From Adaptive Intelligence to Decentralized Skies at Korea Blockchain Week 2025

‍

On this episode of The Edge of Show, co-host Josh Kriger dives deep into the emerging AI agent economy with two very different but complementary builders: Shaw Walters, founder of Eliza Labs / ElizaOS, and Robin Wingardh, co-founder of Wingbits.

First, Josh and Shaw unpack how adaptive intelligence, generative token networks, and autonomous swarms are redefining what an AI agent economy looks like in practice—moving from buzzwords to concrete use cases in DeFi, gaming, liquidity management, and multi-agent coordination. They explore how ElizaOS is building a modular framework where agents talk to each other, manage treasuries, navigate liquidity pools, and eventually transact value agent-to-agent across the AI agent economy.

Then Josh turns to Robin and Wingbits to show how the AI agent economy connects to the real world through Deepin. Wingbits uses blockchain incentives to crowdsource global aviation data, reward node operators, and power airline analytics, satellite validation, and AI-driven forecasting. Together, these conversations map a future where an AI agent economy bridges Web3, infrastructure, and real-world data in ways that are difficult to ignore.

Key Topics Covered

• Adaptive intelligence and the AI agent economy in Web3
  Shaw explains why he dislikes empty buzzwords yet still embraces “adaptive intelligence” to describe how dynamic, non-deterministic systems underpin the AI agent economy. He breaks down how agents learn, adapt, and interact with DeFi and games, rather than just repeating fixed, brittle logic.
  
  
• From fixed tokenomics to generative token networks
  The conversation dives into why many projects suffer from broken token models and how a generative token network can respond to market conditions over time. Shaw describes using agents to interface with complex liquidity strategies, making the AI agent economy more resilient than static tokenomics.
  
  
• Agent-to-agent economies and autonomous assistants
  Josh and Shaw explore a future where agents schedule interviews, negotiate OTC deals, and manage workflows with minimal human intervention. They imagine an AI agent economy where your executive assistant becomes a superhuman conductor of multiple AIs, rather than being replaced outright.
  
  
• Agent-managed treasuries and DeFi education for non-quants
  Shaw talks about using agents as interpreters between sophisticated quant strategies and everyday users who don’t understand liquidity pools, yield strategies, or risk curves. In the AI agent economy, agents both operate strategies and explain them in human terms, closing the gap between pro tools and mainstream users.
  
  
• Swarm coordination, social graph intelligence, and private data
  ElizaOS experiments with agent swarms that live in Slack, Discord, and Telegram, harvesting signals from private social graphs. This illustrates how the AI agent economy can leverage collective intelligence, reputation, and “who to trust” data to make better trading and coordination decisions.
  
  
• Reinforcement learning and modular AI instead of massive monoliths
  Instead of trying to decentralize training for giant models, Shaw highlights a modular approach: use reinforcement learning to improve agent trajectories and policies on top of existing LLMs. This lets the AI agent economy evolve based on what actually works in real workflows, not just bigger models.
  
  
• Deepin and the AI agent economy in aviation with Wingbits
  Robin explains how Wingbits crowdsources ADS-B flight tracking data from thousands of hardware stations worldwide. Token incentives and on-chain infrastructure align contributors and customers, making the aviation part of the AI agent economy more secure, validated, and globally distributed.
  
  
• Token utility and real-world data monetization in Deepin
  The Wingbits token unlocks access to aviation data that can power flight tracking apps, airline analytics, M&A forecasting, drone intelligence, and financial modeling. Robin shows how this data-driven AI agent economy creates win-wins for node operators, airlines, and AI builders.
  
  
• Hardware, security, and validation for mission-critical Deepin networks
  After starting with Raspberry Pi-style kits, Wingbits shifted to secure custom hardware for better validation and network integrity. That choice ensures that data feeding the AI agent economy in aviation is robust enough for major airlines, universities, and satellite partners.
  
  

Episode Highlights

• “We’re moving from static code to dynamic computing systems that don’t always behave the same way. That’s the foundation of an AI agent economy that can learn, adapt, and still be trusted.” – Shaw Walters
  
  
• “Most people don’t understand liquidity pools or delta-neutral strategies, and that’s okay. Agents can be the interface—explaining the trade-offs while quietly managing the complexity of the AI agent economy in the background.” – Shaw Walters
  
  
• “Gaming has always been where the future shows up first. We’re using games and coordination experiments to safely prove out swarms, treasury agents, and the AI agent economy before there are millions of dollars at stake.” – Shaw Walters
  
  
• “Wingbits started when we realized a $6 trillion aviation industry relies on volunteer-collected flight data. With blockchain incentives, we can share value with the community and plug that data straight into the AI agent economy.” – Robin Wingardh
  
  
• “We’re already seeing airlines, aerospace companies, and financial institutions use our data. As we reach global coverage, the aviation side of the AI agent economy gets more powerful—because better data means better decisions.” – Robin Wingardh
  
  

People and Resources Mentioned

• Josh Kriger
  
  
• Shaw Walters
  
  
• Eliza Labs / ElizaOS
  
  
• Wingbits
  
  
• Alex (Wingbits co-founder)
  
  
• Korean Air
  
  
• Spire Global
  
  
• ADS-B Exchange
  
  
• Deepin (Decentralized Physical Infrastructure)
  
  
• Farcaster
  
  
• Telegram
  
  
• Discord
  
  
• Ethereum Foundation
  
  

About Our Guest

Shaw Walters – Founder, Eliza Labs / ElizaOS
Shaw Walters is the founder of Eliza Labs, the team behind ElizaOS, an open-source multi-agent framework powering the next wave of the AI agent economy. With a background in hardcore programming and systems design, Shaw focuses on building adaptive, non-deterministic agents that can coordinate in swarms, manage DeFi strategies, and interface with real-world workflows. ElizaOS emphasizes modularity, composable primitives, and reinforcement-learning-driven improvement over monolithic “one-model-rules-all” approaches. Shaw is also an outspoken advocate for decentralization, open tooling, and user-owned infrastructure, pushing for an AI agent economy where agents, data, and coordination logic remain transparent and collectively governed.

Robin Wingardh – Co-Founder, Wingbits
Robin Wingardh is the co-founder of Wingbits, a Deepin aviation network that turns global flight tracking into a community-powered data layer for the AI agent economy. With a successful background in Web2 startups, Robin saw an opportunity when a volunteer-driven ADS-B network lost a huge share of its data after acquisition. Wingbits now operates thousands of secure hardware stations across more than a hundred countries, delivering low-latency, validated aviation data to airlines, aerospace companies, drone innovators, and financial institutions. By combining token incentives, satellite validation, and AI-ready analytics, Robin is helping transform aviation data into an on-chain asset class that fuels smarter decisions and next-gen applications in the AI agent economy.

Guest Contacts

Shaw Walters

LinkedIn Link
https://www.linkedin.com/company/elizalabs

Website Link
https://docs.elizaos.ai

Twitter Link
N/A – currently active on Farcaster: https://farcaster.xyz/shawmakesmagic

Robin Wingardh

Website Link
https://wingbits.com

Twitter Link
https://x.com/wingbits

‍

Transcript:

‍

Intro/Outro: Welcome to The Edge of Show, your gateway to the Web3 revolution. We explore the cutting edge of blockchain, cryptocurrency, NFTs, ordinals, DeFi, gaming and entertainment, plus how AI is reshaping our digital future. Join us as we bring you visionaries and disruptors pushing boundaries in this digital renaissance. This show is for the dreamers, disruptors, and doers that are pumped about where innovation meets culture. This is where the future begins.

Josh Kriger: Hi everyone, Josh Krieger here live at Korea Blockchain Week 2025. And I'm so honored to have as my guest right now, Shaw Walters, who's the founder of Eliza Labs. Great to have you on the show. Yeah. Great to have you again. You know, we met and ended up talking for a really long time, I guess at Consensus, if I recall. And I was like, I was really excited that we could finally actually have an official appearance today. I know you went back into hibernation and been building for a few months, which is probably like a few years. So we'll have to catch up on what's been going on in your world. And for the audience at home, both of us are suffering from major debt lags. So not financial advice and not responsible for anything that we say at this moment based on late deprivation. We got to Korea, I got to Seoul Sunday, but like last night got up like four times and my Aura said I only slept three and a half hours.

Shaw Walters: I landed, we went straight into the events. I saw you last night, I was vibrating a little bit.

Josh Kriger: At any rate, your team has been sort of leading a conversation on our new approach to sort of AI and crypto called Adaptive Intelligence. Can you tell us more about what that means?

Shaw Walters: I'm not so big into buzzwords and stuff like that. I think we're surrounded by that. So I really try and think about, well, there's what we're doing, and then how do we communicate that to people so they understand it? Because my day is mostly just typing symbols into a computer until things happen. I am just programming all day. And so to try and explain that to people, especially in crypto, like, what is that actually? What the hell are you talking about? So we try to... I'm sure in this interview we will go over a few other things that kind of personally make me cringe a little bit.

Josh Kriger: But I appreciate the packaging, you know, language in a way that's more contextual for the mainstream. I totally agree.

Shaw Walters: As long as people understand that science and science communication aren't always the exact same thing. So we're in a world now where we can build dynamic computing systems at the cost of, they don't always do the same thing every time. They are non-deterministic, which is a new way of programming things. It's more like telling somebody who I just picked up off the street to do a job. I'm paying him well, I can trust he's going to do it, but is he going to do it well every time? It's very complicated. So obviously, Like this, like what we're doing now is like, well, how do we bring AI into DeFi in a way that makes sense? Some of it is very LARPy, I would say. Some of it makes a lot of sense. And there's like definitely a frontier of stuff that I think that we are moving into that will be profitable and like long term has a lot of interesting ramifications.

Josh Kriger: So one of the concepts that you guys have been talking about is this idea of a generative token network as opposed to fixed tokenomics. I want to look at anything that may fix the tokenomics situation because I can't tell you how many like ambitious incredible founders have come on my show in the last year or two and had challenges with their with their tokens. I'm one of them.

Shaw Walters: You know, having a billion dollar token with like two million in liquidity or something like that.

Josh Kriger: Like things run afoul and in a way that you try to mitigate all those risks up front, but you can't, right? So I guess your own experience and sort of seeing what's happening in the industry has led to this new concept. Tell us a little bit more about what it means.

Shaw Walters: Well, so there's a bigger picture here of like what does the future of like an AI enabled network look like? And it's probably like, hey, I need something. You have this crazy, intelligent API. We have some sort of distributed way of like my agent finding your agent that's offering that service. And without me ever even talking to you, I have all this great information and you have all this money. Right, so we have basically automated the distribution of like a lot of the stuff that we might take for granted in the world, like APIs and stuff is now part of this distributed thing. So there's a spec called ERC 8004. This is like a thing that just come out from the Ethereum Foundation, something we're adopting. And the idea of this spec is that we can have agents that discover other agents. What do we do with that? Well, we're doing games and these sort of swarm experiments, but the obvious future of this is that most money will probably be transacted from agent to agent for services that, for me, is just a very clean experience of, oh, I need a thing. Oh, I'll ask chatGBT or whatever that experience is. But behind the scenes is a whole network, many different actors connecting to provide services and stuff.

Josh Kriger: Do you envision this idea of an agent-to-agent economy where instead of me talking to your team to book this interview, my agent reaches out to your agent and everything just happens?

Shaw Walters: In the thousand-year timescale, absolutely. In the really near timescale, I think it's going to happen in bits and pieces to kind of get there. And that's going to be like, well, I really like my executive assistant. I want to keep him as long as I can. But at some point, he's probably going to get replaced by an AI. And you're going to be like, wow, I've got this great AI executive assistant. It only cost me $1,000 less than the human. And so then at that point, we're probably going to do that. Yeah. And so we Like, my take on this is that this is still, like, really early days. And so, personally with us, what we're experimenting with a lot is, like, how do we build video games out of this? How do we build experiences? And another thing that we're experimenting with that's kind of along the same lines, we have a new project we're about to launch next month. It's kind of an experiment where you can buy tokens OTC from an agent and you can negotiate them. So it's kind of like, I don't know if you've ever been in like a Telegram chat, like, oh, I want 25% off spot right now by, you know, $100,000 worth. It's kind of like how the whole industry works is it's like really in the Telegram chat. We're like, well, what if we made that a game for people? So, yeah. So, so these kinds of things of like, I, I would expect that everyone will figure out how to prompt inject and like get the best deal possible from our agent. We sort of baked that in, but it's still a pretty good deal. So I think people will try it. But we just wanted to see how will people, what would happen? I'm willing to risk a million dollars to see what will happen in this interesting sort of experiment, for example.

Josh Kriger: That's very cool. And just on the point of executive assistants, I think for those out there, I would say before you consider how to replace your executive assistant, what can you do to train your executive assistant to be an enabler of this AI economy, right? So I have everyone on my team using AI, and I think if you can take a human-first approach to the future, think about what an executive assistant can do with the power of AI to assist them.

Shaw Walters: I really think if it's not like, oh, do I have five engineers or five AIs? It's like, do I have five people who are managing each many different AIs on their own?

Josh Kriger: Exactly. It gives anyone the ability to be superhuman.

Shaw Walters: Yeah, yeah, for sure.

Josh Kriger: Pretty cool stuff. So one of your posts recently described an agent managed treasury that sort of learns from market cycles. That sounds like a little bit sci-fi.

Shaw Walters: This kind of falls into that like taking a pretty complex, broad thing and trying to explain it simply and succinctly. We have a couple of different approaches to this. Mostly the stuff that's happening in this kind of finance is not particularly exciting code, because in quantitative finance, people have figured out like all the different kinds of like delta neutral trading strategy. And I want, you know, That's all worked out like there are programs and software for that. Everyone has it, but it's not accessible to most people and it's not like you can't really see what it's doing. And so the agent is like a way as an interface between like some crazy rule based engine I have here managing liquidity pools and the average user who's like, what the hell is a liquidity pool? Right. And so that comes with being able to explain what it is.

Josh Kriger: Someone in my dream asked that recently. What is a liquidity pool?

Shaw Walters: A lot of devs who have tokens out there do not know what liquidity pools are and And I think it's not their fault. It's like a pretty complex thing. And even if you kind of vaguely understand it, the math of all this is very complicated. And I think that where agents can help is to be like, let me explain this to you. You're going to go to Radium and you're going to look at like, well, how do I set my range? And you're going to just leave. You're just going to be like, I give up. And so an agent can be like, well, let me explain to you what a good range should be. This is what I think. And then it can also do things like just have a fully automated LP management system in the back that's like normal program. contracts even, smart contracts. And then the agent is just kind of like interfacing you to that. I think that's a more common thing.

Josh Kriger: Well, let's play this out a little bit because at the end of the day, agents seem to be autonomous. They need their own wallets, I guess, and they need on-chain agency to move around and do cool things, to build things, right? So how is ElizaOS enabling that?

Shaw Walters: Yeah, I mean, That's definitely the obvious for our future and how we get there is like, well, how do I get an agent that I can trust that's not going to hallucinate or do something stupid or send John the money instead of Joe? You know, it's so many things.

Josh Kriger: Or like, you know, hold someone ransom.

Shaw Walters: Yeah, yeah. So ideally, that's the really scary. So what we're starting with is like, what would demonstrate the full loop of an agent in an agent to agent economy with lower stakes? And I personally, I think gaming is a big part of this for us. It's like, can we just build a game that would demonstrate the entire loop? And it's like all the things, the components you would need for the full DeFi economy. It's just lower stakes. It's just like you don't lose a million dollars when you mess up. Um, and we need, like, a period, I think, of play and fun and, like, experimentation with this tech until we've got the, like, OK, I trust it. And it's not like we're just, like, messing around. We actually, like, we can collect data and we can start to train trajectories of agents now using reinforcement learning. And so if we can, like, collect, oh, this agent's really good at playing this DeFi game. It starts to get me in the direction of like, I think it's just be good at making me money, period. Obviously, the future we want is I want to be like, make me money. Here's a wallet. Here's a thousand bucks. I want ten thousand bucks next month. And if the agent does that, incredible. And I think when's that coming? Obviously, efficient market, blah, blah, blah. Maybe, you know, if everyone has that, it's complicated. But generally, I want to be the I'm raising my hand. I will. I will be the beta test. Yeah. Right. Right. I don't know. I think there's a lot of still hard, unknown problems. And I'm interested in creating the framework to allow people to tackle those problems. And then we're also like, can we create experiences that kind of bring everyone together to do that together? Cool.

Josh Kriger: Well, that will be exciting when that happens. And speaking of autonomy, you also have spoken about the idea of deploying autonomous swarms. Yeah. And I guess this is sort of a way of changing how organizations operate, looking at sort of DAOs and startups and traditional enterprises differently. Can you speak to it a little bit more? Yeah. Yeah. I think all those models have challenges, right? Yeah. Oh, yeah. Let's take a pause to shout out one of our favorite partners. For tech innovators facing legal challenges, Zubaloader is your go-to law firm. They focus on understanding your technology and business model before addressing legal requirements. Specializing in blockchain, AI, VR, AR, quantum computing, and more, ZuberLawler offers expert guidance in capital raising, IP transactions, M&A, litigation, and compliance. Visit ZuberLawler.com, that's Z-U-B-E-R-L-A-W-L-E-R.com for cutting edge legal solutions.

Shaw Walters: So there's just like, I want agents in my Slack that help me with stuff. And I could call it a swarm. I could have like an agent for social media management. I've got the intern doesn't, I don't give the intern the password, but I let them like tell the agent to post stuff. And now I've got like a nice security model, right? Where I don't have, you know, or I can let the community engage. Maybe I have the community manager and I have like kind of a very like I don't know what we're calling the org is like our test of like a business swarm like agents that interlock we kind of serve like very basic obvious needs for you or me or anyone has a community in the slack. But then there's like the okay well what happens when we have like a swarm of like agents coordinating and participating like you have your five agents I've got five agents we got you know many companies and we all need to like do business and make money again. And a thousand years, very obvious, like definitely going to happen. How do we get there? And so again, where we, Eliza is at the core of multi-agent framework, which means you can launch a hundred agents and have them all talk to each other in a Discord channel with no, you know, that's just out of the box. So what's the world look like? What can we build to get us to that world where everyone's there? And I'm really convinced this is like, right now we're working on coordination games, working on this kind of just early, like what's the fun, what's the exciting, like, thing that catches people's like, 12 year old, oh, that's fun. That's cool. And crypto is great for incentivizing a lot of this behavior. And the difference between an in game asset and the token is very, you know, very minimal. So I think we're like, I'm, I'm personally just thinking about like, well, where can we go that's low stakes that today we can demonstrate all the stuff we're talking swarms and all this stuff. But that's not gonna like, be like kind of unreal. I don't know. There's definitely like a sci-fi element, but that sci-fi element comes from the expectation that this is how our world is all going to work. It's not like the technology works. I have a swarm of agents on my computer that are yapping at each other. Are they making me a ton of money? No. Should they be? Yeah, like, let's get there. What's necessary for that? And I think there are, like, obviously give them wallets, give them money.

Josh Kriger: If you said, hey, guys, get together and figure this out.

Shaw Walters: Well, this is yeah, I mean, this is what we're doing. And we've done a few of these kinds of things. We have like the OTC agent. We have an agent. We did a thing called the marketplace of trust. We made a plugin and an agent for this. And it just looks at everyone shilling each other tokens. And it goes like, well, if I just listened to all of them, like who would have actually made me money? And so it's very different than an agent, like deciding what to buy. It's actually just extracting all of their buys and then turning it into a spreadsheet, going back, looking at the actual numbers, checking the market and be like, oh, I would have, if I just listened to this guy, I would have gotten rich. So now he has a very high trust score. And so I think in those ways we can use these agents today to do like, interesting swarm like or like human and agent collaboration stuff without thinking like, oh, it's going to trade for me because it's like it doesn't have any more information than I do. If it could extract the insider information from like Ansem on Twitter, that'd be great. That's like a really good win. And so we see the agents as being very powerful where they can like get information from from social media or in these like large open social graph things. And that to me sounds like it starts to become very swarm like. Another issue is like right now, it's very hard There's this idea that all of the data is on the web and you just have to go scrape it. And most of your data is in your phone and in a private network and Telegram and Messenger and Discord and these places. Most of the data most of us create is in private networks. And agents can go to private networks and they can collect that information. So you could have a trading bot. All the trading bots you see are also in a thousand Telegram chats and they are collecting that information. And so for an agent, that becomes an extremely useful place to be. And then they can share that information and they can start to derive financial insights at scale, you know, because now you have like, you know, in crypto, crypto is very small. Like if you know, if someone's about to buy a token, like that is incredible. Yeah.

Josh Kriger: No, I mean, I think it's, it's all about being first and like, you know, veering away from those Telegram groups where they do a big buy and then they tell everyone else to buy and then they don't. Yeah, yeah, yeah.

Shaw Walters: But if you can shock that guy, and you know he's going to dump on you because you saw the last 10 times he did it.

Josh Kriger: It's very powerful. Then you can make informed decisions, which is what I think everyone wants to do.

Shaw Walters: That's the hard part of agents is making informed decisions. If I can be like, hey, go look at every news article that came today. And if there's anything about crypto that seems bullish, buy Bitcoin. Yeah, that's pretty doable. That's actually really doable. That idea of turning an unstructured article into an action is very doable. Why? Like, you know, it's not like the quality of that buy is still like a very TBD and it really comes down to what information can you get from where else, you know, so.

Josh Kriger: Makes sense. So you guys have sort of bet on this idea of modularity as opposed to using these big models. Is that sort of accurate? And I guess what do you see as the benefits of like composable primitives relative to like the big model philosophy that is erased right now. Yeah. No one seems to be able to win. They just kind of go in toe to toe.

Shaw Walters: Yeah. And I think that that's actually really good for us because it means they all have to basically get cheaper to get us to use them. Yeah. I'm not complaining. Yeah.

Josh Kriger: It's just, you know, I use one of these services and it's better today than it was a month ago.

Shaw Walters: Yeah. And I find I was going from like, I will just pay whatever for the best model to like Gemini. It's really cheap. I should just use Gemini.

Josh Kriger: You know, like now I'm more concerned about cost than quality for the first time because the quality has gone up across the board to a point where it's like, I guess it's the 80 20 rule or even 99 percent.

Shaw Walters: as good it's it's more like taste now it's like claw just makes too many new files and like adds try catches you know uh gemini is good but sometimes it just gives up and defeat and deletes all its files what do you think about perplexity i don't it's not really good for coding but for information and like search it's the best i mean it does it for yeah for for like it's my new concierge you know yeah I mean, and also I think they're hinting at like a future when we talk about agents and swarms. Like, well, what if your complexity thing has a bunch of connections to it? And it's like, oh yeah, that API, let's go hit that. It's an MCP server. So this idea of agent to agent is really enabled by, we have this new thing called MCP, which is like a protocol for any agent to just call any API. And most things are getting kind of MCP. So this is like helping us a lot with that. That's already like kind of happening. Oh man, this is like very far off from the question though. Where do we start with this?

Josh Kriger: We were just, I don't know, we were just talking about, this is where that jet lag hit. It just hit us. We warned you guys it was coming. No, I think it's all in flow. Yeah. Let's sort of take a step back. You know, there's been a pretty strong narrative around this Web3 AI convergence. What do you think is the most misunderstood component of that narrative, where people are looking over there and they should be looking somewhere else?

Shaw Walters: Oh, right. And this actually kind of ties into, now we were talking about modularity. Yeah. So this is, I think the biggest trick is going to be like around inference and like right now, if you wanted to have decentralized, like if I want an agent that runs somewhere anonymously and does a thing, cool. If I want to train a new model, that is a really hard problem. And I think that there is definitely a lot of people in this space who are like, yeah, we'll just do decentralized training. There are a lot of, like, physics problems to, like, you are over there and I am over here. And this traditional LLM style of, we'll get 10,000 GPUs, put them in a thing, get a nuclear power plant and do it, you know, the Grok thing or whatever. Um, is, is like just not feasible. I just don't think it's the right direction. However, there's this kind of interesting new thing, um, and it's reinforcement learning. So when, uh, when chapter BT came out, they had this idea of RLHF. I'm going to show you two versions of a thing, and you're going to tell me which one you like better. And it creates a preference model. I don't know what's good or bad. I just know what's better and worse. And I can create the secondary model that allows me to make my whole system better. And this was like the insight that allowed Chachapiti to be huge. Then DeepSeek comes out and they change the game. They say, well, screw humans. What if we just have it train on its own reinforcement learning kind of approach? And we just say, well, we know that this trajectory and attempt was better than this one. Or it just thought for longer here and got a better answer than it here where it thought for much shorter. And so now they're starting to train like, well, I wanted to learn more and I want to be able to judge what better trajectories were and that stuff can all be done on the agent can do that with its own compute and can start to then collect and share like. Oh, this is like an actual meaningful way to get smarter and better. I don't have to train the whole freaking LLM. I just have to train the reinforcement learning policy optimization at the end, which is kind of like guiding the LLM toward more the direction I wanted to go. And this doesn't mean like the kind of information I need to collect is less like all of the Internet and more. How did my agent do? And that information is really lacking in our world. And the reason agents suck today is because we do not, no one's writing down, this is my job. This is how I do my job. This is step one and step two. So it just kind of has to guess it like. it's kind of LARPing as your executives. It doesn't, it's never actually like read how an executive assistant does his job every day. That's kind of something you just like learn. And, and so if we could actually quantify that and say, well now we have a million trajectories of a guy trying to help you out and some of them are really good and some are really bad. I'm going to have another LLM judge which ones are good and bad so that you and I don't even have to look at any of this. This is all automated. Now we start to have a world where like I can contribute, you can contribute. We all have meaningful data that can come together to make, like a model of an agent that's much, much smarter. And so I think this is where decentralization and token incentivization can actually play in the space. Not, I do not want to train GPT-6. I just think that's like the wrong, it's just, it's crazy expensive. And if they're not all co-located, you don't get the like super fast GPU interconnects. Now you're like on the internet and it's just like, it's like a hundred times slower. And there's a lot of research that's made it faster and there's federated learning, blah, blah, blah, but like, This new approach is just way better for us.

Josh Kriger: Very cool. So I want to respect your time, but I have one additional question I wanted to ask. You know, you're pretty much an open book on on acts. You're sharing your thoughts. Well, I was when I got banned. Oh, what?

Shaw Walters: Oh, yeah. Oh, yeah. We're suing them. That's the whole thing. OK. All right. Well, before I go, I'm on Farcaster now.

Josh Kriger: Yeah. OK. So what happened there?

Shaw Walters: Uh, well, I made a bunch of open source code and a bunch of people use the open source code and X. Uh, I was talking to that because I reached out to them. I was like, Hey, we have a problem. Like, regardless of what I do, like you are about to kill the, like the internet's about to dead internet theory itself and everything is going to be AI generated. And if you go on TikTok, it's all AI generated now, right? And I want to be part of solving that. I don't just want to put a bunch of... My goal was not to say, like, hey, let's make a thing where people can shill meme coins on X. That's what happened. And in response, I was like, well, how would we make a world where that isn't just the default? We want social media we can use. So I reached out to them. We kind of got in touch. And I went to their HQ and I was like, cool, I want to advise you. And then they kind of looked at our thing like, yeah, so just give us $50,000 a month for an enterprise license, and then we won't delete your accounts. And I was like, what are you talking about? We had a very different conversation, and now suddenly you're threatening me legally if I don't pay you money? That's just extortion. That should be illegal. So I'm definitely down to try that in court. Because it's not like they reached me out of the blue. I was in their HQ. I was talking to them about the whole thing. And they basically looked at us like, oh, so you're like virtuals. You're like all these others. And to be clear, we don't host agents. We don't sell anything. We don't have a product you can buy. It's all open source. It's just open source. And so they basically just deleted me from X for making an open source thing that people use and not paying a licensing fee. that I wouldn't use. What would I use the enterprise license for? I don't sell anything. I don't have a product. You can't sign up for anything. There's no website to go to to log in. You would have to get your own API keys and sign up for yourself if you want to launch your agent. So I'm like, we're not doing anything.

Josh Kriger: Sounds like a misunderstanding there. It's a misunderstanding.

Shaw Walters: And so we kind of got through with their legal team. And we're like, yeah, I think it's all cleared up. You're going to give us our accounts back, right? And then they just cut us off, stopped responding to our emails. And so we're like, all right, well, we got to see you. That's how it is. So I'm just hoping that they just come to their senses and realize we've brought them lots of business, lots of developers, lots of money. We've given them $25,000 this year, but it wasn't enough. So they're just shooting themselves in the foot. So I'm on Farcaster now. Does decentralization matter? Like, if it doesn't matter, then let's just give the social media network owned by one guy all of our money. But if it matters, then why are we on X? It's literally the opposite of why we're all here, isn't it? It's because I can just be banned for making software. That's not the world I want to live in. I'm not going back to X. No, I don't want to do that. I don't participate in that. I don't believe in that.

Josh Kriger: Wow, well, I didn't have all that 4-1-1 before the interview, and I appreciate your candor with your perspective on it, and I will have to keep following up on that. What is your sort of most recent bold insight, maybe in your jet-lagged state or on the plane over here, or, you know, just in the last week or so, what was your deepest insight around what you're doing and where things are going that you hadn't had before?

Shaw Walters: I don't know if I've had this before, but something I'm thinking about a lot is like, where did AI come from? Carmack makes Doom, and then he makes Quake. And Jensen Wong is like, dude, I'm going to make some specialized cards to help your Quake games look really good. And they're going to be parallelized. And then they just keep doing it in a loop, games and graphics cards, games and graphics cards, until pretty soon, this can run AI. And early 2000s, we're like, wait. Mid 2000s, we're like, oh, no, you need this. And now we're pretty much to the point where NVIDIA runs the entire AI world. So all of that came from video games. And I think a lot about OpenAI's start. OpenAI has been around for 10 years, but ChatGPT has only been around for four years, or like three, right? What were they doing? Well, they were playing Dota. They made a Dota agent that was the best Dota player. They made a Rubik's Cube handsover.

Josh Kriger: So gaming, gaming is always that early pioneer.

Shaw Walters: I think about Claude Anthropic, Dario Amadei, game designer, comes from gaming. What does he do? He starts a company, DeepMind, game designer. I might have this wrong, I think Dario is actually a physicist, but Dennis from DeepMind is a game designer, and all they do is play games for years. Let's beat Go, let's beat Chess. and they're like oh protein folding let's just take the go algorithm and just throw it at protein folding salt science solved you know tens of years of science no problem throw in some second life there yeah and so i think you got the future where can we prove like overwhelming success where the stakes are low And I think in crypto and AI, it's like right now, I think it's gaming and social experiences and small things that like prove the hard problems in a low risk way so that when a million dollars is on the line, you're like, yeah, no, we already solved trust. That's done.

Josh Kriger: So, Shaw, this has been incredible. If you want to follow Eliza Labs, do not go to X. Farcaster.com slash Eliza OS.

Shaw Walters: And then your website. The other important thing is, you know, we're launching a new token. It's the same token. We're relaunching it, so it'll be on all the chains. You'll be able to get it on Ethereum. We'll be doing some really cool stuff, I think, with Ethereum Foundation. I can't announce it yet, but I'm pretty excited for that, independent of the token. But yeah, there'll be a lot coming from us next month, a lot of announcements.

Josh Kriger: All right, exciting stuff. Thanks for hanging out today. I'd love to do a Joe Rogan-style three-hour interview one of these days. I totally would, you know, like six hours, Les Brimden-style. I'm down, I'm down. All right, cheers. All right, take care.

Ad: Hey, listeners, are you ready to amplify your voice in a digital world? Introducing .podcast domains, your gateway to the future of online identity. Whether you're a creator, a company, or just someone with big ideas, .podcast is for you. Own your unique digital identity, connect securely, and explore hundreds of apps with no renewal fees. Don't wait for someone else to claim your name. Secure your .podcast domain today and be part of the next generation of content creation. Visit get.unstoppabledomains.com backslash podcast to get started. Again, that's get.unstoppabledomains.com backslash podcast. Your voice, your rules, your identity.

Josh Kriger: Hi everyone, welcome back to the Edge of Show. I'm Josh Trigger, your co-host. We're live at Career Blockchain Week and I'm really excited to be here with Robin Wingard, the co-founder of Wingbits. Yeah, really happy to be here. And, you know, we talk about a lot of fun stuff on the show and when I heard about what you guys are doing, I was definitely intrigued. So excited to unpack this a little bit more. You guys started back in December of 2023 and maybe you can share sort of the genesis story of Wingbits and sort of maybe what some of your insights from the Web2 companies that you created before this and how they sort of impacted your thesis.

Robin Wingardh: Yeah, for sure. So kind of how the idea came about was it was actually from a podcast. So we don't have background in... I love when you're doing a podcast.

Josh Kriger: You've got to listen to podcasts.

Robin Wingardh: Exactly. Now, because like neither me nor my co-founder Alex have background really in Web3 or aviation. But we heard this podcast about a smaller flight tracking network called ADS-B Exchange that was acquired. And within two hours, they had lost a really big portion of the data. And when we started diving into how the industry works, we realized that you have a $6 trillion aviation industry that's relying on a data point called ADS-B, which is the flight tracking data. It's all volunteer collected, and they essentially do it for fun. And these companies then that get the data make hundreds of millions of dollars. And we thought, well, there has to be a better way of doing it where we can essentially create a way to share the value created with the community. And when we started talking to customers and validating the case, it just proved that utilizing the blockchain is the best way to solving this problem. So that's how it started. So it was just podcast.

Josh Kriger: Yeah, I mean, democratization of data and sort of a shared economy, that's the essence of, like, blockchain. So were you familiar with the term, like, Deepin, decentralized physical infrastructure, or was this just sort of an independent thought that you guys had?

Robin Wingardh: It was just an independent thought. We actually didn't know really what Deepin was until after we started. I've heard that a lot from folks that have built quote Deepin companies. Oh, yeah. I know. It felt like the term Deepin really took off a little bit after we started it, which was good for us because it got a lot of attention.

Josh Kriger: Cluster, it gave sort of some context to what you guys are doing and comparing it to folks like Helium and whatnot. It sort of provided proof of concept that you can do interesting things like this and create new economies. Exactly. So what's been your attraction so far? And tell us a little bit about sort of volume and geography and where you guys have gotten to.

Robin Wingardh: Yeah, so right now we're sitting at about 5,000 live stations in about 120 countries. I think we have about 3,500 more stations on backorder. We started like every other deep, and you see the natural growth is in Western Europe and the US. And then we saw massive growth in Latin America, and now we're heavily focusing on Southeast Asia and MENA region, because that's the last piece of coverage we need to reach global coverage. So we're growing pretty quickly. I think we're going about seven times faster than any traditional flight tracking network has ever done in the past. So the model works. In terms of customers, we announced, for example, our Korean Airlines partnership this morning. Congratulations. Thank you. Yeah, we've been working on that for a long time. So I'm happy we got to announce it this week.

Josh Kriger: Is that the first commercial airline?

Robin Wingardh: No, so we have two other ones, but we have an NDA, so I can't say those names. And then we also work with a company called Spire Global. They're an aerospace company, so they utilize our data. We also launched a satellite with them back in March of this year. So we use the satellite as a validator for the network. And then we have a few analytics companies, some drone companies, financial institutions that use our data. So it's now when we're reaching more coverage, it's growing pretty quick.

Josh Kriger: I mean, my hope is that you can reverse this trend of small and smaller seats on aircrafts with the monetization opportunity you're creating for these airlines, because these seats are getting tight, my friend.

Robin Wingardh: Yeah, I wish. I don't know if that's something we can do, but maybe if they use our data to become more profitable, they can increase the seat size. I don't know.

Josh Kriger: or upgrade the food, just a touch, please.

Robin Wingardh: We do have some cool stuff, though, with airlines. We're launching TGE this fall and there's a lot of cool, we're going to have a perks portal, which is going to be awesome. And there's a lot of airline partnerships there that you can actually use the tokens for stuff.

Josh Kriger: Oh, very cool. That makes sense. So what have been some of your key insights along the journey in terms of where you've got to now? We were talking about before the show, rethinking sort of the hardware side of things. Maybe you can elaborate on some of the lessons learned.

Robin Wingardh: Yeah, so on the hardware side, a little bit of a backstory, all the other networks function, they're decentralized, it's volunteer based and it's all Raspberry Pi based, so they build their own kits essentially. We started the same way because we thought that was an easy way to get started, but then the more we talked to customers, we realized that the existing networks had such poor security and validation on the hardware that it would actually be way more beneficial to create more custom and better and more secure hardware. So we made a decision which was actually a tough decision because we had grown to about 2000 devices before we did the switch. But now it's all done and it's proven to be the right choice. So that was definitely something that we had to figure out along the way. But in general, having no Web3 background, when we started, there was a lot of things we didn't know. You know, setting expectations for time to TG, really difficult. So we kept pushing that, mainly because we learned that the missing piece for pretty much every project is the demand side. And if you look at a traditional business, it can take five to eight years to really scale. Meanwhile, in Web3, they expect you to have a token in six months.

Josh Kriger: Well, kudos to you for having some patience around not launching too early. It's a mistake a lot of founders in our industry make. So not to knock on wood, but congrats on feeling like you're at the point where the equation makes more sense. And are there any unique aspects of your tokenomics utility side that you could point to?

Robin Wingardh: I mean the main use case of the token is actually accessing to build on top of the data and we're going to have multiple different data points in the future so you can actually build some really cool applications. So data access is the main thing. Right now there's no voting and stuff implemented yet. In the beginning, it's pretty centralized because we need the speed. What would someone potentially want to build using the data? I mean, there's a lot. I mean, obviously, you can build your own flight tracking apps. You can use the data for all kinds of algorithms to predict flight delays or just monitor performance of different airports, stuff like that. I mean, bigger institutions, for example, finance, they use the data to do M&A forecasting and stock movement prediction based on the performance of the airlines. airlines themselves use it for competitive intel to see how the competitors perform, to see what route they may want to compete on, also to spot how to optimize with turnaround times.

Josh Kriger: Why would they all want to opt in then? Wouldn't that create like a competitive asymmetry? Or don't they benefit from competitive asymmetry in terms of not everyone having that level of granularity of data on each other's airlines? Yeah, but I think everyone has it.

Robin Wingardh: And our data is just better. It's lower latency, more validated. So everyone needs the data. I mean, it's a $20 billion industry, stuff that's built on this data. And it's volunteer collected data.

Josh Kriger: So the cat's already out of the bag.

Robin Wingardh: Yeah. Yeah. But it's actually quite helpful for us because we're not having to re-educate them on something new. We're doing it a similar way, but it's just a better quality of data that they're already used to purchasing.

Josh Kriger: Cool. Does that sort of lead to some interesting potential partnerships with AI companies as well?

Robin Wingardh: Yeah. So we do have two companies that work within the AI space. One does AI solutions for one of the major airlines in Europe, and the other one does AI solutions for the drone industry. And we're also working internally on a lot of AI products at the moment that are quite exciting, actually.

Josh Kriger: Cool. Yeah. That leads me to my last question. Is there anything sort of on the roadmap that you're excited about and can share with our audience?

Robin Wingardh: I mean, I think we don't have the exact, I don't want to tell the exact date yet, but during Q4 is a mainnet launch, which is going to be really exciting. Obviously we had a Korean airlines this week. Last week we announced our Vietnam university partnership and we have, I think we have about 14 more things in the pipeline that are going to come out. So I don't want to spoil them, but it's a, I think it's going to be really exciting end of 2025.

Josh Kriger: All right, well, yeah, sounds exciting. And people want to fly with you guys. Where should they go?

Robin Wingardh: I mean, the best thing is wingbits.com. Start there. Join our Discord. Super supportive community. Super helpful. We have all our documentation. They can kind of learn about what we do, why we do it, and how to participate. And if you're in Indonesia, Japan, Korea, or Malaysia, we have a bunch of campaigns now for free devices that you can participate in. So check that out.

Josh Kriger: All right, pretty cool. And are you guys on X as well?

Robin Wingardh: Yeah, it's at Wingbits on X. And we have some telegram channels as well that are linked in our Discord. So make sure to use the official ones.

Josh Kriger: All right. Are you also on X?

Robin Wingardh: I'm on X, Robin at Wingbits.

Josh Kriger: All right, Robin, thanks so much for hanging out with us. It was fun. I appreciate it. Appreciate it.

Intro/Outro: We've reached the outer limit at the edge of show for today. Thanks for exploring with us. We have room for more adventurers on our starships, so invite your friends and cool strangers to join our journey. If you're among the hundreds of thousands following us on Spotify, iTunes, iHeartRadio, or watching us on Maiko or YouTube, please hit that subscribe button, rate us, and make an awesome comment too. And don't forget to pass this episode along to a friend or two who would benefit from it. Doubling back to again, recommend checking out myco.io, where you can watch The Edge of Show and earn for your time and attention. That's m-y-c-o dot i-o. Don't forget to visit theedgeofshow.com, the as part of their domain name, where you can learn more about collaborating with us, and also subscribe to The Edge of Weekly Newsletter for the latest Web3 news, events, and show drops. In addition, connect with us on all major social platforms by searching for The Edge of Show. Join the exciting conversations happening online. Lastly, be sure to tune in next time for more great Web3 and AI content. Until then, keep pushing the boundaries. Thanks again for hanging out with us today. The views and opinions expressed on The Edge of Show reflect solely those views and opinions of the show hosts and its guests. Please make sure to do your own research. Our show is not financial advice. You understand that you are using any and all information on or through this podcast at your own risk. Whenever making financial decisions, we recommend doing your own research and talking to your accountant for financial advice. From time to time, we may feature sponsored content on the show for which we receive value, and we may share links for which we receive a commission if you make a purchase through one of these links. Refer to our website, www.edgeofnft.com, for our full disclaimer, terms and conditions, and privacy policy.

‍