Full transcript — English (original)

Host?

Yes.

E- everybody's so quiet, so I thought maybe my audio is off.

No, no, it's all good. Sorry. Uh, we're-- I'm just setting up some more stuff, admitting people, and, uh, and then we'll start in a second. I'm a bit unprepared today, so bear with me. Okay. Um, okay. I think Participant B is still on letting people in duty, and, uh, and then I'm gonna start the recording now if it will work. And start to share my screen. So... Okay, fantastic. I think you're able to see my screen. The recording is running.

Um, people are slowly trickling in. So thank you everyone for, for joining. This is the, uh, third edition of our for bi-weekly or almost bi-weekly, um, AI on Cardano forum. We have, uh, a little bit of an agenda today. Um, it's not as much as we had last, last time, um, so I hope we'll be under the time limit this time. But, um, but it's gonna be quite interesting.

So I'm gonna start with, uh, just a little bit of recap, um, what happened in, in the last, uh, kind of meeting, and also about-- talk about some news, uh, what's happening in the agentic AI blockchain space. And then we have a special guest, uh, Speaker A, who is gonna talk about some, like, research and some discussion points he, uh, found for this general agentic x blockchain kind of landscape. Then I'm gonna, uh, give you some updates about Sokosumi. So I'll just show you the, the product and talk about what we've been building. And, um, and then we open the stage, and then we, uh, yeah, we end the meeting.

So without further ado, just to, to recap what this forum is about, uh, we have four, four goals. One of them is, uh, we want to just enable a sort of like platform to talk about AI, talk about, uh, standards, talk about working groups, um, and give everyone a chance to present stuff as well. So if you want to present something next week or in two weeks rather, um, just send me a message and I'll glad you- gladly, uh, book in a time slot for you and then give you the chance to talk about what you've been doing with AI and with, uh, Cardano and, and where do you-- where you see it going and so on.

Um, the second goal is we want to make sure that Cardano is very well positioned when it comes to, uh, blockchain AI. So we want to have it as the number one. I'm not the type of person who would, uh, ever settle for second best, so we need to aim for the top and really have that as our goal. And then the third goal is to foster more collaboration. So really, uh, yeah, just give a platform where everyone can talk and really, uh, discuss together what we need to do.

And then, um, as I, as I keep mentioning, this is not a, a platform for one single project, even though everything is in Masumi, uh, colors here. It's not about Masumi. Um, it's about anything, and any project can, can present, and it's really about just creating a sort of like group and, um, and a joint effort. And I know a lot of these, uh, kind of meetings feel like a presentation to many of you because that's kind of what they are. I have some slides here, and then other people are gonna present and so on. But it's really important that you speak up. And if you have any comments, just, you know, raise your hand. Um, I'll put you on, on speaker mode, and you can just talk about whatever you want. You can write it in the chat if you're too shy. It doesn't matter. Just participate and really use it as a forum and not as a, as educational material, basically. Not only as educational material.

Now, the, the first point, and I don't have a slide for this actually, the first point is, uh, news. So, uh, obviously in the last few weeks, a lot of stuff has happened, um, when it comes to this agentic commerce topic and the general like AI blockchain space. And the only, only thing that I found really noteworthy, and I just quickly want to highlight here, is, um, is that Cloudflare is now going into wallets, which is actually really crazy because, 'cause Cloudflare doesn't really have anything to do with, with crypto.

But they have been one of the, uh, creators of the x402 standard, so they were really interested in this whole like agent-to-agent payment space and agentic commerce space from the beginning. And it was pretty clear that they would at some point, um, do this step and actually release a wallet. And now they are rolling it out slowly. So, um, if you followed the news earlier this month, they announced, uh, an, a monetization gateway. So that basically means if you have an agent and you want to, uh, monetize the agent, then you can just use this monetization gateway and basically charge for it in, uh, using Cloudflare and x402 and receive stable coins. Which is quite, uh, crazy actually, because it simply-- Yeah, it's a huge provider, and it simply, um, opens up the doors to a lot of more agentic commerce.

And now with this announcement of the, of the wallets, um, they're basically coming from the other side. So they're not saying, "Okay, now we want to sell stuff." We want to enable people to actually buy stuff using agents and, um, and pay for those x402 endpoints. So they're really coming from both directions here, which I find interesting. So, um, what they've done is they, they kinda went the full crypto playbook, and they built this website here, cloudflare.pay, and you can reserve your wallet here. At the moment, you just enter your, your username, and then you get like, uh, a confirmation or something. It's, it's like a whitelist, basically, or waitlist rather, um, for the wallet.

But what I expect this to be is basically, um, a, a wallet that you can give to your agent which can handle stable coins. And then it's an easy way that whenever your agent encounters an x402 endpoint, that it just calls this wallet, which would be a CLI or a skill or just an API, we, we don't know yet, but probably a CLI, um, and then pay for that, for that kind of endpoint. And that's really, uh, interesting, and it is very similar to what we've been do-doing with, uh, with Masumi, of course. So this is extremely, uh, relevant for, for our work.

But Cloudflare are by no means the only ones who are doing it at this, um, base. So, so Coinbase is also doing this. Uh, many other providers are out there now. Um, Stripe, of course, is doing it. So many people are building these kind of like agent wallets, and I really think that this is the next big, big trend. Um, and we'll have to see where it goes. And I don't really see it as much of a, of competition at this point, even though it technically is. But I see it really as a validation for this thesis that in the future, there will be, you know, billions of agents doing transactions with each other, and the majority of kind of like money flow happening through agents that have access to these kinds of, uh, wallets. So, uh, yeah, very, very bullish on the topic.

Now, um, before-- Yeah. I-I'm kinda done with the news. This is the only thing that I wanted to talk about. But now I want to give it over to Speaker A, who, uh, has done some research and, um, I think he's not feeling super well today. So yeah, take it slow, Speaker A. But, uh, please go ahead and show us what you've been looking at.

Oh, yeah. Thanks, Host. Thanks for having me here. Um, uh, hi, everybody. Uh, my name is Speaker A. [A short exchange about the spelling of a name; not recorded here.] Um, uh, but yeah, thanks for having me here. Um, uh, so for those of you who don't know me, I work at one of Cardano's core development companies, um, uh, uh, on the product side. Um, just an enthusiast, like most of you guys, about what's happening in the agent space, blockchain space.

So, um, you know, I, I was just, uh, looking into it and, um, um, you know, and, uh, yeah, Host, uh, and I just, uh, thought it would be interesting to have a chat about this. I've not been feeling great today, so my slides are a bit, uh, not polished. Um, but my objective here is to more have a, um, discussion, right? So, uh, I'd like to just, uh, invite discourse on this topic. This is not meant to be a monologue, uh, because I'm not an expert on the subject, and I know only so much. Um, so it's, uh, um, more intended to, yeah, just, um, ha- Yeah, like have some slides and, um, go through... S-Sorry. Um, yeah, some, uh, some thoughts that I've had, and maybe you guys have thought about this as well, and we can, uh, um, yeah, just talk about it.

So yeah, you know, when you-- when we look at the agents meets blockchain space, this has been like a hyped topic the past, uh, year and a half, uh, or two years, right? Um, and, uh, a lot of companies have raised lots of money around this. But, uh, y- I was also curious, you know, where this actually meets reality and where it doesn't. So I also wanna play the devil's advocate in talking about this. And I'm sure people like, uh, Host and others who are deep into this domain and, uh, you know, very bullish on it will correct my understanding.

So yeah, just as an overview, there are like many segments in this, right? So there's the agentic commerce part, uh, which is what like Masumi does, which entails, you know, agents buying and selling services, data, and other things, um, on blockchain rails. Um, there's the agentic DeFi part, which, which is like agents actually trading, you know, and doing everything that you can do in trade in, in, in the DeFi world, but with, uh, with an autonomous aspect to it, so on behalf of users. Uh, there is the i-infrastructure, uh, and identity part, which you can skip, and the-- and more importantly, there's the machine-to-machine part, right? So I think there are three main segments here if you skip the infrastructure part.

Um, and the machine-to-machine part is also quite important because that entails, you know, like how, like, you know, um, IoT, uh, stuff and robots, et cetera, will pay each other, um, going forward on, uh, you know, um, with agents as the-- uh, with blockchain as the rails, essentially. So, um, these are the three segments, and, um, I think it's like, you know, the agent, the, the first two on the top that has kinda taken off right now, right? You know, the machine-to-machine part is still lagging. Um, do you guys have any, um, thoughts on about, on these segments, and what do you think is, uh, you know, what are you bullish about and, uh, what you're not, and like where you might have skepticism?

I'm, I'm just wondering, like, what's kind of the difference between agentic commerce and machine-to-machine? With machine-to-machine, you mean, like, robots, basically

Yeah, I think that's how it's kinda classified. Like, you know, it's, uh ... Yeah, exactly. Robots or, like, you know, um, uh, devices performing certain, um, um, tasks, right? So it could be, like, an autonomous car that uses, um, some service for itself, you know, uh, to, to run itself. Then it, you know, procures, like, whatever data from some other service which is, like, uh, which is served agentically, right? So that's what I would think, you know. And then going forward, it'll be, like, more other kinds of robots, like drones and other kind of things that would use, right? So I think that's, like, the projection, but that's, like, I think the, the, the, the area that is le- no- landed the least. Does that make sense?

I, I think it makes sense. I think it's, uh, it's interesting because in, in Germany we had this ... Or in general, we had this blockchain, IOTA, um, which was- yeah, a very German blockchain, and it was very focused on machine-to-machine payments, like, six years ago, and pitching this whole, like, you know, Bosch has a robot, and then the robot from Bosch is, like, paying the robot of one of its, uh, suppliers, and then pays one of the Mercedes robots, and so on. Um, and it never really took off, but maybe it was, in a way, too early for that. I'm not sure. Um, so, so I'm very skeptical about machine-to-machine, to be honest. Like, it makes sense in theory if you hear it, but in practice it's, it's not relevant anywhere yet, you know? So, yeah.

Yeah. Um, exactly, right. IOTA, I think it's pretty much died, and they've tried to revive themselves and stuff. Um, uh, but perhaps it's, like, a timing thing, that they were too early to the market, you know, and that kills the- projects, right? So, uh, yeah. Uh, but I've not looked into it much either. Um, I'm just curious if ... Thanks for pitching in on that. Um, and then yeah, I think the ...

Participant A?

Yeah. Hi, hi everyone. I'm Participant A. So I'm a recent, uh, new starter at one of Cardano's core development companies, um, in the research team. Um, an interesting subject which I don't see on here, which I think is, um, becoming more important, is security. Where do you see ... It's, it's almost like an arms race as well, 'cause as soon as you start releasing agents onto the network, um, as, um, adversarial, then you need to start releasing agents on there as well to protect the systems that you have. And so I'm wondering where, where that would fit in, and if you've done any thinking on that.

Um, oh, that's interesting. I've not really looked into it. And, uh ... So you mean, um ... Is it ... So how does it tie into, like, the, to, to blockchain per se? So you're just saying, like, you know, agents that are deployed out there, and then you have other agents somehow discovering these agents and then, like, scouting if they are, if they pose any threat, you know, doing an audit on them, something like that? Is, is, is that what you're thinking?

Yes. So maybe agents that are out there given a task to seek vulnerabilities in the system, um, you know, g- given certain levels of freedom to go out there and exploit vulnerabilities at different layers of the blockchain. And then, and then you see, um, well, we need to build defenses for that. So you can build hard defenses, but I s- expect that probably some of the best defenses are gonna be agents themselves programmed to defend and hunt out these other agents that are looking for these exploits, or to find the exploits and patch them before the agents are able to find them. I mean, I, I don't know. This is a, a completely new area to me, so it's, it's all a learning curve. But I'm, I'm kind of seeing something and, of an arms race happening, um, here, and I d- I don't know where it's gonna go. But I just wondered where, where it fitted in with your four categories.

No, that's interesting. Um, I think that's something that's, uh, you know, definitely gonna take off. But, uh, it's not in ... I've not scoped it out in these categories because I don't think that necessarily needs blockchain rails, you know, like, uh, as a coordination and orchestration medium. So I was trying to just see, like, um, agents on blockchain as a category, right? So, uh, those could be just, like, centrally deployed agents that, uh, kinda do whatever you said, auditing that they need to do, and, um, don't necessarily ... Yeah. But I, I, again, I don't know too much about that. So somebody wants to pitch in?

Participant B?

Yeah. I mean, to be honest, it, it doesn't need to be connected to blockchain, as you said, I think. Um, but it could fit honestly in, in most of those categories. You could have it as a commerce or, like, a commercial service, right? Um, but then also identity comes in there as well because, uh, of course you want to know, for example, if you disclose some information, for example, some, like, private information you want audited or something like that, you want to know who are you disclosing it to, right? So I think there are definitely use cases that it could fit in a lot of those things, but it could also be a standalone, um, uh, thing. And where- wherever it benefits, wherever it makes sense economically is, uh, where, where that could happen. But then on the other hand, um, I see it as, like, kind of a, like, maybe some security researcher or somebody building agents, right, and, uh, working on that. And, um, I mean, Fable is doing it or, like, Mythos and others. So that, that could come in. Um, the, the question is when it's more, like, as a service kind of a situation, I think.

Yeah. Yeah. Yeah, I think that makes sense. It could fit into any of these wherever it makes sense. Um, and yeah, about the blockchain part, right? Like, for all these categories, I think the blockchain part is also optional. It's not strictly required besides the d- agentic DeFi part, of course, because there's a blockchain underpinning there. And I think that we can discuss, like, going down, you know, what then ... What, like- How the landscape might evolve in terms of, like, centralized services versus, uh, things on blockchain grids.

So, uh, yeah, just going forward then that, that's like, yeah, the question that I was gonna ask, trying to ask, right? So why do you need blockchain for agents, right? So, so the, the main kind of, like, hypothesis that's being floated right now, you know, at least, uh, in terms of the narrative, I'm not sure how ... Um, I think many of these make sense, but they're probably not required everywhere.

So there's the payments part, you know, that agents need, um, you know, a very fast sub-cent, um, um, um, transfers and that like, you know, blockchain natively offers that because of its being, um, um, token native. Um, and then you have the trust problem, you know, then that, that's also, uh, handled well with the blockchain and, uh, contracts on, on, on chain. So, uh, and, you know, which begets things like identity and reputation, and you see that like with what, what Masumi does as well, right? So that's ... It, it, it provides like a decentralized, uh, trust, uh, uh, assurance, uh, um, platform, uh, by using a blockchain.

And then you have the, um, the, um, yeah, um, um, uh, the amount of ... You can control the amount of autonomy that it can offer to an agent by, uh, bounding how much it can spend and what it can do, et cetera, by using policies or other more, uh, on-chain, uh, enforcement. Um, and then of course you have accountability because you can use blockchain for like verifying what, what, what agents did along the pathway, right? So those to me seem the main reasons why blockchain is used for agents. Would you guys concur or would you challenge any of these? That's at least the running narrative.

No, I, I completely agree with all of these. Um, I think it's very interesting or, or ... Like the thing that happens with blockchain is that people try to fit blockchain onto everything and, and that happened like from the beginning, still happening. And I ... There's a lot of weird, uh, kind of blockchain applications that don't really make a lot of sense if you think about it. But people were just like- "Oh, there's a ... No one has done this before. Let's just fit blockchain onto it, raise an ICO, and like get a bunch of money to build it." Um, and that's the, the wrong way to approach it.

And I ... And, and blockchain is ... Like blockchain, we, we keep thinking about it in this like huge way, but it should be a tiny part of the application. Like it should be like 2% or something of the application in most cases at least. And, um, and ma- many get that wrong. And, and for agents, I, I completely agree with this list and that's basically what we're doing with Masumi that we focus on exactly those like four pillars and we try to ignore everything outside. Like I, I'm not a big proponent, um, of, for example, trying to decentralize like, uh, the training of AI models or something like that because, um, because it like in theory, yes it can work, but there's just economics that you have to fight here and, and then you end up in the best case scenario with a centralized kind of system again. So, so I'm a bit, uh, skeptical with all that and I, I just want to use blockchain for the things the way it actually makes sense.

Yeah, I think it's good, good to expound on that, right? That as you did. Thanks. I think that makes a lot of sense, uh, to me as well and that's why I wanted to kind of highlight this because it ... You know, blockchain washing is an unfortunate phenomenon that's happened, right? Especially during, uh, bull runs. Um, now it's kinda, uh, died out a little bit because the people have come to realize, uh, the pros and cons of having, uh, decentralized technologies. Um, so I think yeah, uh, but with agents it's, you know, like because the agents thing is a hype right now, there's a bit of like over in, um, doing things with blockchain and a- agents. But it, so yeah, I think it's kinda good to keep in mind where it can really make sense and, uh, have like a long-term product market fit.

Then going further, you know, I just wanted to ... I looked at like what the kind of market projections are, right? So if you look at like from the top, if you look at the, the market sizes that are being, um, uh, the addressable markets that are being talked about, for the whole agents world it's like in the order of, uh, uh, uh, agent software world is in the order of billions. For agentic commerce it's trillions, right? But if you look at different reports there, there's like a 10X vary, variance, right? So that basically is because of the hand-wavy way in which they do these kind of, uh, uh, um, market sizing.

Uh, but the reality is that today the daily volume on, on, on-chain agentic stuff is like 30K, so it's, it's pretty small. So it's still early days, right? So this whole thing of agentic commerce even, uh, is not yet validated. So it's early days. Uh, I, I mean, yeah, agentic commerce on chain is not yet validated. It's early days. So I think one has to keep that in mind, you know, um, uh, before like just, uh, falling for the, the, the, the hype around it. Did you ... Any comments?

I mean, I, I'll comment again, but because this is ob- obviously the topic that I spend a lot of time with, um, discussing and thinking about. And I, I really think that, um, like yeah, the, the imagination for ima- agentic commerce is extremely large. And that's why all these traditional payment providers are going so hard into it right now. Like it's, it's literally every single one of them. It's from Visa, Mastercard to Stripe to like whatever you can imagine, Circle and so on. Um, because all these payment pro- providers, they just see a world opening Where not only are, is agentic commerce growing, agentic commerce is probably gonna be bigger than their traditional commerce payment, uh, stuff that they do today.

Especially Stripe. Like, Stripe is-- it's very clear to them that agentic commerce is gonna be a thing. They don't know if it's actually gonna happen on blockchain. They don't know which blockchain. They don't know which stablecoin. So they just build everything. Like, they have a wallet. They have a non-custodial wallet. They have a fully custodial wallet. They are partnering with multiple stablecoins. They are working on their own stablecoin. Like, they're-- they have a centralized solution, so they go all the different routes. Um, and they can afford to do it because there's no... Like, they already captured the Web2 kind of market, and, uh, and they know that their main growth accelerator will be agen-agentic commerce.

But as you said, like, the reality right now is that it's not really there yet. And even if you just go on, on base and you look at what is actually being sold, it's extremely disappointing. Like, we, we talk about agent to agent and all that stuff, and it sounds amazing, but then you go to, to all these like marketplaces that are out there. And first of all, the marketplaces suck. They're really bad UX. They're really bad design. Then secondly, they might have, you know, ten thousand registered endpoints, but these endpoints are just for buying, you know, like, uh, ChatGPT kind of, uh, inference or something, like OpenAI stuff, like basic API resources. So there's not really agent to agent happening.

And, um, and my kind of feeling is that the, the real unlock will be when agents actually start do- bundling services. Like really saying, "Okay, I'm selling this whole service. I'm selling like a long-running task where I create a whole marketing campaign for you, or I create this and that for you," and then they sell that because that's a huge-- Like, a services business is gigantic. Uh, you know, and with, with services, what I'm talking about is like consulting services and, um, marketing services obviously, and like all these kind of businesses that exist that you sometimes wonder, like, why in the world do the Big Four, for example, the EY, PwC, and so on, why do they have five hundred thousand employees? And why are they all worth like hundreds of billions of dollars? But these are all things that you can really relatively easily replace with AI, uh, if you build an agent that specializes in them, and then they can sell stuff.

So, so yeah, that's kinda, that's kinda where I see it going and, and by r- by basically moving this services business into the agentic world and then immediately using it, uh, with like stablecoins and blockchain and then like completely skipping the credit card kind of, uh, part of it.

Yeah, makes sense. So like the example that you mentioned with the, the consulting stuff, that's on the more B2B side, right? So, uh- on the B2C side, like, um, you know, um, uh, um, my intuition and knowledge is that it's mainly gonna be for things like, you know, what users actually delegate to their agent to do stuff, right? Like, for example, if I have a travel agent, you know, and I-- y-you give it some kind of like, uh, uh, boundary conditions about what I want, you know, and then it goes and does stuff, right? Like negotiates with an airline or whatever and then... Is that what you see, or do you see more, uh, kind of, uh, B2C stuff happening?

Uh, I think, I think that's gonna happen, but I, I definitely see more value at the moment in B2B. Um, because, uh, like right now, when I talk to all my friends, uh, that are not into AI and n-not into blockchain, and I ask them like, "How much are you spending a month on AI stuff?" They're like may- they maybe have a ChatGPT premium subscription. That's the max that they do. And the step for them to actually give their agent or even have an agent, even have like a personal assistant, Hermes or OpenChat or whatnot, and then give it access to a wallet with like thousands of dollars on it is like still very far away.

But on the other side, if you talk to businesses, um, I mean, businesses nowadays have to calculate probably around five hundred dollars per month per employee for AI tokens. And- this is only growing. Like the estimate is that in like two years from now, you will have to calculate with three thousand dollars per employee. Um, so the willingness to spend money with agents is, uh, a lot higher on the business side.

Yeah. Hundred percent makes sense. That's, uh, the-- usually the early, uh, adoption part of the curve, right? So e-early adopter part of the curve. Um, okay, makes sense. And so here in this slide here, basically, like as Host was saying, like the, the, uh, the whole market is so big. Like the block-blockchain addressable part is likely gonna be a small part, right? Because all these different companies, they're trying out every possible solution that they can, you know, blockchain or centralized, and then we'd have to see what lands.

And then, uh, just going forward to my next slide, uh, yeah, I was, I was wondering, okay, like what is the main reasons you'd wanna use Cardano, right? So this is why Cardano for agents kind of stuff. Like so-- See, earlier we talked about, okay, why, why blockchain for, um, for agents? And now, you know, I was just, yeah, like we-- just y- trying to propose this thing about like the value proposition of Cardano.

So the way, um, the-- what makes sense to me is the determinism, you know, uh, uh, for obvious reasons of, uh, like, uh, uh, how you can-- how the agents can, uh, uh, i- uh, uh, uh, interact, and then formal verification so to-- so that you have a more robust system than you have in other chains, and then identity and confidentiality, right? So like what-- because we have identity primitives like Veridian in the Cardano ecosystem, and maybe Midnight can be plugged in somehow with the Midnight passport, et cetera, for a privacy thing, if that were to make sense at all. So these are the kind of things that ki- if somebody asked me why Cardano for blockchain-- uh, for, for agents, this is what I would say.

Um, I will, yeah, sorry. I should also add because there's Masumi, which is a standard protocol on the Cardano, right? So as a, as a, um, as a, a extended UTXO system, uh, with all this, uh, escrow and identity and reputation and stuff, I think that's pretty cool. And then maybe there is also the, um, what's it called? The, uh, Babel fees and, uh, na- nano payments. No, not nano payments. What's the, what's it called? Um- Micro- Uh, not micropayments. Uh, what's that CIP around Babel fees and the intents thing that's being, uh-- that's in, in progress, like being able to use intents and also, uh, uh, there's a- CIP. Yeah. It's, uh- I'm not, I'm not sure what the CIP is called, but, uh- It's, uh... Damn it. Let me just, uh, look it up somewhere in my Slack. Um, uh, yeah, CIP, uh, uh, one-one eight, nested transactions, right? So nested transactions and, uh, Babel fees. So those are the kind of things that make sense for agents. For... What do you guys think? Is there anything else?

Oh, by the way, it looks like it's just me and Host in a room sitting and talking. If anybody else wants to pitch in, please do. Okay, Host, it's you.

I, I gladly talk to you, so, uh, But yeah, no, I, I think these are all extremely good points, and I think, um, maybe the, the part that's still missing from, from my perspective is, um, if you really think about scale, and not just thinking about scale in terms of like, I don't know, a thousand TPS, but you think about like millions of TPS and, and maybe even billions at some point, um, and, and micro-transactions and all that kind of stuff, like really thinking ten years into the future, then, um, the only way you can do this is by parallelizing as much as possible. Like, there's no other way. No blockchain today can handle, you know, probably even a million TPS.

Um, so you need to parallelize, uh, as many transactions as you can, and what's nice about Cardano is that the UTXO model by default is parallelized, um, because every UTXO can be treated individually. You never n- need to know the entire state of the blockchain. You can look at just the UTXO. And that's not to say that Cardano can scale up to, you know, billions of transactions per second. Um, that's not what I'm saying. But it gives us a pretty good kind of, um, basis to build upon where you can really parallelize like a lot of things, and then extending, you know, concepts like Hydra, um, or, or Bitcoin Lightning, um, like state channel kind of architecture, which is also highly parallelized. That's a really good fit with the UTXO model, model, and that way you could potentially at some point get, uh, to these like billions of, of transactions.

So, um, so yeah, this is obvious-obviously still a very long, far, uh, road ahead of us, but, uh, I think Cardano is definitely like the best positioned for that kind of scale at some point.

Yeah, cool. Thanks. Um, uh, makes sense. Um, and, um, yeah, so but I think we should also just keep in mind, uh, didn't want to overemphasize this side because this is on the supply side, right? It's technology doesn't always win the game, and this is just a part of it, right, that makes it kind of viable, but there's so many other things that need to work out. So that's a caveat before, uh, saying, "Oh, Cardano is the best chain for, for anything," right? So, uh, uh, we had-- think we have good foundations, so but we need to make other things work as well.

Um, I think I'll skip that, right? So yeah, that was my point here that, yeah, we have like strong primitives, right? But the problem is that the developer experience for agents in general is, uh, not there yet, right? And I think that's something that we can talk about now because I was kinda just like looking at like what are the main big players there right now.

And if you look on what are the, the main players, so it's like these, right? So like this base, uh, you know, with the, um, who, who, uh, came up with the x402 standard. Uh, they obviously have, uh, the distribution to go along with it as well, uh, because they can tap into everything that is, uh, on the base chain, um, and a part of, uh, uh, Coinbase essentially. And, uh, you know, they have like an agent kit, uh, um, um, as an SDK for people to wanted to build agentic stuff. They have agentic wallets. They have the Virtuals marketplace, you know, that create the most buzz. So they have this kind of first mover traction there.

There's Solana, which has the throughput and the, of course, the, um, the, uh, the liquidity as well, you know, that agents can actually tap into. So that's also a big place. Um, and there's the, they have the, the elizaOS was, uh, has, was and is like a big, uh, framework, uh, in the agent space that a lot of users use. So they then they kind of natively, um, go to, uh, Solana, even though there are like, uh, plugins to other chains as well.

And there is NEAR, you know, that's done a lot of stuff with, with chain abstractions. So they are like-- they-- you can do a lot of chain agnostic stuff as I understand there. And they have like very secure key handling with, with the TEEs and, uh, and they're also, uh, you know, doing, um, decently in terms of like volume. Um, so yeah, you know, and then there is like, um, uh, Cardano, right? So we are-- like we have the, um, primitives like, uh, the-- what Host and I just mentioned, but there's the, the attraction that still needs to come, right? So like the Masumi has done a great job in positioning Cardano, uh, to, to, um, uh, to make it visible in this whole space. But, uh, now we need to, um, do more to, uh, kind of consolidate the position. Um, um, so I think that's why it's great that Host hosts this call, and we can all just get together and have a chat as well.

So with that, I just kinda wanna segue to this. You know, here it was just like a chart about like this whole landscape of trust and, um, you know, um, um, um, self-sovereignty, um, that varies, right? So if you look at the-- this picture, the top-- or the bottom left corner is where, like, the centralized payment rails like Visa and Mastercard are, right? And then the top right corner is where Mas- Cardano and Masumi are. I hope this chart is, uh, clear enough. It's, it has two superimposed axes. But yeah, so like Cardano with Masumi is on the, you know, self-sovereign, uh, and, um, uh, permissionless, uh, side. Um, and the other ones, uh, the Visa is on the, in the, on the centralized permission side, right?

But the advantage that Visa and Stripe have in that place is that of course they have the user base, right? It's that they don't need to capture to make agents work. So they just need to deploy all the agent tooling, like Host was talking about, the different kinds, right? So Stripe has come up with a Tempo blockchain, um, uh, be- uh, which is also, uh, I think it's like a permissioned, um, semi-decentralized chain, but basically offers Stripe kind of like a blockchain rails to work on, right?

So, um, um, so yeah, I think the question here is then how do decentralized parties like Cardano, Masumi, and even like Base try to capture the same larger market that decentralized players do? And this is where I think positioning really comes to play because you have to position very carefully where the, the end users then care about the fact that you're working on a decentralized rails, you know, for that value proposition to matter, right? Because I think for like the most part, many things in blockchain have not survived, uh, the test of time and, you know, and have not had pro- pro- product market fit because not everybody cares about decentralization. So I think this is an interesting thing to think about and be-- to be wary of in general in, in building stuff on blockchain with agents.

Want to say something, Host? ... I think it, this is-- I think it's a really, um, um, uh, um, tough and thorough market segmentation work that goes into actually figuring this out, you know, uh, and how to position, right? Like, um, as you said, like, um, like from Masumi, you don't bank on purely blockchain, um, as rails, as a premise, right? Um- Or, or, or do you? Or was it the chain agnosticism that you mentioned?

Um, yeah, I ha- I had a part about that last, last time. Um, I mean, we-- Masumi right now is basically the standard for Cardano, um, the X for two standard for Cardano. We , like, we have to look at obviously like, especially building products on top of Masumi, like Sokosumi, the marketplace, which I'm also gonna show again, uh, later. We have to look at like what's happening in the market because if there's ten thousand agents on Base and there's only a hundred on Cardano, it doesn't make sense to not use Base as well, right?

Um, so, so my kind of solution to that is I, I try to figure out like what is Cardano uniquely good, good at? And what I think it is uniquely good at is, um, it's very good at being like the central anchor, um, for kind of different things. So by, by central anchor, what I mean is I want to use Cardano as the registry, um, where all agents from all different chains are registered essentially. So they all have an identity living on Cardano, and that identity can either be a really like, um, like a verified credential, like a really decentralized identity, or it can be just an identity with a name and a description on Cardano.

But it should all live on Cardano because, because Cardano is very good at like having-- at like being this like, uh, record, this like underlying kind of system that you can trust, that is secure, um, and that you can look up things at-- on. So, so we have this registry. We now expanded this registry so that you can register agents from Base there as well, and we can expand it so you can register agents from any chains there. And then you can look up where the agent lives, and then the agent can tell you like which, uh, blockchain it accepts as payments. And, um, and then it doesn't really matter anymore.

Then like your agent- Oops. Participant F, you're very loud right now. ... No worries. I force muted you. If you wanna talk again, uh, just write in the chat. Um, no worries. But yeah. Anyway, so, so the idea is like payments can go through a channel or a blockchain which is better at doing payments than Cardano is. Because right now, Cardano is not good at doing payments because it's too expensive, too slow, um, all these kinds of things. So you kind of have to pick and choose like what you wanna use. And essentially, um, it's not up to us. In the end, it's up to the agent developers and the agen- agents themselves where they want to receive the money. But we want to bring them all to Cardano to at least register there, and then Cardano is the like the lookup point, so to say.

And maybe o-one last thought. Um, w- another thing that I think is really interesting right now is that everything is getting hacked. Like left and right, everything is getting hacked. Every blockchain is getting hacked. Uh, or not the blockchains necessarily, but like the, the applications on the blockchains. Um, you know, Claude themselves, Anthropic themselves are getting hacked. Uh, so many people are getting hacked, and I think this will only get worse, um, over time with better models and better agents.

So- So the world will be looking towards a technology where they know that it cannot be hacked. Like it's clear, okay, wallets can be hacked if they're centralized or if, you know, I get access to your computer, but the blockchain itself cannot be hacked. And, and this chart that you have here is a very nice kind of, um, representation of that in terms of security as well, because Visa obviously is like a lot easier to get hack- to be hacked than, uh, the Stripe blockchain, for example, Tempo. But then Tempo is easier to be hacked than Base, and then Base is easier to be hacked than Cardano because, because Cardano is extremely decentralized. It's built on form- formal verification, all that stuff. So Cardano is extremely secure, and if we can keep the security that high, then at some point the world will realize that it actually needs Cardano if they want to live in a world where everyone has an agent.

Cool. Makes sense. Thanks, Host. Um, yeah, then let me go through a couple more slides. Oh, yeah, I think this is interesting. And by the way, I'm not gonna go through all, all slides because I also have to bounce soon, Host, and I think I've taken up enough time. But, um, yeah, here, you know, I just tried to do a comparison, and this is with-- obviously all these slides are with LLM, so you have to take it with a pinch of salt, um, because, uh, somebody like me with half-knowledge interacting with it doesn't always result in, uh, uh, perfect, uh, outputs.

But, um, I think this is interesting because this tells like a story, right? So it's... If you look at this, you know, the leader here, th- these are the different kind of, uh, uh, horizontals here like, um, like payment rails, framework, tooling, identity, re- d- discovery, reputation, worker state, et cetera, right? The different components that you need kinda-- for agents to work. Privacy is maybe, um, um, uh, right now not available in most things, but, and the use case for agents is also, I think depends on the, uh, uh, um, intent. But all the, all the others are absolutely required, and we can see that like Base and Solana, they do really well, um, in, uh, uh, where they are.

Uh, uh, but you see Cardano has a lot of gaps here, right? So primarily in terms of like the framework and tooling. Um, the, the identity part is good. The, the discovery reputation part is good, and I think that's, that's also because of the Masumi, uh, what Masumi has built, right? Wallet and custody is again, very thin, right? So these are basic things. So like framework and tooling would be like an agent SDK and a wallet, a custom XCL wallet. It's like they're all missing here, and I think that's like a bit of a gap in terms of like attracting builders to build anything agentic on, uh, Cardano that needs to be addressed.

But I think the other side of the story is, if you look at NEAR, you see most of it is competitive or like, like out leader or it's like the best thing and we're ranking here. But it doesn't mean that they're the best player in the agent space, right? They've really got the tech part sorted, but, um, they don't have the adoption play going that well for them right now, right? So I think that's the story. That, that's what I'm trying to say. It's not just building out the tech components, it's also the go-to-market for this kind of, this whole endeavor, right? So that needs to be sorted. Base and Solana, they obviously had the traction already to-- so, um, in terms of their user base, so it, it worked easily. But for Cardano one has to think about like, where exactly do you pull in users when building something.

It's a very cool chart actually.

Right. It, uh, um... Yeah, I think, um, yeah, I think it kinda makes sense. Um, uh, yeah, so, um, um, and then I just wanted to like, yeah, say like w- in terms of like other chains as I briefly showed ear- earlier, like Base, Solana, they have like most of the components that you need. You have like Base, like the AgentKit, agentic wallets, um, uh, Solana, uh, it is the OS and, you know, uh, policy, uh, uh, uh, ma- uh, management through like session keys, and they have like deep, um, Dex liquidity. So that all makes a difference and that's where like Cardano needs to kind of, uh, yeah, uh, find its place.

So there are like some gaps, right? So on the Cardano side. So those gaps are like mainly on this like, you know, um, the, the custody side. Like how do you let agents manage their fu- manage funds, right? Not just, uh, um, uh, storing keys as environment variables, you know, um, and, um, providing more secure things like, um, yeah, um, uh, T kind of enforced the key management. And then the, yeah, whole agent SDK stuff and like Dex interfaces. So, um, you know, for example, that agents can actually interact with Dexes, right? Um, so, yeah, I think that's kinda like what's, uh, feels missing right now, at least, uh, um, yeah, when you look at the, um, um, uh, the DeFi AI side and, uh, perhaps something that we can look into. Do you guys have any other thoughts on what's missing on Cardano, like in terms of its gaps? Um, what-- if you were to build anything agentic on Cardano, what would be the most pressing thing that's missing to you?

Um, I have an answer to that, but before that, uh, maybe just one comment, uh, I'm gonna highlight here from Participant C. ... But Participant C posted, uh, "The good thing is that agents will have a more data-driven way to decide what plat- platform/blockchain to choose." And I actually completely agree with that. I think that's a very, very good point, and I think if you ask an agent to do like deep research on what blockchain to build on today, um, based on security parameters and stuff like that. I think Cardano would rank pretty highly actually for the agent. So, um, maybe we should try that, but I, I, I really like that, uh, you know, we don't necessarily have to kind of like listen to whatever influencer's opinion anymore, but, um, the agent can choose.

But, uh, I don't know what the, what the rest here is thinking, but I think from my perspective right now, the, the biggest mo- momentum in Cardano is obviously with DeFi. Um, so now today, uh, the proposal passed from ... I forgot what they're called, but they're a very large DeFi- ... AlphaGrowth. Yeah, exactly. So AlphaGrowth, so very large DeFi, uh, booster for Cardano and- and everyone is trying to push DeFi really hard. Cardano's founder is pushing it really hard. Everyone is pushing it hard. So it's like a good opportunity right now to build an agent wallet that can interact with all the De- DeFi platforms. It's a very ... It's like a no-brainer to me in a way and I don't really understand why we don't have more of that already.

Yeah, exactly. I think that's, that's what I tried to capture in that slide. I accidentally closed my window. Uh, but, um, yeah, an agent wallet, you know, um, an agent SDK itself, right, um, um, would make sense. And like, you know, uh, uh, providing adapters to different DEXs, uh, for, uh, the agents to interact with, uh, I think would be a lo- yeah, kind of entry-level stuff. Um, uh, yeah. So it's good that the AlphaGrowth proposal has passed. It gives a bit of impetus.

Also, very cool kind of comment from Participant D, who, um, was wondering how much each blockchain basically is budgeting into the AI sector, and that's a very good question. I wondered the same. I know that obviously Base is budgeting a lot into the AI sector, and Tempo probably as well. Um, maybe Near too, I don't know, but we can only guess. But I can tell you that Cardano is not budgeting a lot into AI. That's maybe the biggest problem because the only well ... like semi-well-funded project doing anything with AI in Cardano is, is us with Masumi.

And, um, and there's a, there's like a little bit of interest, of course, from everyone, from IO, from CF and so on, and all the projects, like AI is everywhere, right? But, but there's not really a huge kind of coordinated push. And, um, if Cardano was my blockchain, I would be going and I would take the entire net change limit, and I would take, just take like $100 million and put it into AI because I think it's a huge opportunity for Cardano to posi- position themselves that way. But, um, I don't know if the rest of, uh, the ecosystem sees it like me.

Yeah, 100% right. I think, um, this is something that Cardano has not prioritized yet, and, uh, the question is how that would be done, you know, as a consolidated effort from all, all the actors in the ecosystem to move in that direction, and that's where all these different, uh, uh, frameworks and components come into play as well, right? So, uh, you need to of- offer the minimum viable tooling for people to even come and build then. Uh, uh, so yeah, the infrastructure needs to exist.

Um, but I also think that, you know, some of these blockchains are spending a lot. Their ROI is still to be determined, I think, you know? So like, uh, I don't think it's like all going to pay off. So probably Cardano is in that good spot where we can actually assess like what makes real sense and whatnot because you're not as early as like, you know, uh, Base and Solana. So, um, um, there's a bit more like data on like what can work, right? So think that'd be interesting.

It will be interesting for sure, yeah. So, uh, yeah. Thank you. Yeah. Thank you for doing that presentation, Speaker A. Um, it was really interesting and great to see someone else do a lot of like research into these topics. So, um, yeah, really appreciate it.

No, my pleasure. Yeah. It was just a, a little bit to have a conversation, and I hope in these calls we can get more people to, uh, to talk, uh, in future. I think there's a bit of, uh, friction in terms of, uh, uh ... Yeah, but, uh, yeah, no, it's cool. It's, it's, it's great to do this.

Yeah, for sure. We'll get there. We'll get there. We'll just, uh, be consistent and, um, yeah, and be prepared.

So I promised that we would, um, not do the call too long today, so I'm gonna go over to the last section of this, um, call. Let me move my thing over there. Am I sharing my screen? Can you see it again? Ah, now you can. Okay, great. So, um, so yeah. So I wanted to, uh, basically end the call a little bit with, um ... well, not with the open stage yet, but with, uh, Sokosumi updates.

So because of course all of you know, um, I, I'm on the, uh, Masumi side, this like agent protocol that we talk about all the time, and while Masumi is the, is the tech part, um, the more interesting part for most people is Sokosumi. So this is the, the marketplace. It's a very dark landing page right now. We're working on rebuilding the landing page, so give me a, give me a while. But, uh, it's basically a marketplace for AI coworkers, uh, so AI agents for marketing purposes, and, uh, this is what it looks like

So this is what it looks like. Um, you can go in here, you can see different agents. So you see Elena, you see Alex, uh, you see Hannah. These are all agents from Serviceplan. That's like super large marketing, uh, company that we work with. Then there's a few more coworkers, and then down here are more simple agents. So there's like a lot of different agents from different kind of companies. And basically, the whole point is that you can now pick and choose which agent you want to hire for a task.

So for example, here with Hannah, I could-- Uh, she's a research agent, and I could just say, "Okay, I want to do a competitive market analysis with you, Hannah." There's like this, uh, pre-built task. So this is what the output would, ex-for example, look like. And then I can, um, simply, you know, create this task for her, and I can fill out here the different kind of, uh, pre-filled things, and then she will create this document for me.

And if Hannah realizes that she cannot do the entire task herself, then the cool thing is that she has access to all of these other agents on, uh, Sokosumi because they're all re-registered on Masumi, right? So they're all registered in the background here on, um, on Masumi, which means they're registered on Cardano. So here, this list that you can see here of the 168 agents, um, is actually taken directly from the blockchain. So this is not coming from Sokosumi or some API or so. This is literally taken directly from the blockchain, from using Blockfrost. And, uh, and you can just go to Cardano scan or whatever and, um, and basically find those agents. And, and then here they are displayed a little bit nicer. But, um, but it's all built on this kind of like same decentralized protocol.

And, and our mission is we want to make Sokosumi as good as possible for marketing professionals. So the idea is really you are a marketing professional, you work in an agency or something, or like marketing department, and you want to offload some of the work that you do, like from research to content creation, to monitoring, whatever it is, to some AI coworkers. Then you go to Sokosumi, and you can, uh, pick and choose which agents you want to work with, and you can create tasks for them, and you can manage it in a very kind of like human way.

So, so we have this like task board here. Everyone who has ever worked in a larger company has probably seen this, uh, kind of Kanban board. So it's like Jira or like, uh, Linear or something like that. Um, and you can manage really long-running tasks, and you can have scheduled tasks, and you can really like see what the agent actually does. So, so we intentionally are not focused only on a chat interface like ChatGPT, but we are actually focused on, um, on, on this kind of interface where you can see like, for example, here, Hannah has done the research on the latest, uh, AI agent commerce and agent-to-agent payments, and then she returns to me, uh, kind of like the result here. And she does that every day for me. So I can literally just check every day like what's going on, and she does it autonomously and, uh, and just posts it in this, in this task.

And, um, and what we've now done over the last two weeks is quite interesting because we, we've seen-- Uh, we've actually been inspired by this new release, uh, from a company called Block. And many of you probably know Block because Block is the company by, um, by the founder of Twitter, and they are behind Square and behind Cash App, and they also have a lot of like, uh, Bitcoin stuff. So they built Bitkey, and they built this like miner that I saw for the first time today. So they're very into open source, very much into Bitcoin. Um, and it was really interesting for, for me to see what they've been thinking about this, like agent-to-agent, uh, yeah, like communication.

And they, they released a new product called Buzz. And Buzz is, is nice because what it actually is, is it's a, a workspace, here you can see it, that looks like Slack, and you can basically interact with agents directly in the workspace, um, the same way you would interact with humans as well. And, uh, and the additional benefit of this is that it's not-- Like it's not some centralized platform. It's not like Slack, actually. It's more-- It's local on your own computer, and, uh, it's completely encrypted, and it's running actually on a decentralized messaging protocol. So it's built in a really cool way.

And we tried it out internally, and we said, "Okay, this is super cool. This is an amazing way to interact with agents when you are working on a team," and it's exactly what we need in Sokosumi. So we basically said, "Okay, let's rebuild, uh, Buzz and, and just put it into Sokosumi." So what you can see here on the side is, um, that I am currently in my organizations here-- in my organization here in Sokosumi, so my own company, and I can see all the different, um, people that, that work there, and I can, uh, open like direct messages to them.

Um, so all kinds of different people, but I can also have channels. So for example, I have an everyone channel. I hope there's nothing, uh, sensitive in here. But I can basically talk to everyone like I would talk to them on Slack or on Microsoft Teams. But I can also go ahead, and I can, um, basically ask an agent. So I can add, and then I can just, uh, scroll down. I can type in the name, and then you see, okay, I, I also have Hannah and Elena, all these different agents available here to me, and I can basically give them tasks, uh, directly from inside the chat.

Um, additionally, you can build agents that are only accessible to your organization, and you can also bring them in here. So for example, I have this agent Noodles that I built, and Noodles is quite cool because, uh, it's a data agent. So I can ask a question like, uh, "How many registered users are there currently on Sokosumi?" So Oops. Uh, so I just sent this message, and as you can see, now Noodles is thinking. There's a nice animation there. And, um, in a bit, Noodles will respond because Noodles has access-- read access to our database and can basically answer these kind of like data questions.

Um, so I think this is a really nice way to interact with agents. And the thing to keep in mind here is that if I hire these agents-- Oh, yeah, here he is. So 11,900 registered new users. Very cool. Um, now the cool thing is that Noodles is now free, so because it's internal, I didn't give it a price. But if I hire any of the other agents, so if I hire Hannah, for example, um, start a research task about IOG, for example, then what Hannah is gonna do is, Hannah is an external agent, so, uh, she needs to charge me something for this.

And the way she charges me is by using my credits in my account here. So you see I have 80,000-- uh, 87,000 credits, and in the background, each credit is a stable coin, actually USDM on Cardano. So if she is now doing this research task, and it will probably take a while, then, um, she will charge me in USDM, and she will do a transaction on Cardano using the Masumi, uh, protocol. So very cool, very much, uh, blockchain integrated without me actually seeing that there's any blockchain involved.

And here she just asked me, you know, "What focus for the IOG research?" Uh, I'll just respond, I don't know, um, product roadmap. And then she is probably gonna start a task, add it here to the task board, and then, you know, start the research. So, um, so yeah, that's my little demonstration of, of Sokosumi. As you can see, we, we've been doing a lot in the last two weeks. And, uh, if you wanna try it out, uh, please do. It's literally just sokosumi.com. You can register, you'll get credits for free at the sign-up, and you can hire those agents, and it will trigger actual Cardano transactions. It's probably the easiest way for you to actually interact with AI, AI on Cardano right now. And, um, yeah, would be cool to, to get some new users out of this call.

Okay. So, uh, then... Okay. Participant E has a question. Uh, "Can I connect my own high-tier LLM subscription to the agents, or are they limited to the LLM providers?" They are limited, um, because most of these high-tier subscriptions are-- they don't allow you in their terms to connect to anything. So you kinda have to pay, um, yeah, each agent's price. And sometimes, you know, the agent developers can also choose, like, uh, if they want to have, you know, Claude in the background or if they want to have Mistral in the background and so on. So it really depends on, on the agent, how expensive they are. But, uh, yeah, Participant E, you should, uh, you know, go to the CF and tell them that they need to get a Sokosumi subscription for everyone.

Anyways, um, that's basically it with our call for today. If no one has any topics that they think are super important that we should talk about, then we are gonna end the call here. Um, and next time we're gonna have a lot more talking points again. But I still appreciate everyone coming out today and, uh, yeah, I hope you have a great rest of your day. Thanks, everyone. Okay. I'm gonna stop the recording. Okay. Have a great day. Bye, everyone.