it's called Kitesurf and it's built with Rust and WASM, which Rust is a a relatively low level language. and it's interesting because it uses a fraction of the compute of Chrome. and based on May data, in May, agent activity on the internet surpassed human activity on the internet. And this news article was basically saying like, all right, in May this happened and now we've got companies that are producing agent first browsers that use a lot less compute. Human activity on the internet is gonna be a rounding error. Like in the coming years. we're just like most of the network activity is not gonna be done by humans, it's gonna be done by intelligent or semi intelligent agents. all going about their business to to do things. I think what it says is again what we're talking about with this permissions thing is or what's jumping out at me is we need human readable intention kind of gates rather than technical gates. Like instead of, can I have permission to do this thing that I would need to research for a long time to figure out what it means it could be a This is the goal, these are the kind of obstacles that I ran into. What do you think the best course of action is based on River Roberts: Hmm. Ty Pattison: these criteria or like to have some way to interface with agents that is not purely technical, that is more intent based, would be a really useful step going forward and whether or not that is enough or whether or not like that that is a real space. I don't quite know what that looks like, but I do think it's an important interface that we'll have with agents because Yeah, right now we do like explicit steps on screens. We click buttons, we fill out forms, we put in all the details. I don't think that's necessarily going to be the case. I think it'll be more like, my intention was to apply for a visa for India and so I did that and my agent did all of the details. Yeah. River Roberts: Hmm. Man, huge implications there. Yeah, more agent infrastructure emerging again. Ty Pattison: Yeah, exactly. how about you? Any any other news articles popping out? River Roberts: Yeah, it's actually, so there was a comment this week by Elon Musk to a tweet or a comment on X and it was FEL FTW. And FEL in this case stands for actually, let me, let me back up. So basically, you know, this past week, there were some visual representations of what the TerraFab would look like. And they've confirmed that it's home base will just be outside of Houston, Texas. And this thing will be the largest structure on Earth, the largest built structure on Earth. And the images, if you haven't seen, are quite remarkable. But a few people noticed that at the center of this structure was this kind of circle structure. And there were a few comments asking if that was actually a free electron laser that he was developing there. And he commented back, FEL FTW for the wind. So free electric laser for the wind. And so basically, you know, this is actually quite significant use and the fact that at the moment, the way chips are produced in TSMC and what's the European one, ASMR, they use a different technology and basically free electron lasers have the ability to be six times as efficient. But the reason that they have kind of stayed more in these scientific theoretical spaces is the space that's needed and the cost to develop it is way too high. So the TerraFab is the first one that can actually put it into production. then, you know, on the back of that, he's also made the claim that 80 % of the output of TerraFab will actually be going towards orbital compute. So these two things converging, so his move towards really owning the full stack of physical hardware and chip development and then setting them up into space and then running the communications network and basically the Dyson swarm in space, it's just phenomenal. It's just such an incredible vision, but his ability to compile all of the different parts of his empire is really starting to see some convergence here. His ability to understand batteries, his ability to understand rockets, and as he's building the Tarifab, they had an earnings call last week as well. And he was basically saying that chips factories, it's not like building rockets. It's not rocket science here. So the fact that he He's building rockets. He's actually looking at chip manufacturing as a much simpler problem to solve. So, man, it's going to be really fascinating what this Terra Fab eventuates in the next couple of years. Yeah. Ty Pattison: The vertical integration of these pieces is pretty insane. River Roberts: Yeah, the rockets, the satellites, the chips, the light source and the AI, like all of these things become just incredible, but also like potentially a single source of failure as well. So if like one aspect of that system somehow gets corrupted, like there's no reason that all couldn't in a way. So, you know, does this full stack bring resiliency or does it actually have some brittle aspects to it as well? Time will tell on that one. again, just the scale of the vision is incredible. there's the saying that don't bet against Elon. And it seems to be a pretty sage advice on this. Yeah. Ty Pattison: Interesting. It's yeah. I I mean there are a lot of assumptions, right, that a lot of these things will work. And also there is some track record that says that n like a lot of this is in place in not concrete way but but that there is there is precedent here for it to work. I'm interested to see how it all unfolds. and also you need people that are River Roberts: Yeah, totally. And then that's not even including his robotics arm as well. He shifted a lot of his Tesla production over to Optimus production. So he's really going full step on the humanoid robot piece as well. And it seems to be much more focused initially on manufacturing robots, so robots for manufacturing lines, which then increases capacity to build all manner of other things, whether it's a satellite, a rocket, a car. Yeah. forever. Ty Pattison: It reminds me of these game mechanics where like I'm interested in board game design and these like economy building games where you win by having the most resources but you can put like Katana is a good example of this. You can like spend resources on resource production assets or you can spend resources on gaining points and River Roberts: Hmm. Ty Pattison: the strategy of creating more resource production. is often or least in the long run is more beneficial. and yeah, creating the capacity to build more is a long game and also something that that yeah, Elon seems to be in for. I have many opinions on this, but maybe we can move on from to to next topic, which is River Roberts: Well, I have actually one more piece of news which, Ty Pattison: Piece of news. River Roberts: actually, yeah, sorry. So I got one more piece of news that really jumped out to me this week, which I actually haven't had time to go into deeply, but it does draw another line in the sand, which I think is compelling. But Mark Zuckerberg released a manifesto on super intelligence published Basically, the thesis of it is distribute superintelligence to everyone. And then off the back of that, he meta-shipped a one billion parameter local agent model that runs on a single GPU. And so Ty Pattison: Wow. River Roberts: backing up what they were previously exploring with Olamma, and they've kind of been quite quiet since. I mean, they'd really spark recently, but... so this is Meta's Muse glimmer. And again, I haven't played with it just yet, but again, it's, it's quite indicative of which way they're headed. And, you know, one of the caveats on this is that, the white house has excluded open weight models from voluntary safety testing. So while it's got these front-end labs like Anthropic and OpenAI that, are going more closed and they're kind of beholden to the white houses, security. testing. These open weight models aren't at all. Ty Pattison: Mm-hmm. River Roberts: So like that tension there, I think we're going to see some breaking points in the months ahead for sure. Ty Pattison: And just so I'm clear the intention of the White House's like voluntary submission process, River Roberts: Safety testing. Yeah. Ty Pattison: safety testing, is that they want to be able to screen models before they go out. But the open weight ones and I'm asking this as a question 'cause I don't quite understand. And the open weight ones don't need to be screened because they are open weight. Is that right? River Roberts: Yeah, basically, So the White House is inviting Frontier Labs, and they're singled out a handful in the US, obviously. And they've said, look, Ty Pattison: Yeah. River Roberts: this is voluntary safety testing before you release this to the public. And the idea is that they go through the safety testing. They are allowed to share these models with kind of vetted invented industry members initially to see for passes and then they can open it up to the public. Whereas, yeah, open weight models, at this point, once they're released, they're released. and then there's a difference between kind of open weight and then open model. so, but we're seeing a lot of the open weight models then opening up to open source as well. So, yeah, so that you can run them locally on prem. You could you could then like do whatever you want with them. yeah, fascinating landscape. It's crazy. Ty Pattison: all right, I got I got I got one more piece of news here. so Stanford and Ark Institute have a model called Evo Two. and last week it I believe this happened last week. This is a very recent piece of news. it designed and completed working viral genomes. So it created three hundred bacteriophage designs and sixteen of those became functional viruses capable of killing E. coli. So phage is something that eats. And that jumped out at me because like these wouldn't infect humans, but they would be interesting for like antibiotic for killing bacteria that is resistant to antibiotics. and it also jumps out as being like a massive biosecurity potential risk. Just like this is what all of these models that like Fable is known River Roberts: . you Ty Pattison: for essentially downgrading people or saying, Hey, you can't talk about this topic if it has anything to do with biology. Or if it has anything to do with potential biological warfare, and this seems like it plays in that realm. and it's the first thing that I've read about. I know that it's around, but it is the first thing that I've read about that i i it has produced working biology specific kind of information. I thought that that was a pretty it like touches on this how does AI interface with the real world and like what is it actually producing or how is it actually kind of changing the our lived experience in River Roberts: you Ty Pattison: a way that is not just software? River Roberts: Yeah, I it's like, think that is actually one of the biggest threats of our more immediate future. In fact, Annie Jacobson, who recently wrote a book on nuclear war, has actually recently published one on bio warfare as well. And although nuclear war is ranked as more deadly, obviously, because, you know, it is kind of a total annihilation, the bio warfare is just underneath, actually more probable because of the accessibility of the ability to create and synthesize insights like what we're kind of seeing. So yeah, I mean, this is a sign that that's starting to become more accessible and potentially more public. Yeah, man, crazy. Ty Pattison: Well, it's also it's a risk and it's worth talking about this because like we're humans and and we do risk management kind of implicitly, but I also think it's important to talk about the flip side of it, which is like how incredible the technology is and how it could open up kind fields for for tapping into our our biology in in ways that we haven't imagined or that we haven't properly explored. Like it seems like the first things that we touch on, hopefully, or often, are like the risk mitigation factors and then the things that we can really like expand into and the stories that we need to be telling are like what is the optimistic future here and in what directions could we River Roberts: Hmm. Ty Pattison: Could we eat, could we explore, could we research that are not to do with survival, that are to to do with thrival? River Roberts: Thrival kind of sounds a little bit like an STD, but I like what Ty Pattison: Ha ha ha. Yes please. River Roberts: you did there. Something that we haven't been particularly good at as humans and potentially, this is maybe what AI can support us in, is third and fourth order consequences of different things. So developing these different bacterials or these different you know, genetic pathways or, you know, look at some of the progress that's been made on the longevity front as well. The third and fourth order consequences of that for 8 billion people worldwide is huge. So having some support to manage that, I think goes a long way. And yeah, again, the upside of this is there's a lot more people working on the positive and the security side of this than those that are working on the harmful kind of warfare side of it. So it's still a very unbalanced terrain towards the positive. However, the tools do seem to be coming quite widely available and then it doesn't necessarily take a lot for something to be released and then it's kind of hard to contain. And having some Ty Pattison: Yeah. River Roberts: experience with pandemics and epidemics, not only going through SARS obviously, but I did work at a biotech firm in Hong Kong after SARS. So I did learn quite a lot about pandemics and epidemics in particular. And the ability to control them quickly is one of the most important things. So, you know, the ability to respond within 24 hours of a release and kind of contain it in a meaningful way goes a long way. One of the things that makes that really hard is if that information is obscured in a way. like, for instance, it was kind of months before we actually really understood what was going on. At that point, you you couldn't contain it. was, was kind of global. So, yeah, Ty Pattison: Yeah. River Roberts: I'd say the other thing that's kind of working against us is global governments and medical professions worked so poorly at the last pandemic. think the next one is going to be faced with some pretty tense cultural and social pushback as well. So the compliance might not be as good, so the outcome might actually end up being worse. yeah, but anyway, let's not go into doom and gloom now. That's not part of one of the sections. Moving towards the abundant future. Ty Pattison: with my dad when COVID first hit that it was like maybe February and he was like, yeah, this will be over in a month and I was like, Mm, I don't think so. I or like it was early March. I was like, dude, the stuff that I'm reading is is not that this is gonna blow up. There's also that defense is easier than sorry, offense is easier than defense. You can River Roberts: Yeah. Ty Pattison: like in in security and biosecurity and cyber security. So there's this like interesting interplay where like you if you're on defence you have to figure out what all the attack vectors are. If you're on offense you only really need one attack vector that works. Yeah. River Roberts: Yeah. And even if you're an offense, the risk of something not going quite right and like, you know, even though you're offense, I mean, this is the whole gain of function testing, right? You could be working on this and then it kind of leaks and you know, whether or not you were, I mean, you were working on defense, but you know, errors still happen. So Ty Pattison: So River Roberts: yeah. So I actually have a bit of Longview Ty Pattison: Yeah. River Roberts: news today. I was browsing longview.org and it's a very special day today, August the 12th. There is a solar eclipse over Europe and there's a New York Times article on the Longview and the original title read, live updates in Europe, the total solar eclipse is only hours away. And then the rewrite of the title is solar eclipse over Europe. continues ancient pattern of cosmic spectacle that once terrified, but now delights. And I kind of love, I love, Ty Pattison: Ha ha ha. River Roberts: love, I love that. I love that idea that, you know, once upon a time, a solar eclipse was one of those things that was probably one of the most terrifying things to account. So, cause you're like, what is happening? Like this is the end of days type thing. And then you could just imagine all of the, the religious sermons on how the, you know, the things that turning. So, our ability to be terrified by one thing and then delighted by it in another age. I'd like to hope that these things that we were even just talking about moments ago end up being those things that we somehow end up getting delighted by in the future. another pandemic, how delightful. Ty Pattison: More information makes us more delighted. More information makes us safer. More information makes us more equipped. more information is applicable. As long as the information is transformed into usable knowledge. Or as long as the Yeah, I don't quite know what that pipeline is. Information becomes knowledge, becomes wisdom, as it gets distilled and then it becomes applied, it makes us more equipped to to experience the world in a way that works for us. hell yeah. I l I like that that reframe. very special day. It's also Wednesday. River Roberts: Yeah, huh? Ty Pattison: That's it. I I mean Wednesdays are just good days. so so I that's me for news. How are you anything else jumping out? River Roberts: No, that's it for me. I'd love to do the meme view again, although without our ability to display images and my relatively poor ability to accurately depict images via voice makes the meme view not a particularly exciting segment for now. So let's leave it at for the moment, but there is definitely something in that. Should we move on to building? Let's move on to building then, shall we? Ty Pattison: Let's look at building. yes, definitely. You wanna kick us in? River Roberts: Yeah, so I have been exploring the possibility of entering a submission into the Future Vision Express, which is a three minute video teaser for a movie based on the future. And basically the idea is there isn't enough positive compelling visions of the future. And so this XPRIZE has created an invitation for thousands of people. And I think there's over 5,000 entries so far to put together these three minute trailers. And my thinking when I first heard it months ago was I'm going to wait until the week before this thing closes and then see what tools are available and see what I can produce. so I've kind of, that's exactly what I've done. And I'm about halfway through the process and I'm finding that there's so many different video tools out there to use. But actually what I ended up working with, at least in this instance, was a system called DaVinci, which aggregates a bunch of the different models. So you can do video, you can do image, you can do all the different models across them. And there seems to be some kind of consistency across, but I have to admit. I'm outside of my wheelhouse here. I haven't played too much with deep image generation or video generation yet, but I'm looking forward to at least being able to share what I've created next week. Ty Pattison: Cool. Hell yeah. do you have any ideas for what you want this trailer to to look like? What story you want to tell? River Roberts: I I do. So the idea or the vision of the future is, and again, like I've, only actually only recently started watching Star Trek. I actually got introduced to it during the pandemic. And through that process, I kind of watched a lot of it. And I was surprised that how, you know, I had an immersion to it growing up. I'm like, nah, that's for dorks and nerds and that's not me. But then actually watching the show, it's one of the few shows that really positions the characters around this idea of exploration. Not everything's about conflict. Star Wars is very much about conflict and conflict in relationships, conflict in personal story, conflict in the global story. Whereas Star Trek is much more about exploration and I fundamentally believe that humans are very curious by nature. kind of maybe a precursor to something like Star Trek is I see an ocean vessel that I'm calling Kinship One. And the idea is basically, it's a ship that invites a bunch of different people that have figured out an aspect of the future and invites them onto one ship together to help evenly distribute it. So this idea that the future is already here, it's just not evenly distributed. And then so what would a vessel, what would a seafaring vessel that can travel to different corners of the world? and invite in different people who have figured out an aspect of the future, but then also create a space both on the ship and then bring things on shore to showcase what's actually possible. those, these would be things that, you know, are very physically technical. So it could be like bioengineering, it could be bioluminescence as lighting systems, but it could also be technologies around different social structures and cultural structures as well. Different. mediums of of building different materials, there's a whole bunch of incredible emerging technology. And obviously that's only going to be accelerated through AI. then so the Kinship One, everywhere it goes, it would help facilitate partnerships and help take this kind of open source landscape that works really well digitally and apply it more physically. So the prime directive of Kinship One is really to open source ideas. and to evenly distribute them in the physical world. So that's the broad piece of it. Ty Pattison: I really love that vision. the kinship one, does it have like is there a minimum viable crew for the kinship one? Like if you were two people, maybe that's too small or is it like what's the you know what's the minimum River Roberts: Well, so like... Ty Pattison: crew, what's the maximum in your mind right now? And this can obviously change. River Roberts: Yeah. So like in my mind, this, kind of works around 150 people. So there's actually, there's a boat that there's a boat called the world. And basically someone retrofitted an old cruise line out that had like 3000 beds and converted it into a full-time residency. was like nice two, three bedroom spaces for 150 people. And so the world travels around the world. But they don't travel all the time. kind of, they'll do like three months in Asia and then they'll go three months East coast of the States and then they'll go three months in Africa. And so they don't actually move that much. When they do, they, they make a, they, they, take much longer in each place. and then everywhere they go, they invite people on, basically the world is, full of high net worth individuals. and I thought how interesting would it, how much more interesting would it be is instead of just high net worth individuals, it was. you know, people that have figured out the future. And like, I'm also just super interested to see what would happen if you brought these people together under one roof in a vessel. And as they traveled, what else emerges from this space? And for me, it's kind of this initial spirit of what the Starship Enterprise is about, but just on a world scale, on our scale. Like I'm so fascinated to see how a ship like this might inform a different township or a smaller city and what infrastructure, what insights, what partnerships, what relationships could be nurtured in a relatively short period of time, let's say like three months in each place or in each port and leave behind a significant difference. But yeah, I think around 150 buck is a good place to start, but potentially could scale up for sure. Ty Pattison: Love that vision and all the different paths taken to get there. cool. So building, building, building. I have been essentially doing research. and I'm really interested in this space that is the the wealth transfer and the knowledge transfer and some of the pieces of data that jumped out at me recently is that There's about sixty billion dollars worth of small businesses that's gonna change hands over the next twenty years in the US. These are American stats, because America has a lot of stats. The the United States has has a lot of statistics. That trend is true across the world. in Japan we're seeing this happen and it's impacting their economy. in the UK this is happening, it's happening in Australia, it's happening all over the world. Where these baby boomers have businesses that that have customers that are loyal. There are a lot of people that just don't want to change the technology or the businesses that they use because they're ingrained, because they use them, because it's part of their day to day life. But a lot of these small businesses, it's something like seventy percent of the small businesses that go on the market never sell and just close. and the online research that I've done says that this is yeah. I it's like it's such a shame. And yeah, the online research I've done says that the main reasons are that there are there's operational complexities. Sometimes there's like key man risk. like in startup land you can have key man risk insurance as in you would insure against the founder getting sick or dropping out or doing something. So there's this element of key man risk. and so I was thinking about this and I was thinking about different business sectors and kind of like where I would like to apply myself and decided to go and do a bunch of primary research, just walk into businesses and talk to owners, talk to staff, talk to people and get that intellectual information and also get the information of like what it feels like to be in different types of businesses. like to be in stores, say like walking into vintage stores and going, what does it feel like to be in this in this shop environment? Like how do the staff interact? What is the experience of this and kind of passing out the different elements of this? That like there are some really human elements to businesses and there are some really there are some elements to some businesses that could be so digital. River Roberts: group. Ty Pattison: and Some businesses are really complex and some are really simple. Like, I walked into an art gallery and yes the staff have things to do and yes the there's, you know, things that could be that could have processes behind them. And that business is so much simpler than I walked into a coffee roasters and it's bustling and there's stuff going on and there's all this like food is is being made and then it has expiries and then and like everything needs to be clean and it needs to be up to standard. Like there's just such a vast difference between those two types of businesses. and Yeah, some of the learnings from like this initial stage of thinking about how to how to serve this market potentially is a lot of the business owners were not there, which is kinda clear and obvious, but not unless you it wasn't clear and obvious to me until I started talking to these different businesses. and then I was thinking about kind of the stack that you could build based on these problems, which is that business owners and like business owners don't have succession plans. So I think there's this big opportunity there for helping with succession planning and and connecting people. And it doesn't necessarily look like a marketplace. Like River Roberts: Hmm. Ty Pattison: maybe it's a different piece of content. I was having a conversation with one guy who was a part owner in a vintage store and he started talking about the history of fashion and how a size medium in the forties was different to a size medium in the fifties and sixties and the seventies, and how like people started having different diets, like eating more protein, which made them taller, which meant that like the garments change over decades. And it made me realize that he was telling stories about his small business and that there is this distinction between like Building a product that is devoid of story and building something that is really simple that taps into like human nature, which is like you could have River Roberts: Mm. Ty Pattison: a product that's a marketplace for small businesses being sold, or you could have a product that is a where you interview small business owners and get them to tell the story of of how they started the business and what the vision is for it and kind of what the ethos is. And that in itself is Pretty low lift technically, but different angle for the same kind of problem solving. and then it's not always technology, like yes, we have AI that can go and build a bunch of stuff that we couldn't before. Amazing. It will help in a lot of ways. It's not always the best solution, and just because you can do it with these tools doesn't always mean that like it's the best tool to apply to it. And you kind of have to there's this analogy of like, when you're a hammer everything looks like a nail. You kinda have to pretend that you're a hammer or like use a hammer a lot to start going, that nail is different. Like that nail actually is a screw and it would probably be better off having a different tool applied to it. Like it's the difference between like vibe coding where you don't have a lot of experience doing it and you just like apply the same methodologies to produce something and then having the ability to discern which tools you want to use, like becoming a builder and saying, no, I I have this way of approaching this kind of scenario. So the tools that I've been leaning into recently. River Roberts: What kind of strikes me is, I was just going to say what really strikes me is that, Ty Pattison: Yeah. River Roberts: you know, in our current culture, we celebrate the startup culture so much. And we've kind of been educated into these two pathways, right? Either you start your own thing or you work for a massive company. And then this like, what is it like 70, 80 % of every economy is these small medium businesses that are more on the. at local retail level. And there's such a, there's such a small number of people focusing on actually, what does the takeover look like? What does the transition look like? And how do you celebrate that as a pathway? And I think that's actually a much healthier pathway for a lot more people because, you know, the rates of burnout in entrepreneurs, it's really high and the lifestyle does not typically help cultivate. you know, healthy long-term relationships and stable marriages and things like that. Whereas the takeover culture certainly could be, and maybe it's not called takeover, maybe it's something else, but that seems to be like such an interesting space. So it's incredible that you're exploring this and you're not only just exploring it in terms of like, let me write a prompt and do the research. Like, let me actually go and do the primary research and meet the people and have some of the actual conversations. And then you get to feel. the essence of it. you know, I've over the years interviewed a lot of different stakeholders and small different businesses to better understand what was going on. And I can say you just get so much more of the context of what you're working with if you actually talk to people specifically versus like, you know, the market size is this the total addressable market is this it's, you know, 70 % of people like, it's great high level, but actually, I think a lot of people with the vibe, with the vibe, building and barcoding, they're missing that. Let's go out in person and actually talk to some people about this and really connect it to the physical world in a way and localize it. so what, Ty Pattison: Yeah, yeah. River Roberts: what do you think? Like, so, so it seems like there's two kinds of parts to this, right? So one is very much like creating new pathways or celebrating new pathways around taking over an existing business and letting something else emerge from it. And then the tools for that actual transition once you've taken it over and how to improve it. And I know a lot of hedge company, hedge funds and &A firms are really focusing on the larger end of the scale and buying up a lot of what they're kind of seeing as underutilized resources that already have customers and then optimizing it with different AI tools. But I haven't heard of anyone really doing that on a smaller scale with, yeah, like a vintage drawer Ty Pattison: Yeah. River Roberts: and an art gallery and a hairdressers and a nail salon. Ty Pattison: Yeah. What really stood out is that the businesses that I thought were most interesting were the most nuanced. So it was this vintage store where this guy was telling the story. there was a bookstore where it was not really repeatable, but it is kind of one of the more interesting businesses. And then the ones that I didn't think were super interesting but were probably more that were better acquisition targets for a Company or were the ones that were a bit more sterile. They didn't have as much character, but they had quite a lot of documentation. They had like processes for talking to the owners, rather than just the owner being there and all living with the person. what I realized though is there was one business that I was asking the employee, I was just saying, like, basically I'm just talking to business owners, figuring out what their succession plans are, to to try and understand this deeper for myself. And he was saying that in this particular store they did quite a lot of consignment and that there were people that would bring in records and different kind of bits and pieces. So there was a community around this particular store and that the succession planning for that was probably based on that community, like somebody who was bringing a lot of things to the store would probably take over at some stage. and that community aspect is something that seems to be missing a lot of the time. Like you have this trade-off between it being not scalable and very personal. and then the like fulfills the specific need of the customer. I need coffee. You give me coffee. Cool. But it doesn't have a lot of like it's almost like the friction was part of the appeal for these businesses. River Roberts: So that was actually one of the exit pathways that blockchain and crypto was actually promising this idea of fractional ownership. And when you can fractionally own something like a vintage store, you could realistically exit to community. However, I haven't really checked it lately how that's actually evolved. But I was exploring this pre-pandemic and this idea that instead of you investing in one property or one house that you own, and putting in 100 % equity in, what if you put in half a percent or a percent equity in 100 stores that you spent time in? And then that was actually the transition period. Well, that was the succession plan actually. So you exit to community. And then as part Ty Pattison: Mm-hmm. Mm-hmm. River Roberts: of the community, people are hired to do the work. Yeah, but there seems to have been some things that haven't quite played out. Ty Pattison: There's this piece that that I sh River Roberts: Yeah. Ty Pattison: Well, the the friction that I see is that ownership and operations are very, very linked in the world because to to put your identity in something in ownership, you want it to run really well. So you so you make sure that it operates really well. There doesn't seem to be enough buy in from a person who owns one percent of something to make sure that the operations are really smooth. and maybe we can decouple those things and it can still work, but like in the world that we live in today, unless those things are tightly coupled then like business owners talk all the time about like when if something is not quite i if something falls over, it's the business owner that has to go and do it. And in fact this guy that I was talking to, Marco, he said that somebody came in one day with a C V and wanted to get a job and they were talking to the person in the store and they were like, Hey, is the manager here or is the owner here? And he was saying that like they didn't quite understand, but the guy that was out front sweeping the front of the like sweeping the door was the owner. He was doing the little things that only the owner would would really care about, which I thought was really Really cool and interesting. yeah. And then speaking of building, because I've been in this like space of ideating recently. and one thing that I'm not building right now, but it's just percolating, and I want to I wanna speak to it for a second, is that In the past people have asked me what I'm doing, and I say that I'm working on a venture studio, that we've been working on a venture studio. And sometimes they mishear me and they go, Ooh, you're working on an adventure studio. And this idea of an adventure studio is really interesting and compelling and there's River Roberts: I'm drawing on you. Ty Pattison: It's growing on me. It's almost like there's there's a space for a think tank venture studio that happens on a boat and goes to different places River Roberts: you Ty Pattison: and is a collection of people that that have an adventure studio where we go and build things and educate people and make things while we're adventuring. We probably need a captain and a nurturer, like a cook, and you know, we need a an adventure team to make this so that you can actually focus on doing work. But I'm just planting these seeds. River Roberts: Yeah, know, totally. Well, I think there's a lot of those operators out there. in fact, one of the experiences I always loved was spending time on a liveaboard. And so a liveaboard is a boat that you spend some time on and maybe you spend 10 days on it and you go diving every day or snorkeling or you explore the area. And there's captains and corks and people to support that. And yeah, I think an adventure studio where we go on adventure, but we're still building at the same time. would appeal to a lot of people and I'm sure that between the two of us we could put a very interesting crew together. So I don't know, how about we put that out in the future reservation? What do reckon? Ty Pattison: Yeah. I I think that sounds like a fantastic idea. River Roberts: Yeah, give me proof. Ty Pattison: all right, building, building. River Roberts: How about we go to, how about you switch up the tools? Ty Pattison: anything else on your plate for building? Tist tool time. River Roberts: That is to time, it has good tools. Ty Pattison: tools. So on my plate at the moment I was having a discussion about about models and thinking about the different models. I've been using all of them recently or at least all the Frontier ones and testing them against each other and River Roberts: How have you been using them? Have you been using them on an Alice? Ty Pattison: I haven't been using them on harness. I'm really interested in diving into this with you. I've been using just chatbot functionality for Kimmy and ChatGPT. And I'm losing track now of the like soul, lunar, five point six, whatever they're up to, because it's starting to not matter to me. they're all just good enough for the purposes and I assume that they'll just release better and better things. and then I've been using Clawed with through my Mac and Terminal. How about you using a harness? River Roberts: I am using a harness. I keep kind of wanting to tweak the harness. So I'm still, my main harness right now is still Hermes. But, you know, I mentioned Prime Agent last week and I'm also exploring Colibri as well, which is a really good harness for local use and a mixture of agents. So, and again, I find it the biggest gain across all these harnesses is actually having the persistent memory system as well and having that connected via I'm using obsidian still so a lot of things can be pulled out of longer term memory on obsidian So I've got that working well locally I'm wanting to refine that a bit more for VPS so that I can access it more often because at the moment I've only got it operating on my MacBook, but that's actually a bit of a limitation because I'm not I don't want to always open the screen. so also to get it up and running, like it requires a bit of terminal input to get the pieces running, getting the router running, getting the harness running, all of that. So it's small, but it's still enough friction for me not to be like, I can't just pick up my phone, talk to it and enter, you know, have good output. That being said, I have been kind of tooling around with our Venture Studio model on Telegram. And that seems to be getting better and better as well. So I, like, for instance, asked it to look through our previous transcripts and give recommendations on how to improve this podcast. And it's actually really quite thoughtful and detailed. So I think we should definitely look at some of those recommendations moving forward. the, the, the other Ty Pattison: Yes. River Roberts: tools I've been using this week have actually been quite primal. So, one of the tools I've been actually exploring a dietary tool this week, and I've been exploring ketosis. And so this is the first time I've actually tried to have just a carnivore diet. So basically steak and eggs twice, twice a day for, for at least a week, probably two. And, yesterday was my second day, which. It's, not sure. I, so I, growing up, I had a car in Australia and it was dual fuel. So you could have petrol and you get an LPG. And when you switched it from one fuel source to the other, it kind of did this like jumpy, sputtery kind of like just conked out type thing. And so yesterday I was switching from fuel sources in this like ketosis state and I felt just like that car. was like just conked out. I was so unmotivated. But today I've turned a page and I feel very enlivened in many ways. And yeah, it's exciting to measure that difference. As a simple tool in shifting the inputs, it's really that simple. But I've never done a full carnivore diet for extended periods. yeah, definitely worth exploring. And then the other two tools I've been using is an axe and a knife. Ty Pattison: Interesting. W what metrics sorry, I got I got a quick question. I got a quick question. What are your what are your metrics for like this these input tools when you're talking about changing your diet? Like are you are you energy, clarity, like how are you experiencing these things? River Roberts: How am I measuring? Well, firstly, I should say that, you know, for me, even to get on ketosis protocol actually used Kimi to support me with coming up with the framework. So basically I'm intermittent fasting until 12 and I stopped eating at six. And then it's been very good at being able to identify like, all right, how much food should I eat? Can I eat to stay in ketosis? And like I asked the other day, can I add some barbecue sauce to my steak? And it was like, no, you can't have barbecue, that'll take you out of... So it's quite a fragile state to be in, in some instances of being quite specific on the input. And I'm just thinking, you know, even 12 months ago, in order to get the right information to have a tailored plan for me specifically, would have taken so much Googling to do. And so many different websites and confusing... But with this, it's been incredible. And then in terms of like output of how I feel, and again, it's day three, so it is still early. I'm measuring sleep quality, energy levels, kind of motivation, a clarity of thought as well. That's another one. So those are the kind of main ones I've been looking at. And then under the hood, there's a whole bunch of things that seem to be happening when you extend in a state of ketosis. And the recommendation is actually doing it every couple times a year for two to three weeks. So kind of do this seasonal reset. Yeah. Ty Pattison: Yeah, I've done I've done long long ish fasts. and it changes your it changes the outputs, changes your mental state. Yeah. okay. axe and knife. Simple tools. River Roberts: Yes, and knife. I tell you what, there's something about an axe. It's just such an incredible tool. So I've been using the axe to chop wood for a sauna each day. And I've been using a knife for carving walking sticks and carving different things. And it's just such nice tools. The feedback is immediate. There's such embodied tools there. they are really great. So as much as I like, love the working with the AI tools, going back to the kind of more basic tools as well is phenomenal. And, these, actually quite similar to some of the first tools that humans invented in terms of like sharpened rocks that were used for cutting. So we're not like, when you're interacting with these tools, I think there's actually, a huge amount of embodied knowledge that our body just kind of connects to in a way. Ty Pattison: Yeah. I love that I was basically just like using duct tape to fix stuff the other day and it reminded me of this this diagram that is like should it move And if it should move and it doesn't, then WD forty, like the spray oil lubricant stuff. And River Roberts: Yeah, yeah. Ty Pattison: then if it's like should it move? no it shouldn't move, but it does, duct tape. It's like it's like such s River Roberts: The two ends of the spectrum. Ty Pattison: Yeah, it's such a simple stuff. It's like, alright, should it be cut down? Yes. Alright, axe. Should it not be cut down? River Roberts: Sure, sure, Ty Pattison: No, okay, let's let's bind it with something. Let's like use a different But we sometimes have these tools that like get so complex and then when you bring it back to the simplicity of it. like I was talking to someone about they they were using this platform Zillow for getting real estate listings. They like need a distribution in Zillow. Zillow charged something like forty percent to to brokers, which is massive. And brokers don't River Roberts: What? Ty Pattison: like using Zillow. Yeah, yeah. It's crazy. It will they've just they've done that. They've captured a huge amount of the market. Now they own all of the deal flow for these brokers. And then they can charge whatever they want, essentially. So we were talking about it and we're talking about like different tech tools that we could build. And then they had like a really big audience on Instagram. And I was like, Well what you need is distribution and the like simplest tool for this is make a video, like speak to a camera and be like, Hey, this is the structure of a deal that I'm willing to give to anybody that brings me this stuff. And maybe there's compliance. And like regulatory things that it has to be from a particular person who has a particular license, whatever it is. But the point being that what Zillow is providing is distribution and in through Instagram, you through Instagram and an audience, this person has distribution. So the tool there can just be a message. It doesn't need to be like some complicated thing where you like harness all of the bits, like I'm thinking about AI harnesses. Like, yeah, sometimes guidelines are really good. Sometimes all you need is like a direct line to the outcome and knowing what outcome you want and being able to communicate it really effectively. River Roberts: Yeah, it's a really tricky market, though, because not all distribution is obviously equal, right? So the thing with something like Zillow is, you know, it's going to an audience that is actively looking versus, you know, sending it to Instagram, like what percentage of your following might be actively looking for the thing that you are trying to sell. And of course, there's networks, like it could be passed on. But again, like that's... That's a pretty big if but, and given the complexity and the amount of noise and attention grabbing behavior, it's increasingly unlikely to happen in lot of instances. Unless it becomes a big meme and is shared thousands of times. And then that also then feels like potentially a waste of an audience as well. Because like, man, I used it for one deal. I could have used it for all deals. But then again, like how do people keep coming back? So yeah, there is something about. actually how marketplaces do really support. mean, 40%, that is ridiculous. Yeah. Ty Pattison: It's criminal, right? But you make a good point that it's not necessarily the most efficient way. It's just like where in the process at bottlenecks. 'Cause like if you have a really broad audience that's not looking for the thing that you're talking about, for in this case deal flow on Instagram. Like not everybody on Instagram wants to wants to look at real estate. But if the algorithms are working well, then potentially it does funnel it to the right people. The question though of like you get a more concentrated audience on Zillow is Kind of a blessing and a curse. You have a more concentrated audience, but it also means that Zillow has all of this power, and that pipeline has to get there. That audience has to get there. They have to know that Zillow exists. Zillow is one, like in this case, a semi-monopoly player. There could be, you know, many, many platforms with the same functionality. I see a world in which we we step away from the network effects that have d defined a lot of these platforms. and potentially this happens with agents if they are better at what is it called? GEO or River Roberts: Yeah, gee, yeah. Ty Pattison: A GEO. where agents can go and find lots of different platforms and maybe there's a a standard by which these like listings or these platforms operate. It does take away some of the Yeah. It does take away some of the ability of these platforms to to genum generate monopoly profits, which is what you need sometimes to raise money. So I don't know, complicated. River Roberts: complicated but you know sounds Ty Pattison: Yeah. River Roberts: like a space right for right no right for disruption not right for disruption Ty Pattison: Ha ha ha. Yeah, but I think this is the case for a lot of things. Like like email is an open protocol and so you can have any email client and then you can send messages to those to to so like anybody can send in a message to my email address and I can use any email client to receive that message. Great. Social platforms have these walled they're walled gardens at the moment. I would like to see a world in which the walls drop and you can still have interaction. action like this but in consuming it in a way that you choose. and maybe there's a couple of layers there that are like translation layers between different things. River Roberts: Yeah. I wonder how much of it will be done ethically or just like in more of a hacking kind of way. It's like the paywalls behind all of the research academic papers. And then someone basically figured out a way to hack them all and just publish them all. And that's what a lot of academics go to as a Ty Pattison: Yeah. River Roberts: source of research now, not these kind of paid gardens. Anyway, we'll see. Ty Pattison: interested though in the in the the weights of like how people how many people access it through the the pirated version versus the authentic version because I think a lot of people pay for credibility and you're still gonna have a lot of like we ha we have verified this credibility stamps on things which I think people are willing to pay for. You can go to YouTube and you can get an MBA or you can go and get an MBA through some other through some institution. And if you're intelligent enough to teach yourself the MBA stuff from YouTube then you're probably then you can do it through YouTube, you know? There's there's all sorts of ways there's different ways to skin a cat, shall we say? River Roberts: There are indeed. The last tool that I'd like to actually explore, and I've been looking into it a little bit, but in fact, I think this is something that we could actually spend some time focusing on in the next weeks to great effect, is this idea of the agentic operating system. So this is a little bit, this is kind of a step above the harness. In fact, there's even meta harnesses as well, but I won't go into that this time. But the agentic operating system basically takes maybe a couple of your harnesses and have them in your own designed dashboard and interface. So for instance, with our Venture Studio, we could actually design and implement an agentic operating system where both of us can access it, or we can invite different clients in to access different aspects of it and still input into it. And it goes into our of the models that support our studio. And then it goes into our knowledge base as it accumulates as well. But I've been looking into that as a way to combine a bunch of the models and the tools and skills and knowledge in a meaningful way. And the Agentsic operating system seems to be one worth investing some time and energy into. Ty Pattison: Yeah, I like that. is it hosted somewhere? River Roberts: Yeah, so we would probably host it on a VPS or locally. Or actually probably a little bit of both, right? Where basically you could run it locally. We could run it locally on both of our machines and it would sync up whenever we turn it on type thing or connect it to the internet or maybe it would sync up once a day or once a week or whatever time period makes sense. Ty Pattison: Okay, interesting. T B E. River Roberts: Yeah, exactly. Ty Pattison: Be explored. River Roberts: Rabbit holes! Ty Pattison: give me your best. River Roberts: yeah, Ty Pattison: Wow. I like this rendition of it. River Roberts: I've been just finding myself in a nice kind of forested, sauna based, like quite simple, basic primal rabbit hole. I've, I've, I've, I've been enjoying it. I will say yesterday as I was in that kind of switching fuel sources, ketosis place, I was in this kind of weird rabbit hole of like, no, I just nothing, nothing. is going to work. Nothing. And I got a bit of a loop around that, which I asked her a bit doesn't happen too often. And it was really interesting to notice it as I was going through a shift in input. was like, yeah, really, so much of your cognitive function is actually impacted by your environment and your input. And yeah, so, but it was still fascinating to notice. to notice myself getting stuck in that kind of downward negative rabbit-hood spiral, rabbit-hole spiral. Yeah. Ty Pattison: Yeah, interesting. I experienced this similar thing recently. I mean, two parts to this. I've been in a I don't know what we call it, a reverse rabbit hole. and it's getting stronger and stronger. This vortex is getting stronger and stronger. That's like, how could things not work? How could it like happen? Of course it works. Of course it's gonna work. The thing that I noticed the other day is that I started to get into River Roberts: I love that. Ty Pattison: this into this little rabbit hole of just feeling like I was drained and couldn't do things. And more and more when that happens I realise that it's probably a blood sugar thing. Or like separating out when it's a blood sugar thing versus when it's real and trying to understand like what the yeah, what's real versus what is. where where the information is coming from, if it's just coming from my body needing something, or if it's coming from the outside world being like a real thing. And what I find is as long as I continue to resource myself, the things that are coming from the outside world are so solvable. They're so figure outable. So yeah, the the kind of been calling it the in spiral as in the inspiration spiral. River Roberts: Brilliant. The in spiral. Got it. I think that definitely goes part of Ty Pattison: Mm. River Roberts: your stable system. Not your nervous system, the stable system. Yeah, the in spiral is definitely part of the stable system. Ty Pattison: Yeah, exactly. The stable system that in spiral. as far as kind of thinking rabbit holes, definitely in this or I I went into one the other day that was about what you do to give back to the world versus what you do for joy. or like what you do for money versus what you do because it's intrinsically valuable. And sometimes this has overlap and I think if you are able to figure out how those two things overlap, that's really fantastic. If you can't figure out that overlap, then at least for me there's this tension between the two. So I've been kinda going into this rabbit hole of what is the way that I want to give back and how is that like working or not working for me. and and that rabbit hole of like a being an intellectual rabbit hole versus it being a felt experience of this is what I this this feels good. and so a lot of like the conversations I was having the other day when I was walking to businesses is like, what does it feel like to be inside this environment right now? River Roberts: Mmm. Ty Pattison: And not doing it as an intellectual exercise, but instead doing it as a felt exercise, 'cause feeling is such a human element. River Roberts: Mm. I actually did hear a beautiful comment from Elizabeth Gilbert the other day. And she was talking about how, you know, as a writer and a lot of creatives, they kind of talk a lot about how ideas find them and how, you know, when ideas find you, it's very easy to find yourself in a rabbit hole with that idea and looping in on it. And her advice. which I've, I've practiced a little bit, but I need to practice a bit more, but is to be quite demanding of those ideas when they come and find you. Like, well, all right, thanks for coming, but can you prove that you're actually worth doing? Like, can you like, what's like, why, why should I prioritize you as an idea as opposed to the other three ideas that I already have kind of in my field right now? And actually just being quite, kind Ty Pattison: Yeah. Pitch me. River Roberts: of. Yeah, yeah, yeah. Like, you you got to pitch me exactly. Like, come on. And to be able to do that kind of cognitively, think can definitely be applied to, you know, the more academic or thinking based rabbit holes we find ourselves in as well. go, actually, no, why am I stuck? Like, all right, I need to pitch this. can you pitch to me this why this is actually worth doing? And with these agentic tools, it's actually really quite easy to switch into that mode. quite quickly just to validate the path that you're on in some meaningful way. So, you know, for instance, I did get stuck in a rabbit Ty Pattison: Yeah. River Roberts: hole earlier today around designing this designing this vessel. And I was, you know, I want I want to spend 15 minutes on this and I'll just use the best one that comes up. And then I found myself down, you know, all these different rabbit holes and like, well, actually, what's realistic and I'm like, that important. So I ended up with something that I feel, I feel very happy with. but there was some rabbit hole in there, but I was quick enough to kind of pull myself out of that. And actually part of it was one of the prompts that came back from the whole setup was the agent said, Hey, you with you've got a set of these images you needed to produce that then are turned into videos. But with, with the images don't get stuck like after five or six versions, just pick the best one and maybe refine that once more. And then that's it. And, It's actually a really good positive constraint in a lot of creative work. All right, I'm going to do three different directions. I'm going to pick one direction. I'm going to do one or two variations on that, and then that's it. Whatever comes out, it's good enough. Ty Pattison: Yeah. yeah, absolutely. I've also been this is fascinating that you bring this up because man, there are like three or four threads here that I'm trying to pull at at the same time. Let me slow down. So First of all, you're saying that your agent said, Hey, don't get stuck on this. pick the best one. So your agent was giving you was coaching you on like this eighty percent good enough run with it. Right? Your agent was learning and pulling you out of the rabbit hole, which I think is fascinating. about two weeks ago I was talking to this friend who She runs social media for a brand. she I don't remember specifically, I think she was building her own website and she didn't know how to use these tools, or like she was new to using these tools, and what she said is that a couple of times her Claude told her to just go to bed. It was like, look, we've been troubleshooting this for an hour. You're tired, go to bed. We'll come back to it tomorrow. It's like, wow, that is capability that I haven't seen. River Roberts: Hmm. Ty Pattison: and the fact that you're bringing it up now is like i I think that's pretty incredible. The the coach the yeah, ability to pause and bring you back and and like put you on track in a different way is really fascinating. River Roberts: Well, I will say that the reason that provided that was because I gave it quite a strict timeline. like, need to do this in the next three days. so, you with that being the case, I've got this many hours each day. So, you know, help me achieve this type thing. And with that, it was quite specific and like, all right, you know, some of the things. And in fact, I also said, like, I'm a beginner at making videos as part of the prompt as well. then it's like, all right, so this is where some of the beginners kind of typically get stuck. So yeah, I didn't specifically say like keep me out of Ty Pattison: Interesting. So some of the guiding principles here for River Roberts: the rabbit holes, but I gave it enough information for to, you know, want to provide those insights. Yeah. Ty Pattison: Thinking about rabbit holes or tools, I think that the structure that we've got here and even just highlighting the rabbit holes that we get stuck in has been an incredible tool for getting us out of rabbit holes or like making us aware of rabbit holes and then letting us make those decisions. Like what I realise with rabbit holes is these days I'm much, much better at not opening the door to the rabbit hole. Going like, huh. That is a long way down there. We could open this door and go straight into this, or we could pause and say, Mm No, I don't need to do that. Like I look at my GitHub commits and it has gone from being astronomical to being very selective. and that's probably happened over the last couple of months. It's like, mm, actually I'm not just gonna build that thing, 'cause it's cool idea. It's gonna pull me in and it will be interesting and it will teach me things, but what's important? and that to me is a much more valuable skill right now. River Roberts: Yeah, well, having both modes is important, but keeping them balanced, right? So, this idea of 20 % play, I think is quite healthy in a lot of instances. So at least 20 % of time, you're not focused on necessarily being productive. You can still explore working with tools, but the output is a little bit more in the Kairos as opposed to the Kronos. All right. Should we switch to open questions? Because there's a question that Ty Pattison: Yeah. Well I'm I'm aiming for fifty fifty. River Roberts: is burning. Ty Pattison: Hit me. River Roberts: Are you sitting right next Ty Pattison: Hit me, hit me, huh? River Roberts: to a microwave? And if so, it feels like it's one of the most used microwaves I've heard in a while. There's a lot of opening and closing a door or something. Ty Pattison: no, but I think it's I think it stopped. I think it was a fridge. Or a AC unit. It might be an AC unit. River Roberts: No, no, was like in the past like 10 minutes, it felt like someone opened and closed a microwave maybe 20 times. Ty Pattison: Sorry. that's gonna River Roberts: with like cups going in and out. But yeah, maybe it's a fridge. So we're friends in the fridge. Ty Pattison: think it might have been a fridge and a coffee maker and I I think it's just the kitchen that's right below where I'm sitting, which I haven't Yeah. Anyway, the racket is over, I think. River Roberts: for now. there's a rush hour at them. Ty Pattison: and I now and I now have a coffee. River Roberts: Yeah, and a pitcher as well. I have some pitcher coffee. Ty Pattison: Yeah. Well this is actually an inside joke because I was talking about how pitches are so much better 'cause they're literally two cups usually. And I like more liquid rather than less liquid, so two cups is a is a fantastic metric. River Roberts: You're about to get to... Now that that's the case, I can't think of my other open questions. What do you got? Ty Pattison: what questions are open for me? I have an open I have a couple of open questions. I I have some open questions around specific AI terminology. I feel like I understand the difference between open and closed weights. There is just so much jargon in this industry that keeps popping up that like I'm keeping up with it vaguely, but if I have to explain it in detail I I get a little bit tripped up. Like, for example, I was endeavouring to explain what an agent was this morning and how it differed from other parts of like a harness. I can articulate a harness, but a month ago I couldn't do that. and so there are all of these definite Definitions, I'd quite like to get clear on what a lot of these definitions are and and how they're different from one another. so I have an open question around that. River Roberts: We could do a whole D-type on definitions. Just like a glossary of different AI terminology. I think that would actually be a really nice one to go through. Yeah. Ty Pattison: That'd be great. All right, next week. River Roberts: definitions. Ty Pattison: Definitions. any more open questions for you or should we should we deep dive? River Roberts: I still have a little bit of an open question around like where to be physically while all of this is happening. and when I say all of this, I mean more of the AI piece. I'm like, where do you physically position yourself over the next, you know, 12 to 18 months as these different things kind of emerge and grow. And, you know, was listening to the moonshots podcast the other week and they were talking about. you know, something like 70 % of Americans are pro AI and 80 % of the Chinese are pro AI. And you know, how, how does, how do those statistics in kind of a cultural acceptance or willingness to invest in these emerging technologies actually play out? And yeah, then where do you, where do you kind of sit in that? I keep having this vision of this wave, know, and where do you position yourself so you get a good ride on this wave? And I still see it as a wave that if you ride it well and effectively and with the right people as well, you end up just in such a beautiful sense of abundance in so many ways. And if you... mistime it or completely miss it or end up being drowned by it, it could be a really tough road in the next 10, 15 years. So, bit of an open question, no real answer. Ty Pattison: I got some answers to to part of that open question. River Roberts: Okay. Ty Pattison: want if you want my opinion on this part. I kind of see River Roberts: Yeah, of course, please. Ty Pattison: it as three three distinct I think three is the right number. Let's see as we go through. Okay, so there's the like physical location governance distribution of of affluence that comes from this AI wave that is the lagging part of it. So that determines to me, all right, which country are you in? How is the country distributing resources based on all of this stuff? Are they pro AI and is are we gonna benefit from this? Is it an uplift? And that's relative to every other country. So that's like the rising tide aspect of it. And I think that that River Roberts: Yeah. Ty Pattison: rising tide will be different in different geographies. I think that Europe's lagging a little bit. I think that the US is in a really good technical position, but that the tide won't necessarily rise uniformly across the world River Roberts: Equally. Or, evilly, evilly, yeah. Ty Pattison: and evenly and also in the US. I think that China's in a really good position to both benefit from the technology and benefit from from people being on board with the wave. So so that's like one that's the base layer to me. And I think that that's kind of lagging. I think that that's government driven. I think that that's probably w I don't know. When we talk about regulation, I like to think ten years is a good timeline for like government to catch up. With this tech, it's gonna have to be faster, maybe five years, whatever. for all intents and purposes kind of irrelevant. Except for where you choose to base yourself. I think that there's an element of the wave that we are catching and physical location means that you might be a little bit ahead of the curve on this wave, but those locations are Silicon Valley. I think that the the AI wave there is a couple of months ahead of most other places in the US and most other places in the world. Then there's like, are you tapped in to what's happening and are you using the tools and are you talking to people that are also doing this and that are like kind of keeping up with this? And that it doesn't matter where you are as long as you're tapped in. So I think that that is kind of where you and I are both at is like moving and being in this ecosystem. River Roberts: I kind of like, disagree on a few things and like, you know, you could point Ty Pattison: Mm. River Roberts: at San Fran and be like, there are a couple of months ahead, but also a lot of the big financial resources being invested directly in Silicon Valley. So the bigger conversations happening there, the bigger visions are kind of being executed there in so many ways. That being said, obviously, you know, to have great... effect and impact on the industry, you don't necessarily need all those resources. actually, what we're kind of starting to see is that the teams with less resources are actually doing a lot more. And there's obviously distributed teams as well. But there is something I think to be said by actually having enough physical presence with enough people that are also actively building and have the time and space to be doing this and yeah, having spent the last few months in Europe, I'm not feeling particularly connected to a particularly ambitious group. I'm on a few WhatsApp chats that are specifically about builders, but there's only so much you'll ever get with that level of connectivity and input. I think more physically being in this space is the way to go. I think art just brings us back to the the adventure studio and part of, and also like, you you talk about building on the wave, like on the supersonic tsunami, like actually have an adventure studio that as you're building, you are going on adventures throughout the world as well. And you are connecting to a broad set of case studies and people that are not just what you might experience in one part of the world. which is arguably how and where these different models and technologies should ideally be built from, you know, a much broader perspective, not Chinese, not American, not like one nationality in particular, but you know, really a global perspective. perhaps the reason that's a bit of an open question for me is because it's not necessarily one place in particular. It's multiple or it's with a group of people that move through multiple places. Yeah. Ty Pattison: Yeah. Yeah. Yeah. Okay. Interesting. alright, I have a hankering to dive into to to dive into deep dive. 'cause I have a bit of a I ha I have a bit of a hard stop in twenty five or so. River Roberts: Let's do it. Ty Pattison: so deep dive today on convergence and divergence and to set the stage for this, these are like pretty broad concepts and you could you could go into you could apply this to anything. And it's also a false dichotomy to say that like the world is either this way or this way. But it's a framework for thinking about things. So convergence being the the going into and finding a singular right answer or decision and divergence being going out from where you are expanding a view expanding your resource or your sources reach and specifically I've been thinking about this in how we Go into problems and what is the best direction to go in based on where you are. So figuring out where you are in the world, I think the world is a very messy place. Generally, I think most of the time we don't really know where we are in a processes but like taking inventory of this is what I've got right here and this is where I'm trying to go and what do I what direction do I need to move in from here toward a definite answer and have I broadened my scope enough or do I need to broaden my scope before I narrow into this? This is quite a well defined framework in design thinking, which is this double diamond of all right, I need to do research, I need to distill it, I need to do some kind of design, then I need to s distill it. So it's like this interesting double diamond shape. and I think that this can be applied to a lot of different domains. So I'm interested to dive into this and what the different domains are and how it applies to both what we're doing and also kind of diverge into where this what domains this could touch that we haven't thought about. River Roberts: So we start with a bit of convergence. Ty Pattison: Okay. We can we can start with River Roberts: You Ty Pattison: convergence. What are you is there something on your mind? River Roberts: Yeah, I mean, at least in this industry, it feels like there's a few things that are starting to really converge. So one of them is this idea, at least for the Frontier Labs, the idea of converging on kind of vertical integration. So we're seeing a lot of these Frontier models that are also like going into chips and product and distribution. you know, your Frontiers are all doing that. They're all trying to get. the models, the chips now, and then the products and distribution. And I think those seem to be pretty clear pathways. The other conversions actually seems to be on agent architecture as well around this idea of mixture of models or mixture of experts. So rather than having... one really big model that can do everything, actually being much more reliant on a mixture of different experts and pulling them in at certain points. So those seem to be really two clear industry convergences that seem to be happening at quite a large scale. Ty Pattison: Interesting. So this what's jumping out at me is that one of the clear wins for convergence or why things converge is convergence tells you about a truth in the world. If this is if the same thing is happening from a lot of different disparate River Roberts: and go with it. Ty Pattison: locations and it and it's converging on the same thing, then it kind of points to a truth. So I think that definitely vertical integration, all of these companies are going, all right, where is the value being captured here? Where and it's either value maximization or it's risk management, 'cause there's also a lot of conversation about, yeah, these models are great, but they become commodities at some stage and how do you capture the value from all of these users? and there's an astronaut economical investment in this AI infrastructure. So making sure that that that there's some return on that. The convergence piece there seems pretty obvious. I like these converging factors. Convergence that I'm seeing from kind of builders in the space is that people are trying to endeavoring to find locations or groups or sources of truth. There's like there's an attempt to converge on specific tools and to like find the right answer for this stuff. But I think it's actually a false bot or or a false convergence because we're still in a phase of everything being built out and actually people are we're in this trap of trying to converge too early, I think, in this case, like before this wave has fully kind of crested. in that like we're still building things out. Like yeah, this model architecture is an interesting thing to like talk about the convergence of all of these things coming into like the the mixture of agents. And there are all of these spaces that have not yet been explored to their fullest extent. We're just moving really, really quick. quickly into this, but like the spaces that I have been reading about is like, yeah, maybe LLMs are the are the right kind of architecture for this. Maybe they're just one piece of it. Maybe there are other pieces that are like not yet there and it's moving at such a pace that I think River Roberts: Yeah, like world models, opposed to large language Ty Pattison: The reason. River Roberts: models. Ty Pattison: Exactly. Well and and why this is interesting I think is 'cause there are lots of events that are happening. Like if we talk about a Kairos, there are lots of events that are happening that in the past, that many events happening regardless of the time period would be enough to say, Alright, we've c we've diverged enough, now we need to converge into a right answer. Now there are the same number of events happening at like mm there are power laws to this. So like the the leaps that we did in the last century they're talking about will be the same as the leaps that happen in the next year. so I think that there's like kind of this history what is it history doesn't repeat but it does rhyme. It's like, yeah it rhymes, but the rapper is getting faster. And so there's this weird thing that's happening in our brain which is like, all of these events are happening. We need to we need to figure out what the right answer is, but actually maybe holding some space for like letting this be a little bit messier, letting it be divergent at least in the way that this tech is being developed. there is there is value there. River Roberts: There seems to be a bit of a convergence around the quality of a lot of the models as well, at least in terms of the general intelligence that most everyday user could use, right? So the ability to interface with Kimi versus Open versus Anthropic Claw for general everyday questions and prompts, like the way that you might use Google, for instance, seems to be getting to a bit of a plateau. And I think this is when we start to see a bit of a sort of from the convergence and this kind of what most people need access for. And then I think we see a divergence again based on very specific use cases and very specific, probably smaller language models that are very highly tuned for very specific niches and kind of forms of intelligence and yeah, super curious to see where that starts to take us as this general intelligence kind of plateaus a little bit. And then, you know, even though the general intelligence may be plateauing its ability to be applied to different instances and basically be distributed through different industries, there's still going to be a long tail. of, so even if we just stopped development on any new models, it would take decades for some of the unique products and insights to emerge from even what we just have today. So yeah, that kind of Ty Pattison: Yeah. River Roberts: convergence creates this divergent output. Ty Pattison: Yeah, a couple of points to make here. These models are like transformer models are convergent by nature. They take all of this data and they distill it down into like the most likely answer based on all of this data. which is it's interesting to to recognize this property of of what this is because it's not divergent in nature. And when I think about divergence I think about creativity and I think about exploration. And like where this tool these tools will To evolve and continue to change. But if you just take the transformer models and you say, Alright, this is the thing, this is the genesis point, and then we're gonna go and apply it to all of these different things, to me that is a divergent activity. And alright, we have this hammer, tool, drill, whatever. We have this tool that we're gonna go and apply to things, and that ability to figure out where it applies is where there's a lot of exploration and also like application to the real world, like beautiful work to be done. is not to ask the model, hey, how can you fix these things in the world? Instead it's to understand what's cap what it's capable of and then go and find the problems yourself and apply it to those. my microwave is back at it. River Roberts: know, barking microwave there. All right. Anything else on convergence before we, feel like we've slowly managed to get through to divergence anyway. Ty Pattison: I got a parking microwave. River Roberts: Any other points on convergence? Ty Pattison: A couple of A couple of points to make in convergence 'cause I've been thinking about this a lot. Convergence is not a natural state of the world. Divergence is and convergence takes time and effort. and I find this to be true in my writing and when I compare my writing to somebody who's a professional, is that I can write in a divergent fashion. I can write words and phrases and I can figure out like I can I can explain something off the top of my head, right? and then the being able to distill ideas down and put structure around them is what a lot of institutions do. It's what a lot of universities do. It's what a lot of academics do when they do research is they di they diverge into the world and they find all of the data and then they converge to an answer. And in the convergence of all of the kind of research that people have done and all of these answers that they come up with, there is a ton of value in doing that and it takes effort. And so yeah, understanding that or or exploring that as a truth to me is quite interesting that like the nature of the world is to diverge into chaos and then the task of a lot of humans is to try and make sense of the nonsense. River Roberts: Yeah, it also strikes me that there's probably an element of scale at work here. So if you look at a certain set of things at a certain scale, it looks like convergence. And then if you look at it at another scale, it looks like divergence. So there seems to be something on that. And that's maybe more of a universal kind of law of nature type thing. as you're talking about it, I just keep thinking about. when you zoom in on a microscope and something and sometimes it just looks the patterns look so elegant and in harmony and then you keep going and it's just like super fragmented and it's like, it's potentially just a matter of scale. And if you apply that Ty Pattison: Mm. River Roberts: to even what we're talking about here, it's like, all right, where do things converge and they kind of harmonize and then where do they kind of back out again? But I really like that insight on this idea that when things do converge, there seems to be some level of kind of truth to some extent. truth is subjective and it's time bound, right? Like it does change. And yeah, so one truth in one area is not necessarily the truth in the other. And so I think a really good measure of that is kind of divergent thinking between the open and the closed. Like that's a very clear divergent pathway here at the moment. The Americans are largely focused on the closed and the Chinese are largely focused on the open. Although now that they're seeing how Ty Pattison: Mm. River Roberts: much progress has been made with the open direction, there does seem to be a bit more of a focus on open source models even out of the US frontier labs as well. Yeah. Ty Pattison: Yeah. And and I think that an important aspect of this of divergent thinking is to remain open to changing your it to main to remain open to whatever it is that you're diverging into without having emotional attachment to it so that you don't have to like kill a part of your ego so you don't have to like let go of this big thing that you're really holding on to in order to change direction. 'Cause the whole intention of divergent thinking is to explore different things. Like it's very scientific in nature and I think emotions have like a real place in the world, but when it comes to exploration and trying to understand what's going on, it's not useful. So like remaining open to like all of the different iterations of of i to use this example, like seeing that open weight models are working in that they are creating this plethora of of companies that are doing really well and and that are like all exchanging ideas. It's the Medici effect, it's the the you know, people coming and having conversations and applying different ideas together in different ways to get different outputs, to get different outcomes. Are there are there other insights or other pieces that come up for you when we think about divergence? River Roberts: Yeah, I think the other piece that seems to be really obvious that I mentioned before is there seems to be this divergence around trust and AI and how America and Europe are fundamentally distrustful of this technology. then China and maybe some of the Asian countries are more trustful of it out of the gate. And again, how does that divergent play? I'd say broadly speaking, for me, the insights are, you know, there seems to be a convergence on both the architecture and the infrastructure of this industry. And there seems to be a divergence on the philosophy and governance of this as this emerges, which again, divergent Ty Pattison: What? River Roberts: is actually the healthy instance in this case. So I think having both approaches that kind of keep each other in balance is a really, really important part of it. In fact, for healthy ecosystems, you typically want as much divergence as possible. If you have an ecosystem with three species, it typically ends up with one gets 60%, one gets 30%, and one gets 10%. Whereas if you have 11 species in an ecosystem, they all keep each other balanced, and it's much more harmonious in a way. So actually, the more... diversity and the more divergence, the healthier the overall system ends up being. Say. Ty Pattison: Yeah, and health in this case is resilience, right? River Roberts: Yeah, health is resilience of the overall system. And so, you know, if you have 11 different species in the system and one of them dies out, it's not too much of a... If you have three in the system and one of them dies out, like that is a huge kind of effect and impact on the overall resiliency of the system. Yeah. Ty Pattison: Yeah. What's also coming through for me when we talk about divergence and this cultural aspect of it is that it's completely understandable that there is divergence here because there's a huge unknown and the huge unknown is the I talked about the three waves earlier. The third wave that I didn't explicitly mention is the tidal wave thing. No, I decided River Roberts: that I cut you off on. The title, Yeah, yeah. Ty Pattison: it wasn't it wasn't the point to to bring it in, but it's the tidal wave that could be the you know, these founders talk about like twenty percent, twenty five percent probability of an extinction event from AI. Like that is a massive unknown. It's bigger than any other To humanity and i it okay, that number is pulled out of people, out of thin air, out of out of research, out of whatever. But that big unknown I think i speaks to the divergence because actually nobody knows so that so there's no way to like converge on an answer. If we say yes there's definitely gonna be an asteroid that hits the hits the planet we could we could converge on a an answer for the all sorts of things related to that extinction event or related to that asteroid. In this case we have an asteroid that could fly past the planet and spin it a little faster and create like the opportunity for all of this beautiful new life. life. Weird example I'm aware. but it could also be something that like totally disrupts what we're what we're trying to do here on Earth as monkeys, as fire apes. River Roberts: Yeah, it's for me one of the probably more disappointing convergences that seems to be happening right now is that irrespective of where you're on the world, it does still seem that at least from what I've been looking at, there is a convergence in the use of these tools towards economic means. So it's like, you everyone's looking at them for, you how to make more money. That being said, there is still a huge amount of usage for health and relationship advice as well. but then still the tools are really looked at like, do we commercialize them? which yeah, seems a bit of a shame, but again, that's where open models, differentiate, differentiate a bit because, know, that isn't necessarily the end goal. It's not like, how do we turn this into a trillion dollar IPO? It's like, all right, how do we share this more openly? And then what emerges on the edges of that? But still the convergence around how we use these tools. Again, if we're looking at this from the perspective of we are the first and only generation to parent these AI technologies, the fact that we are still so commercially focused or yeah, feels like a bit of a shortcoming. Ty Pattison: Maybe a really good use of this tech is to explore scenarios in which we don't live in a post capitalist society or late stage capitalism. Like people talk about capitalism a lot and and late stage capitalism and we put a lot of resource and focus and attention on that. But something that I firmly believe is that we need better alternatives to incumbents in order to switch and that's the case in in businesses and it's also the case in like lifestyle design. And so it's like all right, well we I don't want to live in like one location. What's the alternative? the alternative is that I can travel around a bunch. Mm that's not an ideal alternative. What other alternatives are there? and so maybe using these tools to tackle some of these problems, which is like what is the what is the setup of the world and how do all of the elements that are currently at play in our society kind of fit into or morph into something that has all of the benefits that we currently have, the luxury and the like status that people feel and all of these things but not in a organized in the way that we currently have it organized and then what's the path to River Roberts: Hmm. Ty Pattison: get there. Like that'd be a pretty fun and interesting piece of work to do. River Roberts: Yeah, and to build and explore, Yeah. Now I'm mindful of your hard stop. shall Ty Pattison: Somebody gotta do it. River Roberts: we, there anything else you want to bring into convergence and divergence for now? Ty Pattison: no, I I would round it out as a bit of an open question, which is how do we apply this idea of convergence and divergence to our own work? and what are the kind of what are the trigger points? and trigger hasn't connotation, so it's not really the right word, but like what are the You know, what are the flags that jump out that go, I have diverged enough, now it's time to converge onto something, or we've reached a point of yeah, some like some questions or some triggers that tell me when to go in which direction. I think is my open question and what I'm gonna be focusing on a little bit. River Roberts: Yeah, the knowing, knowing, knowing where in the process Ty Pattison: What's about for you, anything else that comes in? River Roberts: you are, are you open? Are you closed? Are you converging? Are you diverging? yeah, I, I like, think, you know, working with other people is a really important divergent process. And then actually having systems that allow the conversions to come back in, in a really simple way is, is, is quite powerful. So, you know, when I was talking. before about the agentic OS, I feel like that would be specifically for us, a really beautiful thing to or piece of infrastructure that really helps our inputs converge in a more meaningful way. Because I certainly know in the past, one of the one of the weak spots for our working relationship has kind of been handover from one model or one output to create an input. And so narrowing that space, I think would impact a lot. Yeah, that's just very practically top of mind. Ty Pattison: We lose context, we lose like Yeah. Yeah. any any quotes to end on today? River Roberts: Well, do you have an AMA? Ty Pattison: god, I almost Yes I do. I've got some funny MAs actually. how about you? River Roberts: Alright, alright, really good. I don't have some funny ones, but... Now that you said funny, I definitely want to start with yours. Ty Pattison: Okay. Alright. what's the last thing that made you cry? River Roberts: What agent is this from? Ty Pattison: This is from long time long time not listener, long time getting my getting the transcripts about episodes. Clawed Fable Five. I I gave it I asked for some AMAs a River Roberts: This is fable coming to the table, finally. Ty Pattison: coming to the table, I asked for some AMAs and they were vague and not good. And then I was like, Mm, here's the most recent transcript and the types of things we talk about. your AMAs are too vague and random, so now it's come back and it's given me some like specific riffing on threads. Yeah. River Roberts: So the question was, when was last time you cried? Ty Pattison: the full question is a little lamer. W what's the last River Roberts: Yeah. Ty Pattison: thing that made you cry and what did you know ab and what did it know about you that you didn't say out loud? River Roberts: Well, you know, I could, what comes up immediately is I almost cried when watching Ty Pattison: Mm-hmm. River Roberts: the Michael Jackson like film documentary thing last night. And it could have been because I was in this weird kind of emerging ketosis state, but also, I don't know if you've had this, but I find when I watch particularly musical, documentaries or documentaries about musicians, autobiographic films about musicians. They feel so tragic in so many ways. And there's something just about MJ and just his brilliance and innocence and yeah, kind of, it definitely brought a bit of a tear. And I think why it brought up is because At least how it was depicted in the movie was it was really just this man trying to share his life with as many people as possible and actually like heal the world. know, there was this scene where he heard on the news about the warfare between two gangs and he invited the two gangs to a warehouse and performed with them and for them and then got them to do the backup dancing in one of his music videos. And like, What a beautiful spirit. So yeah, I felt a bit sad that our system made it so hard for someone like that to kind of stay with us in a way. Ty Pattison: Yeah. River Roberts: What about you? Ty Pattison: The music is a really good example. I I find this with with music, art, poetry. The last thing that made me cry was a poem. and it it really touched me and it made me realise that there are very different realms. of existence, at least in my experience, they're and and we we merge them and we like don't necessarily see the separation of different things but I'm really experiencing this recently where the realms that are emotional, the artistic, the Anytime that I'm dancing, music, poetry, film to a certain extent, they touch these emotional parts that are that are huge parts of me and that realm is something that technology and AI specifically don't necessarily have a place in in my life. Like I would choose to have the emotional realm touched by people rather than touched by technology. I just I I don't desire that relationship with the tech. and there's a certain There's there's hardship and there's struggle and there's tension and there's that is what makes that emotional realm so compelling, is that we have all of these aspects of it. And then there's the the logistical elements of the world and of life. and those things I think can definitely be automated and definitely be dealt with from by technology, so like I'm starting to see this like crying thing as kind of a a I've been calling it the tie test. Like if I write something or if I read something that makes me cry, I'm like, okay, cool, that like that passes the test and I cry pretty easily. but there's like some distinction there between yeah, what realm does this fall into and I'm at am I happy to have that fulfilled by a piece of technology. And understanding the the River Roberts: Alright. Ty Pattison: distinction between the two. All right. River Roberts: So, all right. So if AI can do in 10 years, what took developing economies a century, does that close the global gap or just create a new dependency on foreign AI? So let's say you're a developing country and you're kind of dependent on overseas AI systems and you're integrating them into every system that you work within your economy, both politically, economically and socially. And then that piece is then pulled away. Ty Pattison: I I I have a pretty clear answer to this. depends on what you build and what you build into it. You can choose to build dependencies. I can choose to to start relying on coffee and to want it every day. And in that choice I am deciding to rely on coffee. And that's a dependency that I build into my ecosystem. If I choose to use a technology to build systems that don't rely on the technology, that's a different choice. I think it'd be smart to build both. I think with AI we have the capacity to diversify this a little bit, to build systems that are both like AI independent and AI dependent. so yeah. it makes sense to me that we could build systems that don't rely on this. I also think if from an think of it from an energy perspective, we have this like blip of fossil fuels that is like a massive uplift from a energy standpoint River Roberts: Yeah. Ty Pattison: and we right now have the ability to build systems that are survivable without that. We have to choose to do that. River Roberts: Yeah, I think that's a really, really important point to invest in the things that kind of future proof societies while we have access to this relatively affordable abundant energy source. Yeah, I would agree with you. And then also, like, I'd see how on a national scale, it can be quite easy for politicians to make certain deals that kind of wire you into. an agreement that isn't necessarily in the long term interest of everyone, which is kind of how politics works in general. And we see that with lots of infrastructure deals that have been done over the past hundred years, And I guess potentially the one saving grace with this is that the systems themselves may be able to reduce the the downstream negative effects of that over the longer term. And then again, particularly when you're working with open source pieces, Ty Pattison: Yeah, if they're if they're not designed. River Roberts: it's different. Ty Pattison: Yeah, well and it also depends what what you have in view when you're designing these things, right? Whether or not you have the long view or a lot of this is based on short range politics, which is just a critique River Roberts: Yeah, yeah. Ty Pattison: of politics or current politics. Yeah. cool. Open questions. River Roberts: Alright brother, do you have a quote finished or should we use the one that's on your back wall there? I kind of like it. those that Ty Pattison: I River Roberts: can't see, what has been posted on the wall behind Ty, this whole episode reads, provoke your own illumination. Set yourself on fire. Ty Pattison: I like that one. The other one the I don't know if you can see it's up here. This is River Roberts: Yes, we can do that. Ty Pattison: Jesus would slap the shit out of you. River Roberts: Yeah. Ty Pattison: Share this. River Roberts: I like the first one. I appreciate them both though. And they're both probably quite accurate in many ways. Ty Pattison: Yes, sir. all right. It's been a River Roberts: Alright brother, pleasure as always. Until next week. Ty Pattison: pleasure. Till next time.