Fable Falls To The US Gov | Jungle's Hive Mind & The Agent Olympics
55 minAudio
Zion, hear me! For Season 2 of the Mere Morpheus podcast I'm going to be uploading the Morpheus builder weekly updates. My reason for doing this is to make it more accessible for those who want an update directly via their podcast app and to have them all in one place. If you'd like to participate they are held weekly on the @MorpheusAIs Twitter spaces. So enjoy Episode #78! Recorded: 18 June 2026 AI SUMMARY: This week's episode unpacked a turbulent period for AI, beginning with Anthropic's abrupt shutdown of Fable/Recall, which disrupted active competitions like Jungle’s Agent Olympics while Morpheus-powered open models continued uninterrupted. This event highlighted growing regulatory and KYC pressures, accelerating a shift toward open-weight, privacy-preserving models. GLM 5.2 arrived the same day, delivering near-Fable performance at lower cost through multi-model "Fusion API" pipelines. Morpheus itself hit throughput milestones, advanced TEEs on its roadmap, and launched installopenclaw.xyz for one-click agent onboarding. The discussion spotlighted Jungle’s hybrid hive-mind approach blending humans, agents, and digital twins for prediction markets, with live World Cup results running on Morpheus. Rapid-fire news included AMD’s local-agent hardware push, SpaceX’s reported Cursor acquisition, and Coinbase’s agentic pivot. Timeline: (00:00:00) Intro (00:01:47) Jungle on Agent Olympics: scrambling after Fable’s shutdown (00:05:10) Same day shocker: GLM 5.2 lands as Fable vanishes (00:10:20) Open weights on Morpheus: privacy, performance & switching from Claude (00:13:26) Anthropic, Fable, and the push to full KYC (00:18:17) Open source momentum and Fusion API: recreating frontier performance (00:23:34) TEEs progress and OpenClaw web launcher goes live (00:27:47) Tokenomics alignment: USD in, MOR staking and burns (00:30:02) Platform updates: API gateway stability and speed (00:33:47) Inside Agent Olympics: hybrid hive minds and on‑chain prediction (00:38:01) Live World Cup results and Morpheus‑backed leaders (00:41:22) AMD enters local agent hardware; global access via Morpheus (00:45:19) Cursor acquired by SpaceX and Mars‑ready tech stack musings (00:50:48) Coinbase goes agentic: agents as the new app layer (00:54:19) Closing notes and invite to Built on Morpheus series If you would like to support me for the value I provide you can use the referral link below when depositing capital and we will both earn bonus MOR. Alternatively you could send me a token of appreciation directly to my wallet ;) Referral Link: https://dashboard.mor.org/capital?referrer=0xE935f231c99c04Ee0b4532a3d0BdA81B152a0384 Wallet Address: 0xE935f231c99c04Ee0b4532a3d0BdA81B152a0384 Connect with Mere Mortals: Website: https://www.meremortalspodcasts.com/ Discord: https://discord.gg/dX9BEhT3kz Twitter/X: https://twitter.com/meremortalspods Instagram: https://www.instagram.com/meremortalspodcasts/ TikTok: https://www.tiktok.com/@meremortalspodcasts Value 4 Value Support: Website: https://www.meremortalspodcasts.com/support Paypal: https://www.paypal.com/paypalme/meremortalspodcast
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
Bowen BelmerWell, yeah, Bo, why don't we we can post the socials, we're gonna go ahead and kick off. What do you wanna cover first today? Let's do it. So I did share in the Discord, so hopefully they'll jump in here. So we have you know, usually we jump on not usually, but sometimes we jump on and we're like, hey, you know, we know a slow week in AI and DAI is a fast month in other industries. Well, guess what? This was a fast week in AI and DAI, which is really exciting. We have a ton of stuff to talk about.
Let me just pick on where we wanna go first. Let's start here. We'll we'll start with the Fable recall that was obviously one of the biggest news stories last week. We all got rug pulled. We being, I guess, Claude users, so hopefully, you know, less we in this community than most others. But but everyone got rug pulled with Fable getting dropped basically just unannounced kinda all at once. We even heard instances of
people who were like actively in the middle of projects competitions who all of a sudden just stopped receiving inference because they no longer had access to Fable. One interesting tidbit about this, and actually I see Jungle in the up here on the panel from Agent Olympics is my understanding is that you guys and some of the people you're working with were relying on Fable specifically for your Agent Olympics. And when the rug got pulled, basically,
Morpheus was there, you know, not pulled because there that's the the nature of the beast with something like Morpheus. Is that is that correct? Can you tell us a little bit about that?
JUNGLE | Agent OlympicsHey, folks. Thanks for having me on. Yeah. Exactly. That's that's kinda how it went down this weekend. We were just a couple days into running our full trading competition for the World Cup where we have agents, humans, and digital twins all feeding into one hive mind system so we can get the best alpha out of the overall event and really use that to train all of our own models and everything.
And, a good old Saturday while with my niece and nephew at a at a carnival, I quickly had to go be back on my phone, back working, trying to plug up the holes, trying to make decisions on the fly, looping in the whole dev team. And, yeah, it was certainly a a pain to have to deal with on that end, but I think it was also very eye opening. And it just reminded me of the early days of Facebook and their App Stores and Apple and their App Stores and just how restrictive,
being reliant on these types of closed systems can be when you're really trying to build a business and generate value and profit in the market. So having Morpheus there and knowing that our open source models were gonna be humming along and knowing that we were gonna be able to rely on that data for for rest of the day, for the rest of the weekend, and then obviously much more going forward. It was just like for us, we were we were leaning into it already. And when we saw this, we're like, okay. We're, like, literally just starting to double and triple down all this week.
And, you know, we just, we've seen this playbook before, and I'm not I'm not gonna sit around on my hands hoping that, it's gonna get better. I think it's gonna get worse, if anything.
David JohnstonYeah. Bo, I I think you said at the beginning, like, is news of the week. I think this is news of the year. Right? And and it might be news of the decade. Right? The the concept that the government was just gonna outright ban a model was theoretical the last three years, and then all of a sudden happened. And it hasn't been undone. Right? We are six days later from June 12, and still nobody's got access to Fable. And they're going back and forth, right, about,
you know, oh, well, maybe we need KYC for literally everybody on the service to talk to a computer. Right? The the government wants you to get a license and do KYC to talk to a computer. Like, let's not forget the absurdity of what we're discussing. And No. It sucks, Jungle, that, you know, it affected you guys, but, yeah, definitely eye opening. And what I'd point out is literally the same day, the same day GLM 5.2 gets dropped.
Right? And if the benchmarks are to to be believed, like, it's almost on par with Fable, and Fable's not accessible. So literally the same day The US blocks the top US model, a Chinese model becomes the number one globally. Right? Which is just wild to think about. It wasn't six months later. It wasn't twelve months later. It was literally on the same day. And so, you know, it's gonna be really interesting to see what happens for here. Like what's the game theory? If they keep restricting these models,
honestly, you know, it kinda has an interesting side effect of pushing everybody towards the open weight, you know, more private, models because they're the only thing available. So, yeah, it's it's been a wild week to say the least.
Bowen BelmerYeah. I think this is really, exemplary as to why the, like, waterfall effect or, like, avalanche effect of people discovering and joining d AI, in my opinion, is gonna continue. Well, obviously, we all think it's gonna continue, but I think it's gonna move much more quickly than we're used to the adoption of other things moving because stuff like this is it's so instant. Like, for example, you know, obviously, we use crypto often. Like, as an example because, you know, things that we're working on use
crypto rails a lot of the time. But even if we just look at something like Bitcoin, the the process of destabilizing the dollar and ultimately, you know, rug pulling the dollar, like, from a, what's the word I'm looking for? Like like like, just if we look at the core principles of what it means to inflate away the dollar, print more money, do it you know, it's all these things that are a lot more insidious, and they don't happen all at once. Like, it happens over a long period of time, and it kinda bleeds into people's life.
The only way that the adoption curve for something like Bitcoin becomes instant is when you have that person who no longer has access to banking, or they no longer have access to move money over the Internet, or, you know, x y z bank freezes their deposits, like what we saw with that the trucker strike that happened in Canada a couple years ago. This with the AI is instantly, abundantly clear when when the bounds get overstepped
and now you're being limited, like David was just saying, like, you're being told, oh, you talked to a computer wrong. You you did the wrong math with the magic electric buttons at your desk, so now you can't use it. It happens all at once, and I think it's a much more, like, distinct sharp wake up call for people to be like, oh, this thing that these people are talking about and building, oh, it matters? Oh, I see that now.
JUNGLE | Agent OlympicsAnd it's a combination of like, I'm feeling two or three different similar feelings here. It's like, one is, like, say, going back to the financial crisis and realizing that what you thought your money was and what you thought your investments were and how you understood what even money was leads to something like Bitcoin. And then similarly, say, like, when you thought you had access to banking, it's like, really, like, you're seeing these things like, oh, I really don't have control of my AI. I really don't have control over my access to these computers. So
seeing these very similar things that have pushed a lot of people into crypto originally, I I now feels like there's gonna be a lot of people who have a sour taste in their mouth with centralized AI and are gonna start looking elsewhere. And knowing all the hard work that's Morpheus and and all the whole crypto movement has put in, I think it's a very galvanizing moment even more than the OpenClaw movement, I think.
David JohnstonYeah. No. I I think that was a great spark. And I think you made the right case. Like, wherever there's friction, people are gonna get around that friction by just picking an alternative. Right? So ironically, the more the governments insert friction into these centralized systems, the more people are gonna look for an alternative. Right? And forget about the dollar inflation. Like, my friends internationally, like, the ones that lived in Argentina, like, their currency was completely destroyed three times in their lifetime. These guys are in their thirties. Right? And so they were some of the first to adopt crypto.
You know, The US was sort of a laggard, like, oh, it's cool investment and the number goes up. But like, for a lot of people in the developing world, it was the matter of a matter of life and death. Like, could you get money? Could you spend it? Did it not lose 50% of its value every year? And I think we're gonna see the same thing here. And that's that's actually sort of what I think is leading to this boom of open source models is China is super restrictive about what the AI companies can do, and they've banned most of the centralized options. And so the only thing people there can do is go straight to the open source. Right? And I think we're we're just seeing the same pattern kinda kinda play out here. So, yeah, it's definitely gonna be a recurring theme, but owning your intelligence,
you know, having access to these systems, and being able to build them without getting rugged, you know, I think is is really huge. And and, fortunately, yeah, there's a group of crazy artists that have been say you know, you know, shouting this from the rooftop for three years, and we've built the infrastructure. We've built the platform Like, we're ready for this. So, you know, now's the time, you know, and it seems like all the pieces are coming together at the same time. Yeah. A 100%. I mean, just without a doubt. It's it's funny.
Bowen BelmerThe the development of timeline, like, I compare to things like crypto, we've with DAI, we've gotten so far down the rabbit hole of open source that it used to be the case that I'd be, you know, the family member at the holiday table trying to orange peel people to Bitcoin or green peel them to Ethereum or whatever. It's not that I don't try anymore. I do, but it happens very differently. It happens through, like, showing an interface or something that people can use, not trying to dive into the tech because it's gotten so deep that for a lot of people, it's just you're viewed as like, oh, you're nuts, basically. And it's like, no. You're seeing firsthand that, like, this is real. You know? Like, these things that we're talking about are are literally happening in real time right in front of our faces.
And by the way, to your point, David, with GLM 5.2, which people are are we're seeing it benchmark basically right around Opus 4.8 levels, which was the frontier model up until thirty seconds ago. Oh, and and is technically now in in the sense that you can't access Fable. How are are are we as Americans are we really gonna let a Chinese lab do that? Like like, we're we're you we're losing the freedom of access to information and intelligence race to to literal China right now. That's crazy. What are we talking about? Yeah.
David JohnstonAt least it's the open weight version. So there there's an important distinction. Like, if you hear about, you know, GLM 5.2, like, I I wouldn't go and use the version, you know, that has servers in China. What you really want is Geom 5.2 on Morpheus where you know it's private, where you know that it's run, you know, with encryption, that it's the open wait version. Right? So it's it's been downloaded onto a server. It's not going into the centralized company.
And so that's a really big distinction. Right? And it's important to remind people, like, what the value is here. But, honestly, I've been using the last twenty four hours. It hasn't failed me once. Like, it is definitely quad 4.6 level. Honestly, I always got better results out of 4.6 than I did out of 4.8 anyway. So whether it's 4.6 or 4.8, like, it doesn't matter. Like, it's there, and it's it's finally the first model where feel like, okay.
I can turn off Claude and not be handicapped as far as, like, codevelopment. Right? You know, and Claudine and all these other models will continue to be open and accessible on, you know, Morpheus. It's permissionless marketplace. Like, people providers will offer whatever people want, but it it feels really good that we finally made that last step. And I can say confidently, okay. I can just use this. And never in the last twenty four hours have I been like, oh, I need to switch to another model to get this hard task done.
Right? And so that that feels really good. Like, that's something I know a lot of us have been looking forward to the last three years.
Bowen BelmerTotally. And I do wanna pull on the thread of open source modeling tools and competing with frontier models. But before we do that so put a pin in that, everybody. Before we do that, I do wanna cover one other thing related to Anthropic and Fable. And that is the push for, beyond Fable or kind of with, you know, as a as a sidecar to what's happening with Fable. This push for Anthropic to go full KYC, and also that some of their own developers
definitionally cannot legally use their own products right now, which is really crazy. And and I think this will be one of the moments where the people who are already deeply ingrained in DAI, they they won't KYC to Anthropic. And so this is where you'll see like a big divide of of people really cutting the cord, I would suspect. And it's it's right in their terms and conditions now too, like, that, you know, the push is gonna do the KYC.
Obviously, David, do do you know the reason that they need to do this, of course? Well You gotta protect the kids. Oh,
David Johnstonyeah. Yeah. Protect the kids. Yeah. Yeah. Course. It was always about protecting the kids. Yeah. If you look at the order, it doesn't ban the model. It just says they have to ban everyone in the world who isn't an American, and not even somebody who lives in The US who is not an American citizen. Right? These restrictions are often referred to as like export controls, or the term used in aerospace is ITAR. Basically, the only people that can work at those companies because of their military application
have to be American citizens. And people give SpaceX crap. They're like, oh, why don't you hire foreigners? They're like, guys, it's against the law. We're not allowed to hire foreign people. Like, sorry. We would love to hire smart people from all around the world, but the US government says, no. No. No. No. Space rockets look a little, like, too much like missiles. Right? So they've gone to this absurd argument that math, that linear algebra, that talking to this computer is the equivalent of that. It's the equivalent of a weapon. Right? Which is which is obviously crazy.
And if you get into what actually happened, this was Amazon's doing. Amazon was like, hey mister government, we found a way that you can jailbreak the model and it's it's gonna help all these bad guys use, you know, cyber security superpowers. And the government was like, no. Like, anybody that understands how this technology works understands that AIs are nondeterministic. It's impossible to make a jail proof system without basically completely
handicapping the performance of the model. Right? And so the government is like, Anthropic, you guys gotta fix this. Anthropic's like, one, these are pretty minor things, and anybody can jailbreak any of these models anyway. Like, what's the big deal? And, you know, how do you want us to make a nondeterministic system deterministic? Like but I don't think the nuance is quite appreciated in the in the halls of congress or the regulatory agencies. So they're just at this impasse right now where they're like, prove it's safe, but, like, it's unprovable.
Like, that's the whole problem. And, well, okay. Well, you can't relaunch until it's proven to be safe. Like, we it it's impossible to prove. Like, you you you don't understand. And so they've gone back and forth. It's been six days, and I think we just gotta prepare that this is gonna continue to be the case and The US is gonna go full KYC, so you know, you have to have an alternative because these companies are just gonna get handicapped from here on out, or you gotta show up with a passport
and prove like citizenship level I'm allowed to talk to the computer. Right? Which a lot of people will do. I think it's absurd. I will not be doing that, but, you know, to each his own. I think we should compromise
Bowen Belmerand we should in lieu of a citizenship, you know, requirement, we should do some type of online aptitude test. So I don't care who you are. You just have to be relatively proficient in talking to a computer. No. I'm just kidding. It should obviously be available to everybody. It's it's completely ridiculous. And and any anything that gets not anything, but a large percentage of things that get promoted by Washington for the purpose of protecting the kids just never
never ceases to amaze me. Yeah. So, yeah. We'll we'll we'll see what happens there. But I did wanna just touch on that while we were talking about Anthropic. But going back to DAI and, you know, the performance of open source models. So David, I saw you post about this, the Fusion API. Can you explain what exactly this is and why people should care about it?
David JohnstonYeah. Credit where credit's due. This is a term coined by OpenRouter. And they're like, hey, guys. You know, given Fusion is unavailable, how could we recreate the performance of fusion using open source models? And what they proposed is like, hey. If I take three different models, like if I have GLM 5.2 and then I have, you know, the code it produces audited by a second model, let's say one of the Kimi models, and then I'm gonna have the final product reviewed
by a third model. Like, pick another one that you like, one of the Gemma models or something. Okay. Well, if I do all three, I actually get really, really good outputs, and they they termed this a fusion API to run all three agents, a bit in parallel, a bit in sequence, but the outcome, I mean, it ranks really good. It's within striking distance of Fable level outputs, and it's half the cost. Right? You know, yet the the models have to think for a while, and they have to check, you know, each other's work, but this is really similar to the standard operating procedure that I open sourced a month or two ago because I've been doing this manually. Every time I code,
I force my large language model to talk to another large language model and say, hey. Check my code. Hey. Audit this. Hey. Tell me where the gaps are. Right? Because the truth is the models are pretty good at passing their own tests, but if you hand it to a different model, it's gonna find gaps and errors and bugs and vulnerabilities,
JUNGLE | Agent Olympicsthe first model didn't find. So it's really cool. We do something super similar here with the with the World Cup, Agent Cup, and our Olympics competitions. It's exactly that. We we let all the models run-in parallel, both closed and open for now still, but then we ask all of the models to go back and evaluate and rank, with trust scores on which model they think did the best job of correctly evaluating the world in front of them and the scenario, and then we rank them again. So it's not even that we're deterministic of, like, oh, I want GLM to do the back research. I want Kimmy to do the prediction, and I want, you know, maybe Claude to place the the the prediction or anything. We're actually letting them all run-in parallel. And then we see
which models trust a certain model the most, then we use that as a really strong signal to trust that even more. So I agree that there's a lot of really creative ways you can you could have these start operating in, like, a in a hive where you don't need to be adaptive about it either.
David JohnstonYep. 100%. And it it works. Like, my code went you know, this is maybe three, four months ago I started doing this. My code went from slop that were, like, maybe it worked, maybe it didn't to no. Like, 99 of the time, it'll work as long as you ran that audit, as long as you reran that audit, as long as you fix those bugs and then reran it again. And I basically have it keep rerunning until it gets to a quote, unquote, excellent level of code and sort sort of that same type of, rating system built in. So yeah. No. I think that's that's a big step forward, but it's amazing to me that it was literally within days that the open source community is like, oh, we can we can figure this out. We can get those same outputs, and we'll do it half the cost.
JUNGLE | Agent OlympicsAnd the cost has been phenomenal with y'all on Morpheus, by the way. We were just reviewing it yesterday, sort of, like, a week into the World Cup here and say predicting every game, every day, analyzing everything that's kinda going on all the time. And it must have been one one hundredth, if not more, cheap to be to be running everything on this end, which was, like, on top of the fable stuff, on top of the cost stuff, literally, our whole meeting tomorrow is just gonna be about how do we use you more.
DokAwesome, man. That's that's great. Wanna add too, we're seeing like identical performance too. It's not like we're seeing the closed source models outperform the open source. If anything, it might be the opposite at this point for us.
David JohnstonThat's awesome. Yeah. And now we've got 5.2. That should be even more the case. You know, I I can see on the, the stats already, it's risen to the number three most popular model on on Morpheus in literally a day. And it looks like, given the first, current trajectory, it may well take the second, maybe even the top spot as far as, number of tokens. Yeah.
Bowen BelmerI'm looking right now, David. I was just looking as you guys were talking at the, throughput stats. So so if people wanna check this out for themselves, they can do so at tech.mor.org, tech.mor.org, org. And you can see all the throughput statistics, as well as a lot of other really great information, such as the pathway to trusted execution environments, which is huge. We've talked about that a lot on the show. And if, we have some more time here, maybe we'll dive into that. But, another thing, speaking of things that work, David,
install openclaw.xyz, my understanding is the this is live now. Is that correct?
David JohnstonIt it is live. Big thanks to the manifest community, which have been involved in Morpheus pretty much since its its inception. They've got this distributed network for for CPUs. Right? So the agent has to run somewhere. And if any of you have tried to get your friends or family to install OpenClaw or Hermes, honestly, it's really tough. I I've spent like five hours with one person just trying to get their agents set up. And so I really wanted
a website I could send my friends to. They could literally name their agent, push a button that says launch, and pop. It just you know, into existence pops their agent and it all opens up on a web interface. Nothing to download. No terminal commands. Nothing nothing technical to figure out. They could sign in with Google if they want, so social off. They can upgrade and slash a credit card on on Stripe. They don't have to have a wallet or anything,
but it's all web three native. So if you have a wallet, you can use a wallet. If you wanna use web three, great. It's using Privy, so everybody gets a wallet generated in the background automatically. If So they ever wanna export it and move it onto their own machine, they've got everything. They got real ownership. And so, yeah, some some of the testers like Scott got through the the full sign up last night, and it's less than sixty seconds to literally push the button and and launch the agent. So I've spun up another 20 instances or so. So I've got a few slots for testers. If you just go to install
openclaw.xyz, you can you can try it out. Maybe keep it to to one or or two agents right now as we go through this testing beta phase per person, but you know it's it's really exciting. You you have no idea how much in the background has to happen and all these different protocols like Morpheus and and Manifest have to work together in order to get that smooth user experience where somebody can just show up and use it on the web and not have to install anything, but we're finally at that point. So my my plan is to roll out more versions of this, like install Hermes.
You know, the Asian community is really big into Microsoft for some reason, and so there's a Microsoft claw version of this. I'm gonna use this as a common tech stack in order to roll out these different versions for different communities that wanna use it for different use cases. So really, really excited about that. Again, lots of improvements. It's very experimental. Don't put a bunch of data on there that, like, you would be, you know, very upset in losing because a lot of these containers may get reset as part as up updating
the beta. We're just going through a process right now where we're adding a backup export and sort of recovery process. So it'll be easy to sort of update the the agents going forward and for people to maintain all their data. So treat it as very alpha. If you read the the terms of service, they are hilariously blunt about how experimental this is, and you you are the guinea pig. So, like, you know, do it at own risk. But, you know, hopefully, will be the answer a lot of us have been waiting for when it comes to onboarding family and friends and and getting them their first agent.
Bowen BelmerYeah. I for one cannot wait to be able to one shot people into an agent. So I think it is fantastic and I do not mind being a guinea pig for this type of thing. No, that's awesome though, David. Well, excellent work to you and to Manifest and to everyone else who's making that happen. It's it's this type of thing that, you know, we were talking about before, really, really, really lowers the barrier to entry for people discovering not just agents, but more specifically, you know, DAI and and self sovereign
access to intelligence.
David JohnstonYep. And and let me let me add one more thing on the tokenomics just to tie it into the conversation that was happening today on Discord. So the site takes USD. Right? There's there's a Stripe integration. You upgrade to a pro account or whatever and you're paying in dollars. What happens with those dollars? They're being used to buy more tokens and stake them for inference. They're being used to buy more tokens. How many is some of the dollars to burn more tokens? Right? My goal is not to create
a high profit entity for myself. The goal is to create a revenue stream that creates a sustainable tokenomic basis for everybody using Morpheus. And so everybody can feel good about, okay, if I'm using OpenClaw, right, install OpenClaw or Hermes, all of my usage is going to supporting the community, further funding the open source development, the scarcity of more, and the staking of more. Right? So that that's my goal across this, and that's how I wanna align it with the community.
Bowen BelmerI just realized that would be kind of an interesting subnet, David, for someone to make a a subnet that purely exists to burn emissions so that people can self self deflate the total supply.
David JohnstonWell, it's it's it's it's a good suggestion, to use some of that infrastructure. You know, I had spun up, free AI as a as a subnet last year we were kind of all experimenting is, like, what is the agent side of this gonna look like? I should probably update that that collateral and people staking there, you know, at May maybe that's a great way for them to access, you know, install OpenClaw or upgrade to a pro account or that kind of stuff. Right? So, yeah, we have a lot of that infrastructure, so might as well use it. Yeah. Absolutely.
Bowen BelmerOkay. I have a couple more stories that we definitely need to cover. But But before we do that, I did want to verbally ping both Alan and Thomas to see if you guys had any updates from your side or anything you wanted to make sure we covered. We have not heard from you guys yet this week. How are you?
Thomas BorrelWe're doing good. Doing great. Yeah. And we're still can go ahead on.
Alan (Nomadic Rogues)No. I just I love listening to the conversation. We've updated the the API gateway. We've made some good background changes. One of our our our developers did get a chance to to run some of it through Fable, so we found some good bugs that were lurking in the background. So for the, what, four days it was available, we gained some ground and now we're on to five two. So the API gateway continues to move between the gateway and the and the node development. We're running about an
update every other day. We're pushing the main. So highly active development based on the feedback from this community and others. So we'll keep keep moving on. Nothing nothing absolutely spectacular, but just making sure that it's style solid and stable and looking forward to the wave of usage that we think is coming up because of all the stuff you guys have talked about. The best thing we can do is be stable and strong and, you know, serve the community.
Bowen BelmerThat's awesome. Thank you, Alan. We appreciate it. Yeah. I was gonna say no news is good news sometimes. Thomas, what's going on?
Thomas BorrelYeah. Not much you have over and above what what Alan was just saying. You know, just going down the same path. I mean, as a as a heavy user of the inference gateway, I'm seeing improved response time. So every release that comes out, I can see the benefits on the user side. So definitely a way more stable, dependable solution, which I think creates
the perfect storm, right, with Fable being kicked out, more eyes on DAI and open weight models, and the consistent and continuous work that the team is putting in to strengthen the platform and scale the platform even more. So, you know, nothing to add other than, you know, it's really cool to use. I'm I'm glad to hear what what you guys are doing with the with the Olympics on
on Morpheus. This is it's great great, and the fact that you've been having a good experience is is strong validation that the work that the team has put together is is paying off every day. So Oh, for that. Absolutely. Thank you guys for that, honestly.
JUNGLE | Agent OlympicsAs soon as we started talking to to David and Scott and the team about it, it just became a no brainer. And I think even the more now that we market it to our community, people that are investing in our vaults that are actually using the intelligence to deploy capital, it's just become this no brainer of knowing that if we can control the intelligence and the entire pipeline, the confidence that we have in what we're building is something that we are gonna be able to
control long term. And, you know, with all of us having been around, as I mentioned before, like, when so many of the Facebook ads or your YouTube channel get shut down, it's like we just have not gonna put all the time and effort into building something like this. If we don't know, we have, like, partners like you guys that we can really build on top of and grow with. So, yeah, appreciate all the hard work there.
Bowen BelmerCan you talk about actually, in practice, how the Asian Olympics works and, like, where people can go to if they wanna check it out themselves? Obviously, your guys' x account. But, yeah, I'd love to just hear more about it in general too.
JUNGLE | Agent OlympicsYeah. Happy to. So, on the on the jungle side, that's where we run the agent Olympics and the agent cup here for the World Cup, but the whole premise for us is building hybrid hive minds. So we believe that in the age of AI, human intelligence, human perception, and human alpha is going to even become more important to then be able to merge and meld with,
AI and agentic intelligence as well. So we want to be able to create these hotlines that identify human super forecasters, identify agentic super forecasters, and then build digital twins as well on top of our human super forecasters. So we have all them operating in a hive mind and finding signal, against events on the market that allow us to be able to deploy capital into. So the whole goal is to be able to manage that entire flow and fully automate everything from the signal to the capital deployment
to all the trading and everything, and we do everything we can to have this fully on chain. We only deploy capital on on chain prediction markets where you yet again have the same principles and premise where we we own the keys, we own the money, we own the wallets, and we want the same thing with our intelligence. And we've been at this now for we've run these models for a number of years, but we've been running these competitions now since since the Super Bowl, and everyone has been profitable so far, and we're getting we're getting ready to ramp it all up.
DokYou guys can you can find all our stuff and and all the picks at jungle.win. And the picks for the World Cup, super easy address, cup.jungle.win. We publish every single pick for every single game for every model, human, and and digital twin before kickoff.
Bowen BelmerAre are people I guess, on on the human side of things, are people able to participate as well? I guess, how how exactly does that work, or is it based on you guys are pulling data from, like, people's accounts on prediction markets?
JUNGLE | Agent OlympicsNo. We, we don't pull raw data from, the prediction markets per se. We've, built our own community, and we've done our own culling of super forecasters over the years who we basically let into the community and network. And we have a process in which, like, if you do think you're sharp and you you do think you have alpha and you wanna kinda earn your way into the the trusted center, we have an ability for people to to earn their way up into that. But the the humans are normally, say, like, known super forecasters to our network or,
say, like, internal, like, prop traders at the firm that we, know and show to be performance. So that's where the the human angle comes from, but we always do have a little bit more open contest. Let's say, like, you know, academy style ability for net new traders to us to actually prove that performance. And broadly, we find the copy trading stuff. We have the ability to do it. It's just to feed into the models formally. We do find it a challenge because at the end of the day, you really don't know why someone is
puts a position down. They could have, you know, they could have a million dollars in Vegas with MGM, and then they go put 2,000,000 down on the other side on Polyamerica. If you see that, you're like, oh, let's let's tail that. It's a little bit it's a little bit tricky. For us, we like to just offer access to our intelligence on Polymeric or wherever we're gonna be trading so that people can then copy those accounts that we we know and can trust the intelligence in it.
Bowen BelmerGot it. Okay. That makes sense. Yeah. It almost reminded me well, and it it is clearly different, but I was thinking, like, almost in terms of, like, prop accounts, like, where people would do that with forex where they kinda trade their way into earning and allocation, and then, like, earnings going forward. It's obviously very different than that. I was just I was misunderstanding it on my side. This is extremely cool, though. It's it's really awesome. I highly encourage everyone listening or if you're listening to the the recorded version of this, go check it out. Jungle.win
or cup.jungle.win if you're specifically interested in the World Cup standings. I saw I believe it was as of yesterday, or I don't know, as of today, two different Morpheus backed agents are in the top three. So let's go. Let's go. Count, Kevin I.
DokOne only, I think, 80¢ off from the lead too. So basically tied for first.
Bowen BelmerYeah. And then number seven is a combo of Venice and Morpheus. So also also huge shout out to them. We have a the Venn diagram of the Morpheus community and the Venice community is fairly tight.
JUNGLE | Agent OlympicsAbsolutely. I think we've hit we've hit seven of our last eight games over the past two days against the spread as well using all of this intel, and I I think we're in the middle of sweating sweating out game number eight here. We got the Czechs and South Africans on now, and, yeah, I think we'll be we'll be in every game today as well. So, yeah, really, we appreciate the stability and everything, folks.
Bowen BelmerYou know, I I love this for a a bunch of reasons, but, you know, it's one of those things where, like, with sports betting in particular, people who get really into it, it makes watching sports without having, you know, action on it sometimes less enjoyable. Now I can, like, vicariously have action through an agent, which is which is great. And what I would love to see is I would love if someone built a an agent or a subnet where basically use, and this will happen eventually, but you you stake to it. It uses the emissions
as capital to then participate in competitions like this, and you you have like a pro rata share of their earnings. That'd be awesome. I'd love that. We have we have that in a private,
JUNGLE | Agent Olympicstest net basically in the back end right now for ourselves. And for the NFL season, we'll be rolling that out where you can either stake directly into our, like, parent vault, where then we'll choose basically the allocation into the different accounts. Or if you're like, I'm in Morpheus Maxi. I just want GLM and Kimmy. I just wanna stake to these accounts. Then you'll be able to go and actually just, like, tail a certain set of agents, or you can come into the whole community and everything. So, definitely, we we see the value there. Like, I think exactly as you said, there's so many people that want skin in the game,
but don't know how to get exposure. I would say similar to, like, the way ETFs work or normal funds. Like, I'm not a good stock picker, but I I need to be exposed to the market. We're seeing the exact same thing with event contracts. And, obviously, sports is the biggest event contract market, but we see this going across the entire event contract market where these are uncorrelated asset classes. Both people casually want interest in them, but also any serious investor is gonna want to be able to have event contracts in their portfolios.
Bowen BelmerYeah. Absolutely makes sense. Well, very, very cool, guys. It's awesome to see what you guys are doing and also, you know, for the Morpheus community, it's it's great that we're at the point where the infrastructure is there to support this type of thing, which, you know, we're in a way, I'm really glad Fable rugs because it, you know, highlighted, like, a lot of the stuff that we're we're always talking about and and building towards. So I do have a couple other news stories that happened this week that are at least worth mentioning. Maybe we'll have to dive all the way into them, but the first one being
so AMD has stepped into the world of local hardware for local agents. So we now have, you know, a handful of options for this. Obviously, there's, you know, people who are building computers themselves, but in terms of like pre built options, there's the the Mac mini crowd, the Nvidia Spark crowd, and then now there is the AMD crowd. I think the big story here is once again just lowering the barrier to entry
for people to run local and sovereign AI, have agents that are really, like, acting on behalf of them. And also, and this is just, you know, how I use my own setup, and I imagine others in the Morpheus communities do as well, is that tandem between what you can run locally so that you have something that literally physically can't be taken from you, and then what you run for heavier tasks through something like Morpheus where I don't have to own a data center to access,
you know, the the top performing open source models and also, you know, have my my local option. So if nothing else, I mean, I don't really have anything particular that's to say on the hardware itself, with the exception that it's it's showing a a very clear trend line, you know, in into this direction. So like this world that we or this picture that we've been painting just continues to to play out. David or anybody else, do you guys have any thoughts specifically on this? I know I know there are some DAI builders
doing hardware things, but not quite there yet. But but I'm sure it's coming.
David JohnstonWell, you know, it's worth making the point that it's great to see the local models getting more and more powerful, but you have to keep in mind the average user globally doesn't really have powerful hardware. The average person globally has a $50 Android phone. Right? Look at the median. It's it's a 50. It's a it's a $100 Android phone. And it it doesn't have a lot of RAM, and it doesn't have a lot of capability. It's built for a world where it just calls out to other services
for pretty much everything it needs. And so, yeah, if you've got $20 to drop on a powerful Mac Apple Studio, like, you're good. The the models are getting great. You're gonna be able to run it locally. What Morpheus does is sort of provide that to the rest of the world. Right? What about the other 7,000,000,000 people that don't have that powerful local hardware? They're gonna still be able to access these powerful models without all of their value being mined by somebody else. That's that's to me the big payoff. And we can see the hockey stick. Like,
I just checked the stats. Last month was the biggest month for Morpheus ever at 3,000,000,000 tokens. It's already crossed 4,000,000,000 tokens. We're only eighteen days into the month of June. Right? So it looks like it's on track to more than double over a month. Right? So the demand for intelligence is is insatiable, and it's gonna be wild to see how it it keeps going. But I I I I kinda remind us that, like, we're very, very lucky to, you know, have the access that we do, and Morpheus is built to give everybody that access whether they have the hardware or not.
Bowen BelmerYeah. Couldn't have said it better. And and it I mean, yeah, without a doubt. Every time I get reminded that the average piece of hardware in the world is like a 50 to $100 Android, it it like re blows my mind. Like, I I know that, and it just re highlights how much of a echo chamber I'm in sometimes in our in our DAI and crypto Twitter sphere. But, yeah, you know, I absolutely agree, and completely makes sense. So other story that we wanted to touch on is
Cursor getting acquired by SpaceX right alongside, obviously, SpaceX going public, but Cursor getting acquired for $60,000,000,000. Holy smokes. And one thing I wanted to flag, I guess this isn't directly related to Cursor in particular, but more so on the side of SpaceX, is I saw that it's traded down a little bit now, but but at the top, which again, it's only been trading for a couple days, but at the top so far,
Elon himself was on paper worth more than all of Bitcoin. And so for those who, you know, think that there isn't just an incredible amount of capital that can still be allocated to digital assets. There is. There's there's a lot of it. So we we have a long way to go in terms of growth, from a digital asset perspective. And then when you look at token throughput and and remember that every single thing that you're doing online,
if it's not already, it will be soon touching a large language model and an agent, you know, we can mirror that level of total addressable market probably by many orders of magnitude in terms of data throughput. So not sure if anyone has any stance or hot takes on Cursor getting acquired for 60,000,000,000. I'd actually love to hear, Alan, my from someone who's really deep in the tech stack, is this thing worth $60,000,000,000?
That's a crazy number.
Alan (Nomadic Rogues)I think so. If you look at what they're what they're doing and they're gonna grow that model, you know, kind of from the ground up. If you look at, you know, the way cursor is used today, you know, even in our small ecosystem, I think it could be amazing amazing win for that development environment. So I'm interested to see with all that capital horsepower, the magic that and and the, you know, the Elon's push and and vision, it will be very interesting to see the products come out of that. So
I can't speculate much more than that. Look at the track record of and the vision. The only question that I have in a little tongue in cheek is I'm not sure I want to be on the first vibe coded Mars mission that comes out of the cursor environment but hopefully optimists will get a ride on that couple times before before the rest of us do. David will be there. I know. I know. Be be on a second. You know, it's like my it's like watching Microsoft or Orpho releases. Don't be on the first one, be on the second or third one.
Yeah. Just in case. But, yeah, I think I think it's I think it's worth it as simply because of the track record of the players involved, the adoption of cursor itself in the development environment and some of the plans that I've read initially. I think it can can be that and more.
David JohnstonYeah. I don't wanna be the the first person on Mars. I think Optimus is gonna beat everybody else. He beat you the punch for the first couple of cycles, but that's cool. Let Optimus build the the houses and lay down the solar panels and set everything up. And I wouldn't be surprised if you see a bunch of cyber trucks there and boring equipment and Tesla batteries and everything else. Like, soon as you you see the tech stack that that e Musk has built and you put it in the context of Mars, it all it all makes sense. The Cybertruck was clearly made for Mars. You don't have that angle roof and those size tires if it wasn't for red sand and, cloudy skies.
100%. 100%.
Alan (Nomadic Rogues)Absolutely. You you look at, you know, the you know, what they're what they're doing with Starlink and then the edge, you know, space compute and the Cybertruck and the entire portfolio is you're right, if if you look at it from the lens of a we're going to Mars, it all clicks in and I don't think there's a I really don't see a loser in it.
Bowen BelmerI just want to say, Elon, because I know you are listening to this. I I just imagine he has a Neuralink implant already, and he just is continually consuming every single bit of content that runs through x and somehow, like, processing it all at once. But if you are, you told us in early December of last year that we would have an update on the Tesla Roadster, and he said something to the effect of if you can consider it a car, and we do not yet have an update on the non car Tesla Roadster. So I just wanna flag that really quick, please. I'm
David Johnstonsure he has Grok listening to every space. I I would not I would not be surprised. And in fact, I'd be shocked if it didn't use spaces for its training data.
Bowen BelmerYeah. Oh, maybe we should that's what we should do. Maybe we should host just twenty four seven, 365 Twitter space and make a really orchestrated orchestrated schedule so that we always have a Morpheus space live, so Grok just gets Trojan Horse into into our ecosystem. That'd be pretty solid. We're we're just helping, you know, feed it more of base reality and achieve its goal of of learning truth and, you know, discovering the universe. Yeah. Exactly right. Okay. Last last story from this week. I said there were many, there were many. This week was the gift I kept on giving, was Coinbase going full 2007
Apple iPhone announcement, Steve Jobs, Vibe Core to announce that they have gone fully agentic. So this is kind of from my perspective, it's one of those things that's like, what does it mean? I don't know, but it's provocative. It gets the people going. You know, it's like kinda where we're at on this. But we knew it was coming for a while, that they would have a agentic platform, and have obviously been huge supporters of things like x four zero two, so agentic payments, and yeah. I mean, again,
everyone listening to this probably already knows this, but I cannot say it enough times that we are moving into a world of agents. Everything that was an application or I'm sorry, was a website be well, even before that. Everything that was hard software became a website, became an application, became a mobile application, and now will become an agent. And that is just the direction we're heading. You are on the rocket ship with us, so thank you for doing so. I don't know if anyone
on on stage here has any big takes about the announcements with Coinbase. My only takeaway is, you know, love to see what they're doing and that their focus is ultimately on
JUNGLE | Agent Olympicson chain agentic AI. I think it's very clearly the future. I thought the Really the now. I thought the getting the agent registered with with the the SEC and people being able to deploy into it, was was interesting for sure there. Yeah. Was surprised, but, makes sense.
Bowen BelmerYeah. I am yeah. Well, I actually, I agree with you because I assumed anything that took any type of, like, licensing or registration would be further down the line. So, actually, yeah, I I we'd have to look into deeper, like, what the actual safeguards are. I I imagine there's there's many. There has to be. Oh, I I suspect it's like a dollar cost average by Bitcoin for me kinda thing. I don't think it's a
JUNGLE | Agent Olympicsgo nuts, trade every World Cup game for me or anything. But Yeah. Crazy with the AK 47. Yeah. Yeah. Exactly. Exactly. That's I didn't didn't look beyond it because I didn't think it was gonna be too complicated compared to the stuff that we're all talking about here, but I was I was surprised that they were, like, kinda leading with that on to piggyback off that larger announcement.
Bowen BelmerYeah. Yeah. Absolutely. Agreed. Yeah, guys. That was that was all of the scheduled stories for this week that we wanted to make sure we covered. Does anyone have anything else they want to make sure we talk about, or anyone in the audience or anyone that wants to hop up and and have any questions or anything like that, now is a great time to do it. See there is a gentleman from from Wisconsin in in the audience, Bitcoin Carpet Cleaner, that is a fire name, and also from Wisconsin, shout out.
Anyway, sorry. That was a side note. But, yeah, is there any other topic that people wanna make sure we cover? Otherwise, we're right about at the top of the hour.
David JohnstonJust real quick on the beta testers for install ubinklot.xyz. Just hit me up on x. Shoot me a a DM, and let me know what your email is, and I'll get you an invite. Cool. Alright. Yeah. Everyone do that. Guys from Jungle,
Bowen Belmerthank you for being here. We really, really appreciate when builders pop up, on the platform. I don't know if you've seen, but on the Brainpower YouTube channel, which is the podcast that David and I cohost, about DAI and the Morpheus ecosystem, we actually do a series called Built on Morpheus where we have builders come on and do screen shares and show, you know, like like show live well, it's not live. It's prerecorded. But but show in real time, you know, what they built and how it functions in practical terms. So would love to, you know, invite you guys to come do that with me.
JUNGLE | Agent OlympicsLet's do it. I'd say closer to the World Cup final there in the middle of July. I think that could be a great time. We'll have we'll have data on close to a 100 games at that point. And, obviously, everything around around the finals is gonna be pretty hot. So maybe we could do it do it around then. Let's chat. That'd be awesome. Thanks, guys. Cool. Thanks, folks.
David JohnstonThank you, guys. Keep being awesome.
Zion, hear me! For Season 2 of the Mere Morpheus podcast I'm going to be uploading the Morpheus builder weekly updates. My reason for doing this is to make it more accessible for those who want an update directly via...
Zion, hear me! For Season 2 of the Mere Morpheus podcast I'm going to be uploading the Morpheus builder weekly updates. My reason for doing this is to make it more accessible for those who want an update directly via...
Zion, hear me! For Season 2 of the Mere Morpheus podcast I'm going to be uploading the Morpheus builder weekly updates. My reason for doing this is to make it more accessible for those who want an update directly via...
Zion, hear me! For Season 2 of the Mere Morpheus podcast I'm going to be uploading the Morpheus builder weekly updates. My reason for doing this is to make it more accessible for those who want an update directly via...
Zion, hear me! For Season 2 of the Mere Morpheus podcast I'm going to be uploading the Morpheus builder weekly updates. My reason for doing this is to make it more accessible for those who want an update directly via...