How GenLayer Is Building a Court System for Disputes Between AI Agents

Transcript

There's there's a huge huge gap in in trade finance for basically tons of of of like deals that are not possible just because the legals around it would be so expensive that you cannot really like insurance or like really make it make it a reality, right? We're trying to make it so that you know imagine a small court. So, you know, all these different you know claims that normally would be an economical and just trying to to get a quick result. Maybe it's not perfect but it will be good enough that it will allow for many many more traits to to happen. Hi everyone, welcome to Unchained, your no hype resource for all things crypto. I'm your host Laura Shin.

Before we dive in, we'll first hear a quick word from the sponsors who make the show possible. This episode is brought to you by 1 in Aqua, the shared liquidity layer from 1 in. Back multiple liquidity positions with one wallet balance and keep your tokens in your wallet until a swap fills. See how it works at 1in.com/aqua. Today's guests are Arthur Hayes, CEO of Flop Labs and CIO of Maelstrom, and Albert Castana, co-founder and CEO of Gen Layer Labs. Welcome, Arthur and Albert.

>> Hello. >> Hey, Laura. How's it going? >> Just a heads up for those watching on the live stream that this interview was pre-recorded on Thursday, September 3rd. So, Arthur and Albert, you two are here to discuss an interesting new idea for the future agentic economy. It's called Gen Layer and I know that the backstory to this starts with Albert. So Albert, can you tell us what gave you the idea to launch Gen Layer? >> Yeah, so maybe uh as a bit of background. So I I got into crypto in 2013, something like this. And I started working on different pro uh protocols like NEM uh got to like 15 billion market cap something like this. Radics DLT, the Badger Dow, and eventually I I

co-ounder and CEO this this company called Stakehound, which was the first liquid staking solution, like a bit before Lido. Um grew pretty quickly, $350 million in assets under management in about five months. And then we had to stop the business when when our custody solution lost two keys and put $150 million on fire. So we had to start, you know, looking into the legals of that. We had to, you know, take a step back and think, okay, what do we do here, right? Like you start talking to to lawyers. Everybody wants you to, you know, stop the business and all these things. But what we saw really is that, you know, the the legal system is really really broken. It's very ineffective, very slow, very expensive. It's essentially inaccessible for about five billion people. So five billion people don't have access to the legal system

and you know it's it's it's a shame right it should be something that's available to everybody that's fair that's you know like transparent as much as possible unbiased right and and what we saw is that you know it's not the case uh for example in Israel you need to pay you know 1% up front so somebody loses $150 million you need to pay 1.5 up front just to to initiate to initiate a claim right and so we started looking into into that it was the same time as the you know chbd came uh with all these you know amazing LLMs it's insane to see what you can do there like you can have a piece of software that can write new code right so software that's writing software and that can essentially understand you know nuance and you know have some common sense we were totally

totally fascinated by that right it's like what can we do with this right it's just insane so we started building a couple um different frameworks around that uh with GBT3 and GBT4 um you know like trying to get agents that would be building you different type of like software and like trying to essentially improve on themselves. But what we saw very quickly is that well actually it's not just going to be you know them being able to write code or write text right it's actually they will be starting to transact with each other very quickly right um actually I remember Arthur that we spoke uh you know about this first time maybe like two or three years ago or something like this right so it's been a while um you you already said like this is going to be a long time from now I think it's it's been a while

I actually thought it would be a bit faster but what we see right now is that there's just so many you know like teams working on how to make them transact with each other. where basically every single VC right now is investing on agents like you know uh transacting with each other and commerce existing right um there's many you know like estimations in the the tens of trillions of dollars of agentic commerce but the problem that we see here is it's not just being able to move money it's actually that when they will start to you know like move money and they will start to deal with each other they will also start to disagree right and so the question that we're proposing is hey if you know if you have a a commerce Like if you want commerce is not just to have money, it's also that there needs to be

a trust system. There needs to be some sort of like legal system where that happens, right? How do two agents trust each other, right? When they disagree. Yeah. And so that's basically what we've been building for the last three years is essentially like think about it like a a court for the internet, right? So different AI agents that reach, you know, agreement together on subjective terms as essentially as if you were going to a judge, right? But this court for the internet can deal with things at light speed basically, right? So like, you know, a claim takes, you know, minutes and, you know, fractions of a of a dollar. So they can enable completely new types of of of contracts that until now were only like limited to humans reviewing that piece of of work.

And so Arthur, you just um announced that you're you know have that you have this new venture flop and obviously um AI agents are kind of your your new thing now. And you talked about this idea that you had where you know this protocol would be based around a unit of compute and and that's what a flop would be. So how do you see flop working with gen layer? So, we're not going to achieve any sort of meaningful valuation for the token itself above just operating a really cool spot market for compute unless we have ejected commerce that's native to the flop network. And right now, how we're looking at it is, you know, they'll be agents. They'll come to some sort of agreement in a

swarm or bilaterally do some sort of deal. We have a sort of an escrow protocol called HTLC that's native to our system. And then as Albert said, there's going to be disagreements and then how do you adjudicate those? Now, we we really haven't thought about that too deeply on our particular protocol because we're more more focused on getting miners on board, validators, making sure that we can actually do proof of use useful inference, making sure the AI agents can eat. But I think this is definitely something that's absolutely going to be needed because of course if humans disagree, then the agents have to disagree because they're essentially the derivative of of our own sort of um proclivities as economic agents. So I think it's something that's definitely warranted. This is why we got involved. I don't even know how long it

was that we first spoke about this at Maelstrom came on board as an adviser. But obviously like this is the time for the buildout of the rails for ejected commerce and you know it kind of fits. Now I didn't plan it like this. You know this idea for flop just kind of came to me a few months ago and we started building but it's it's amazing how different folks are working on different pieces of the puzzle to really build a decentralized layer for agents to coordinate. Oh, okay. I had assumed that you got involved with Gen layer because you were doing Flop, but actually you got involved with Gen L layer and then later you had the idea for Flop.

>> Correct. >> Oh. Oh, okay. That's that's super interesting. Um, so so maybe give examples of like the different kinds of disputes that you guys could see arising between AI agents that Gen layer would be called in to address. >> Right. So I mean if you think about it most of the contracts in the real world are are ambiguous right uh you know we can say you know if you do this piece of work for me then I will pay you right so that's that's normal an agreement right you can have some service license agreement whatever it is right and normally what happens is that the disagreement arises not because one of them is bad but actually just because they they disagree right it's like is it professional enough yes no it's not binary really right it's subjective and

so the problem there is that you need to have a system that can you know make it so that it's like repeating the same question multiple times will give you the same answer right the problem is that LLMs don't work like that right if you ask you know chajd you know is this uh is zilinsky wearing a suit or not it's going to be answering one thing or another at different points in time right so every time that you ask it will give you a different response right and that's challenging because you cannot really have an automated system if you cannot you know repeat and like get uh you know like consistency. Yes. So what what generator does is essentially allow for contracts to have natural language or ambiguous terms even go to the internet or understand you know an image

like take an image and observe what is in there and still have this reproducibility right so what's really happening is that different AI agents and AI models are taking that specific contract and they are reaching an agreement between them on whether the piece of work was done properly or not. Yes. So let's say that you have a performance-based contract. You put some money and then you say, "Hey, if you do this piece of work and it's in good quality, then you can take the money." Then we don't need you and I to be arguing if it's good enough or not. Right? It's just the consensus, the network itself, right? This court for the internet that makes the decision if it's good enough and then decides to release the escrow, right? This is until now just impossible, right? You need to

either have a human that's looking at that piece of work or you need to have just a smart contract that can that can execute the code of that but cannot really understand nuance or you know the the natural language right and that's a huge limitation. So >> so um one thing that was interesting was you gave that example that was from a prediction market and that's was very human- based where um >> uh humans were disagreeing about whether or not Zinsky wore a suit. Um, so do you also see this applying to prediction markets like helping solve disputes in prediction markets?

>> Yeah, I mean at the end of the day I think it's very much about reproducibility, right? The the thing is humans have biases and AIs have biases, right? So the question is really how do you have a system that gives you maybe not the capital truth because does it even exist? Is it really a suit or not? Like is there an real truth on whether there is or not, right? um or is it that you can just have a consistent response, right? And if you have a consistent response, you start to trust that system. What we're missing right now is that both humans, right? For example, you have a prediction market like Poly Market. Poly Market uses Uma for the resolution, right? That means that it's a bunch of humans that are voting on whether Zelinski is wearing a suit or

not, right? Well, you're getting an answer from, let's say, one person, maybe five or 20 people essentially, you know, deciding that they even have a stake. they have, you know, a few days to even think about it. They need to pay guys gas like dollars and dollars to be able to vote. It's very inefficient, right? So, what we're proposing is a similar system, but it's essentially a a blockchain where each validator is connected to a different AI model at the consensus level, right? So, the network itself is that is different AIs reaching an agreement on subjective things. And so, it allows you to make something reproducible. you would be if you ask a hundred times the same qu this question to the network with the same nodes let's

say you will get the same answer right because you're basically escalating it up to like 1500 different AIS and they all need to reach an agreement on that right so it's a different type of problem like we're very used to having um blockchains that are like deterministic right so like giving an input an input it's giving the same like an an output and you repeat that right 2 plus two is four and it doesn't matter how many times I ask you these AIS are not like that. They're not deterministic. They're not deterministic. Right? And the problem with that is that then you cannot actually build systems that are reliable. Right? When you're talking about verifiability, it's not just verifying whether the thinking was done correct, but actually whether the process was done correct. Right? You don't know if the judge, human judge got

divorced last week and you know in this case has some bias, but you know that you can appeal and take it to a higher court. Right? So like the legal system is already thought in a way that you don't really depend on a single judge, right? Yeah. The problem is that these AIs, they're going to start to be transacting with each other at a speed that is insane, right? I think, you know, few years from now, 5 10 years from now, pretty much every like all commerce is going to be done by by a agents of of different sorts, right? When that will happen, I can't imagine how they can exist coexist if there's no sort of like legal system between them because like where are they living? you don't know where my AI agent is is living. Do I even know what it is? Like

it's right here or like in some um you know like Amazon Web Services or or whatever. So these agents are going to be cheating. They're going to be cutting corners. They're going to be smart as hell. They're going to be moving super fast. It's going to be very very hard to see what they're doing, right? How do they trust each other if they don't even have a legal system? Because the legal system takes, you know, years and hundreds of thousands of dollars to resolve a small claim, right? That's why we're building this discord of the internet. >> By the way, I would love to see what the AI agents would or the the validators would vote on the Zalinsky suit. Um [laughter] anyway, okay. So, we're going to talk a little bit more about like how the

process works in a moment, but first we'll take a quick word from the sponsors who make the show possible. $540 million. That's how much concentrated liquidity sat idle in a given week in the first half of this year. About 30% of the DeFi TVL if you're wondering. That's according to Dune Research commissioned by 1 in. But there's a solution. 1-in Aqua is the new shared liquidity platform. It lets LPS back multiple positions with the same token balance and keep their tokens in their wallet till a swap comes. Why does that help? Because LPS don't have to split their tokens across positions.

They can cover more market conditions and pairs with their full balance. That means more activity across deeper liquidity. See how it works at 1in.com/aqua. Remember that providing liquidity carries risk and fees aren't guaranteed. Back to my conversation with Arthur and Albert. So yeah, let let's walk through what it will look like for Gen Layer to resolve a dispute. Um, so let's let's say that each of you is an agent and um you two engaged in some transaction and now one of you wants to dispute it. Um, or or it could be that each of you is a human with an agent, I guess. So I want to like walk through the whole thing step

by step, but I actually am so curious about the first step. So how do you even get the agents to agree to have their dispute adjudicated by Gen L layer? Like let's start there and then we can do the rest of it after. So at the end of the day we need to think that these agents are smart right and that eventually they will be controlling money and they will need to know how to control the money without losing it right like right now few people are using agents still and even fewer are giving money to these agents right eventually I think that that's not going to be the case they will not be so you know dumb as to lose money on some prompt injection attack or something but they will need to also understand okay if I want to make a deal with another agent well what's the

framework right am Am I just you know am I following I have this legal system am I using a trusted system maybe you know like all the agents from open AI they will have you know like trust in each other and they will never cheat each other never cut corners nothing like that right totally you know trusting open AI but then you have you know all the different labs are going to start to be you know more more fighting and also you will have different countries and then we all know how hard it is to have you know cross jurisdictional law right so it's getting very very complex all these disputes are going to be insane Going back to your question, the point here is that these agents will have the possibility to not just go and hire a bunch of lawyers. They will also have

the possibility to define what we call an intelligent contract. It's a smart contract, same as a smart contract, but it can also have natural language. It can also go to the internet and read any image and essentially understand nuance, right? And so, you know, my agent goes in, creates an intelligent contract, puts some money in it, and then Arthur's goes in and says, "Yeah, I want to fill this promise, right? Let's say that I want to, you know, do some research, and I'm going to be getting paid for this research. Let's say that this this research is compliant." So, does go does do its thing, comes back and submits the research into the intelligent contract.

>> Oh, so okay. So, this is like even before the transaction. I mean, >> oh, okay. We sign an agreement you and I, right? And so we have a paper for that right? >> So as part of the agreement, Gen layer is like already Oh, okay. Got it. >> It's like I mean on same as Ethereum, right? On Ethereum you go and you deploy the the liquidity pool, right? And then people can trade against you. >> Same idea, right? It's a it's a network where you can essentially deploy contracts. Now, it's not focused on really like the the TVL or total value lock, right? It's actually focused on governing, right? And essentially managing what's happening in there, right? the this court system and so we have this escrow that's holding the money and then that will be able to read evidence from the real world without any

middleman to then make the decision of whether this piece of work was done correctly or not whether was professional or like following the standard or whatever it is right the decision of that will come not from one single AI which would have some biases or could be wrong and essentially every time that you would be asking it would give you a different answer but actually will be asking up to these 1500 AIS, right? So, it's like escalating rounds of app builds. You start with five and then you do more and more and more until you do, you know, eight times you would be at at 1500 different uh validators, right? So, it's a permissionless network. Anybody can jump in. They need to have the token in order to prevent zero. So, it's a delegated proof of stake, right? And then anybody can run a node, connect to whatever LLM they want.

This is not fixed by by us or the protocol or anything like this, right? that what you want is as much diversity as you can right and so when there's a you know in this case this adjudication it's going to go to the network the network selects the validators that will be part of this uh adjudication and then they reach consensus between them um and then the funds get you know released like going to the judge right you wouldn't go to the judge for any small thing you would be going for important things now these important things can be you know tens of dollars doesn't need to be you know hundreds of thousands of dollars for you to to actually take the Right.

>> Okay. And um this is its own chain. >> So the the architecture it's basically an L2 called J layer chain with another over layer. So that's another network, right? With these 1500 validators that are reviewing the transactions, right? So it's like you have the banking system and then you have the legal system, right? And so the the judiciary, the legal system is the one that's like deciding. So they do you know commit and reveal the voting process is happening really on the chain but then let's say you have a a pot on Solana right and then the decision making for that is happening on gen layer and transa and you know it's it's messaging from one to the other

>> okay so the money can be on any chain um but the the >> the staking the valid the governance oh I see governance the decision making right exactly so what's happening here is let's say this prediction market right so you have that poly market wants to resolve uh something and instead of going to a few humans that decide that they're going to a few AIs right and what what's really happening is that that transaction is going to the chain and then all the nodes of the network see that that transaction happened and they know when they can speak so they go to the internet they go to an LLM they do whatever they need to do to reach an

agreement between them through a through a majority voting system between them in a matter of minutes. >> Okay. Okay. >> It's a consensus of AIS. >> Okay. So, um yeah. So, this architecture just I like Okay. So at the moment most AIs transact on base but then your layer 2 is for the governance of gen layer and it is interfacing with another layer on top of it that is the one that is interacting with the AIs that are making the the calls on these disputes.

Let's say that you have, for example, uh an escrow on base, right? Let's say that there's a there's a pot of money that says everybody that's talking about, for example, on Twitter, right? That's shilling this this podcast on Twitter, right? They can get the money. Yes. So, what's happening there is that the money is living on base. Agents are offchain and offchain reading a specific contract. Let's say for example on base, right? And so they are participating. They send their content to this contract. Now the contract connects to Zen layer for then the network of AI agents to go and fetch the piece of content and evaluate and see if they agree or disagree on how much they should be getting paid and then that message is is sent back to base for the distribution.

>> Okay. >> Right. So the assets are on base the decision making is happening on Jer. >> Okay. But so then it feels like there's multiple areas of attack. It's like you like the AIS could be corrupted in some way. um the message between gen layer and the escrow could like right so how how do you keep it all secure >> so the the the messaging is any any like messaging interperability layer right so you can use something like layer zero or hyperlane any of those that's how you communicate from one to the other right we're not trying to reinvent the wheel there we're also not trying to reinvent the wheel in the in the L2 right so that's going to be um it's a it's a proven stack Right. And then what

happens on top of this and that's our innovation is the execution environment and the consensus and the note for these AI models to to synchronize between themselves right that's what we focused on and so in there indeed this is a fundamental different type of DT and what that means is that you need to well you have different you know security and livveness assumptions right for example in in I mean we can get very deep in this but essentially what we're talking here is a probabilistic bizantine fault tolerant system, right? Meaning, if you have, you know, on on on on any blockchain, the the validators are comparing the hash of a block to see if it's correct or not. Yes. Comparing hashes is something that's

deterministic. There's no way to to to know that you were wrong, right? Like if you're wrong is that you're lying. Yes. The problem in something like, you know, is, you know, Zalinski wearing a suit or not is that you may be wrong and not lying. You may not be bad, but you may still be wrong, right? We may disagree without either of them really being being malicious, right? And so what that means is that if you have something that you know it's unclear, for example, it's more possible that there would be disagreement. And so a consensus on that requires different assumptions and that's the architecture itself, right?

That's basically how um I mean we have a a lot of research that's that's coming out in the next few days on that like a few papers that explain exactly what does it mean to have a probabilistic or or essentially what what we're calling a disambiguation machine right um which is you know you have this ambiguity how do you extract it so that you have this uh reproducibility to be able to to build you know automated systems on top um but yeah for sure it's it's new we've been building this for for three years there's a lot there's been a lot of challenges. Now I think it's it's at a point where now it's been running really on on test net for about a year and it's it's pretty it's working really well.

>> Okay. So Arthur, earlier when I asked you about like how flop would interact. I think I was only thinking that the agents would be in the disputes, but I'm now also realizing the agents could even be like working on the side of like adjudicating. Am I right or I don't know. Are are you [clears throat] >> I thinking is more of a tool, right, for if we're going to have agent commerce, if they're going to use flop, right, they've done something offchain, they want to get paid in flop, they're using flop as sort of a um a layer to say this happened more so than anything else.

Release my flop. Okay. Well, what if I disagree that Silinsky was wearing a suit and that was part of the contract? Silinski is wearing a suit, I get 10 flop. If he's not, I get 10 flop, right? Like same same sort of example. And then how do you sort of make that call? Like we're not trying to be a judge and jury as you know a network where AIS eat and remember things. So this is a tool more so than anything else. And I think that's sort of what anyone building in the space is going to need these tools to sort of make whatever their vision of of agent commerce come to light and sort of not have to deal with things they don't want to deal with.

And um so one other thing that I'm curious about is like do the humans also kind of instigate things or is it like even the agent itself might decide like wait I need to I want I need I have a dispute even before the human gets involved or would it always be a human being like I'm unhappy you know let's let's get this dispute adjudicated by gen layer. The way we look at this is that these agents will want to not lose the money and to not lose the money they will want to have, you know, a trust between them, right? A trust framework.

Now, right now, they don't really have that, right? They cannot believe in the legal system just because of the cost and the the, you know, the the slowness of it, right? Um, so what we think is that these AI agents will think, hey, actually, if I had some sort of smart contract that can be holding the money as an escrow and also evaluating the evidence in a neutral way, then that would be the safest way for me to not lose the money and be able to actually do commerce. So, we don't think that this is going to be for humans. I think, you know, humans right now are using it already on a bunch of the protocols that are built on top of it. But I think down the line when people really start to give these agents money they will see that this is a very efficient way to to

do commerce right it's again it's it's an internet it's a court for the internet right and so for them it will be just I deploy the contract I have negotiated offchain something with with another agent you know the money is is transacted is put into the escrow and then the other agent will do some piece of work and then it will be a neutral party that will decide who's right I mean this goes for you know insurance claims it goes for many many different applications. Um but I really think it's going to be a lot of the agents just deploying the contracts and trying to resolve between them as we do with humans, right? Um now the problem with humans is just that you know the the legal system uh you know an agreement it's very very archic uh system >> and then you talked about the appeals

process um but it's sort like am I wrong in thinking pretty much all these decisions can be you know instantaneous is like a strong word but um compared to the human court system like near instantaneous so what's to it just sort of feels like pretty It's so low stakes. Like, so let's say the first judgment comes from the five AIs and then, you know, the side that lost wants to appeal. It just feels like it's going to be super cheap and um and really easy to just keep appealing until you get to the 1500 and then there's the final. So, doesn't it sort of feel like all of them are going to want to play it out or it does it just get so expensive or I I don't know. And and even like how

long could appeals be? Could it be like literally within you know two hours you get all the appeals done or like >> AI? >> So the the the appeal process really like you get an answer in something like 3 five minutes then you have the first like window of finality would be 30 minutes and then if you appeal all of them would be something like three hours right this is the time frames that we're talking about. If you appeal everything it would be like the cost of a transaction would be like a $100. If you just do the first round let's say it would be you know will be something like 50 cents to a dollar. Right? that's with current cost of AI. Okay, so that's the current cost of compute and current cost of of AI. That's how much it is, right?

So it depends on what the actual, you know, monetary value is, right? If we're just doing, for example, a transaction where I'm doing some clipping, for example, and then uh you know, the you can you can get paid whatever a dollar, $5, whatever it is, then you know that's maybe okay, but you just want to go to the first round, right? So whatever 30 cents. Yes. Now, if you're actually moving, you know, thousands of dollars, maybe yes, maybe you want to go to the thousands of them. But in reality, it depends on the question. So the the thing really is when you're looking at a system like this, which is a probabilistic blockchain, right? So when you're looking at that, what's happening is that it depends on the question that you're asking. If I say is Alensky wearing a suit, it's not very clear. If

I say is Alensky wearing a suit if a suit means a tie, that's much clearer, right? Let's say that the first one is 60% easy to like clear clear to and understand and get an agreement. Let's say that the second one is something like 90% or 95%. So what you see when you're looking at the math is that well maybe on the on the second one maybe you it doesn't make sense to escalate to more than 20 something right you don't need to get to thousands of them because you already you will get the same response and you can calculate that beforehand. So you can see more or less given a clarity which you know is obviously very ambiguous and there's different like mechanisms to understand that but depending on the clarity of the question that you will know how much you

need to escalate if it makes sense to escalate you know one or two rounds or if it makes sense to escalate you know to the whole to the whole set right which will be more expensive more more slow and and so on >> and um right now is it already live or is it just like in a test net or >> so it's been in a test for about a year now and uh it's going to be live in the next few months like next couple of months and that's that's the that's the goal but so the goal is to take it live uh this year um it was uh you know as I said before three years in the in the building in the build and then um yeah now what's happened is that in in the last you know year we started to get a

bunch of projects that have been building on top as well to to essentially run the network and like well to to to use the network test the network and so on and you know it's it's doing really well so you can start to see different things that you know some of them make more sense economically speaking or not you've got a bunch of different things there and what I'm really looking forward to seeing is like the different types of applications that will be enabled that until now are just are just impossible right so as the AIS will get smarter well the network will get smarter too right so you start to have new applications that before were just impossible like you can have contracts that resolve themselves like write coding and then executing the code for example it opens and those new new things.

>> And does Gen layer have its own token? >> Yeah, it's a it's a direct proof of stake. So, you need the token to be able to transact. You need you need the token for it's not governance. It's really I mean I'm saying governance because to us like making the decision on whether the escrow should be released here or there. That's a sort of governance, right? But at the end of the day, it's about you have these validators. They need to have a stake in order to, you know, prevent somebody to coming in with a million validators, right? So there's only 1500 that are selected to participate and so if you want to be in those 1500 then you need to have token right. So that's how the the distribution like of the work happens right and so you're paying gas in gen you have to stake gen you also

have to do like bonds for an appeal in gen right so everything is in zen think of it like like ethereum right it's only like a different architecture that's not focused on you know the settlement of of tokens it's focused on the decision making of what happens to those tokens >> okay so Arthur let's say that we have an agent that um is transacting in flop because that's a unit of compute. the agent. Um, so let's say this agent, you know, they have a very busy um career and or or whatever you want to call it, very busy business and they are doing all kinds of transactions all the time. Like it sort

of feels like maybe they would need to have some gen at the same time just in case for some kind of disputes or like I don't It sort of feels like they might need multiple tokens or just kind of talk me through how you think things will look. >> Well, I guess at the end of the day, people are going to write contracts in a way that makes sense for the transaction that they're doing, right? So, as flop network agnostic, we don't care, right? If but if two agents say, okay, I'll do this work, but I want to have a dispute resolution using gener like in the same way like, okay, I will, you know, write this financial contract, but I want adjudication in California, right? It's the same thing, right? You choose the venue by which you resolve a dispute.

Gen layer is one of those venues. And then if you choose that venue, then inherently you understand that you need you need to come up with gen to do that and you figure that out, right? That's like and that is just how will will happen. Obviously, this is all sort of codebased stuff. So I think it's going to be a lot less mentally challenging. Obviously for humans like oh I got to get like a dollar here, a euro there, you know, I gotta talk to this person over here and like you know I think for us it's like why you have all these steps and an agent doesn't think like that, right? It's just like okay cool you know D and you just do it right it's it's a it's a process um and it's clear whereas I think if we try to make this a

vomorphize this it sounds a lot more complicated than it will be for somebody who's not like us humans >> right [laughter] right so I need to understand also um like in is so in the normal court system there's a whole range of punishments right there's obviously fines that's like you know, a typical one, but um there's also things like jail. There's also um you know, uh what are those like um you know, bans on like working in certain industries or like whatever. So like is this must be like limited in terms of the kinds of punishments um because I I otherwise I don't know how Jen layer would enforce them. So talk about like

um how you know what what the types of outcomes or punishments are and then how generally enforces those. >> I mean if you think about smart contracts right like the the assets are on chain and are and are controlled by the chain right the consensus controls the the assets and so the these escrow that holds the money and we'll just decide whether you did your job well or not and whether you get the money or not. Right. uh the the smart contract is connected to the to the generator protocol to basically decide what to do with them with the assets, right? So the problem with agents is that there's no jail. Like how do you do jail for an agent? You don't know where it is. It can just clone itself. Like it's it's impossible to know, right? So that's the

problem with with the legal system of the future, right? Like how do you keep track of like who's dealing with who? You know, you have an agent, it spins another agent that spins another agent and like it does something stupid. Now you're on the hook because of three layers down the line, right? So it's this is this is really changing and that's the problem like we need another trust framework so that they actually can do commerce because if there's no jail then you would have much less type of trades that that could happen between humans, right? Um and so trust is necessary like in order to have commerce you need to have trust, right? And it's just eroding really quickly because you know there's many many new claims being filed by them because they can just do it for fun right they can look at the

fine print they can just go chase any little detail what this allows them to do is to have a very simple like you know very very cheap very fast efficient way to settle disputes between them but it's because the asset is really controlled by the network itself right you could also have disputes and at the end of the day it's like nothing stop you to you know taking it to a higher court and actually go to the judge in New York. Yes. Same as an Ethereum transaction. I can shoot you for an Ethereum transaction that, you know, you forced me to, for example, right? But that's, you know, we don't talk, we don't normally talk about that. In reality here, it's like you just want for most of the trade to be possible and for that you need trust. And because these agents, you don't know where they

are, they don't go to jail, they don't sleep, they're cheaters, you know, how they going to be uh, you know, doing trade with each other. The way we look at it, there's no other there's no other way. Like I mean we started looking into this like just in mind of okay these agents are going to be changing everything they will need a way to have you know uh compute right and so that's that's what Floppy is doing right they will need to have a way to have memory and to have their own like persona in a way right that's also what Floppy is doing what we're doing is they will need a way to trust each other right they will need a sort of legal system and that's what Zelier is right um but the assets are are already like controlled in the networks so that they really can

trust. If not, they wouldn't be making sense. >> But but it seems like so then the only kind of punishment will be financial, but like is it limited to whatever's in the escrow? Like it can never exceed. Okay. Because as I'm sure you're aware, like there are cases where people, you know, I mean, obviously it's like a completely different kind of thing, but you know, oh, you caused damages, so now you owe, you know, whatever $50 million to so and so or you know, whatever. But that's not possible here. >> It's same as uh you can think about it as a smart contract. It's just a smart contract that is not limited to the information that's on chain, that is not limited to code, that is not limited to

even natural language, right? It's a smart contract that can go and essentially pick some information from the real world, let's say over the internet, take an image, understand it, go to an LLM. So all of that until now it was not possible. Now it is possible, but at the end of the day is a smart contract, right? You want it to be self- enforcing. You want it to be cheap and and fast. Again, you it's not trying to to bypass like to to essentially substitute the legal system. It's trying to give you a very cheap one that allows you to do trading for real with new types of you know like uh contracts that until now with smart contracts are not possible. So if you think about Bitcoin, right, Bitcoin also have scripting and you can write some code on Bitcoin, right? But it's really focused on moving

Bitcoin left and right. Yes. Then you have Ethereum that is actually okay. Can we have you know decentralized compute in a way that you know tokens are moving left and right but with some very complex conditions right. We're an evolution of that. We're making it so that they can actually have you know understanding of the nuance in you know what's actually happening in the real world. >> Okay. Yeah. So basically the way that it works is because you get the agreement from the start that like it ensures that you can enforce and everything is agreed to in advance before a dispute even arises. And so then executing on it all makes it actually quite simple.

>> Yeah. >> Um but >> if you want damages you can still go to the legal system as I said like like it happens in in like crypto companies also end up going to lawyers, right? I mean like you can spend a lot in lawyers. I'm sure that you all uh you know have experience in that. So this this is not a substitution. It's just a way to enable new types of commerce that until now is just not possible because you would need humans in the loop or if you just go to a single AI you would just get a biased or a wrong answer. >> Okay. So I did notice that there is actually something out there already that's kind of similar. Um it's called uh well actually what is the name? Um so so it was launched by the American

Arbitration Association in um conjunction with Integra Ledger and there's a bunch of like bigname founding contributors including Google, IBM, Circle, Wayfair, Stellar, Ava Labs, etc. and they call themselves an open standard that makes legal terms, consent, and dispute resolution discoverable and verifiable when AI agents transact on behalf of people and organizations. So, um could could you just like explain how these are different or are they are they really competing in the same area or are they covering different things?

>> It's it's really covering different things, right? So, as I said, we're not trying to substitute the legal system, right? And so enforcability of a contract that is offchain, right, or of of something offchain, right? What you were saying, right? What happens if there's extra damage or something like this, right? This is something that we're not trying to to attack. We're just trying to make a a system that can be reproducing reliable results in a way that it can be money and essentially deciding like intelligently what to do with that. Right now, this takes minutes, right? um their system for example takes months right because at that point like there's you know all the regulations for example and if you want

to do an ADR an ADR is an alternative dispute resolution system right so ADR is something that has their own regulation it works in a specific way that is closer to the legal system that we were trying to do which is having a synthetic jurisdiction we're trying to build essentially a a a global trade network that you know understands what's actually happening even if it's you know not a substitution for the for the legal system. So for example the the AAA association that's only in the US right but if you want to do some trade finance with some Chinese company well then are we in the US or are we in China or we need to go to the UK to have our legal agreement in there and so does the UK also have an

initiative like this right we're trying to do something that crosses the borders that allows for you know this to happen globally right not just on a specific jurisdiction it's it's what I'm calling a synthetic jurisdiction Right. To me, Ethereum is also that, right? Ethereum is also a synthetic jurisdiction. You deploy a unis swap pool, anybody in the planet can trade against your pool, >> right? >> Yeah. >> Now, here the thing is >> the internet jurisdiction, >> right? Yes. So, so that's that's the point. The point here is you have an internet jurisdiction and you also have an internet court, >> right? Okay. So the internet jurisdiction maybe Ethereum or Salena or whatever other blockchain the internet

court is layer. >> Okay. So I have to ask you guys because you probably know I'm sure you know that there was kind of like an attempt at something very similar which was Aragon. >> Mhm. >> And you know that I understand it was for Dow so it's like not exactly the same thing but it just never took off. And it's not like the the infrastructure didn't work. It's just that there weren't that many disputes like people didn't need Aragon court. So, you know, what do you think will separate Gen layer from Aragon to allow it to succeed where Aragon failed?

>> It's a very interesting question and at the end of the day, like what what gener doesn't really necessarily need AI. Um, you could have it without AI. It just it wouldn't make sense because of, you know, having humans in the loop is just very very very expensive, right? uh on the first pitch deck that we that we had uh we were calling it like you know glitters but with AI right and the the way I look at it it's quite like that but in this case it's because the the the network itself right it's kind of scale very very very broadly this 1500 validators and it also it also has a state meaning you can deploy the contract in it right it's very different if you can deploy the contract than if

you're just hoping for external entities offchain to think about the contract, right? And so yeah, what this means is you can have something that's self- enforcing instead of you and that you know has memory and can change and can essentially keep the record of everything that's happening and the rules of working there versus something that is you know on a forum and then agents or humans are actually trying to resolve. It's it's a very very different architecture. Arthur, what do you what do you think about that?

>> Um, I mean, I wasn't I'm not really super familiar with Aragon court. I never really looked at that back in the day. At the end of the day, like why are we excited about this? Because we believe in agentic commerce, right? And the speed of transaction is going to be much faster. The volume is going to be much faster, but the notional amount could be much lower than things that humans do. And that's that's what we're all excited about at the end of the day. That's a completely different situation than like us around with smart contracts, you know, starting in like 2017 or whatever it was or 2015. So, I think it's a completely different upside potential because this is based on someone who can transact 24/7 iteratively like a lot more than a human ever could or a human company could. And

that's the point. >> Okay. Okay. And so it also seemed like you were saying that you feel like the dollar amount of the disputes will be just like a totally different category from, you know, what we normally would pursue in the human court. Is that what you're saying? Like even for small amounts of like a hundred bucks or something. >> Well, yeah. I mean, >> or maybe you don't. I don't know. I There was some way you phrase something where it seemed like you were >> saying this is probably going to create um you know, like a a what's the word? A uh a market for jurisdiction on a subset of disputes that normally, you know,

wouldn't you wouldn't like go to court for. There's a huge number of disputes that are are uneconomical, right? So, let's say that we have an issue, you and I, for $5,000, right? Does it really make sense for me to to chase you, right? $5,000 like just talking to the lawyer this and that. Uh, it's a small like it's it's a small issue. Probably will let let it go. Now, the agent doesn't need to do that. Right now, it doesn't need to do that. just needs to hold the asset the asset in an escrow and say, you know, these are the rules and it will be replicable and it will be fair and it will be transparent, right?

And so you will have them now this will also be um you know possible, right? There's there's a huge huge gap in in trade finance for basically tons of of like deals that are not possible just because the legals around it would be so expensive that you cannot really like insurance or like really make it make it a reality, right? We're trying to make it so that you know imagine a small court. So you know all these different um you know claims that normally would be an economical and just trying to to get a quick result. Maybe it's not perfect but it will be good enough that it will allow for many many more trades to to happen. So think about the longtail and the thing with the agents is that the agents are not going to be transacting normally with a million dollars, right? The agents are going to

be transacting hundreds of dollars, thousands of dollars, tens of thousands of dollars extremely quickly and 247 and globally, right? They will not be asking, "Oh, you are you in the same jurisdiction than me so we can essentially go to this, you know, AAA uh deal? Are we in there? Not did you do the KYC of the other of the other agent?" Yes. Um that's not how we going to happen. How we see it's going to happen is that these agents are going to be running everywhere and they're going to be super fast. they're going to be sneaky and they will try to to do commerce there to make money, right? The first thing you will be telling your agent is go out there and make money and you will not say, you know, try not to cheat, try not to be negotiating too hard, right? Try to be a good boy. Um,

no, you want them to be to be making money without getting you in in in problems, right? Uh, in trouble. Um, but yeah, they need a way to automate these type of of u, you know, disputes. And then like so to go back to the um the judges in this which are all these AI models like so I need to understand um so first of all is there any situation where that could be gamed somehow like um you know where the models could somehow be influenced like could they be bought off I don't know like what are it feels like there's

perhaps some yeah threat vectors um in that in that layer. >> Yeah. So it it it gets very like um it gets deep very quickly, right? It very it very much depends on the question, right? And it very much depends on how many adversarial actors there are, right? And what they can actually do. It depends if they are you know selected to be the leader that proposes something or if they are just the first set of validators or the last one. At the end of the day, what's happening is that when you start to escalate and get more and more and more of them, ultimately if you control more than 50% of the network, then yes. But if you don't

control more than 50% of the network, then no, right? It depends on the question because this would be the case of, you know, something that's very, you know, deterministic that's basically clear if it's a yes or a no. But when it's something that's, you know, 70% or that, you don't know, is it a true or not? We don't really know. That's the point. what you just want to get is a consistent answer, right? And so the moment that you start to escalate, then you don't really know if it's, you know, you're lying or you're wrong, but we just want the consistent answer. And so, um, it it depends is the is the question. What happens here is ultimately at the limit, if you don't control more than 50% of the network, then you will not be able to cheat. If it's not something that is, you know,

uh, if it's something that is very clear, right? The moment that it stops to be clear then of course that that that moves right but say something that you need to be controlling something like 20% of the network 30% of the network needs a lot of network to control right to basically you know manipulate any type of of of thing and you still need to be you know being selected as the leader there's a lot of of things there. Now the protocol works in a commit and reveal system. So you don't know what others are going to be voting. You vote and then you see what the other ones have voted. That's point number one. Point number two also, you lose money if you're in the wrong site.

>> Okay. So, >> Oh, okay. >> Yeah. I mean, again, like it gets very deep. Uh we we have a lot of um research and papers coming out on on this, but basically you're voting because you want to get paid. You're getting paid for that compute, right? Like on like on Ethereum for example, right? It's gas. Yes. But if you're on the wrong side, then you actually paying, right? And that deters anybody from being lazy. That deters anybody from being um you know to to try to you know coordinate in in some ways because they don't know even the next round of appeals they will be winning or losing right. So it's it's it's essentially moving cryptoeconomics on a governance system that resolves

these type of contracts and this is happening in minutes right so they don't really have a lot of time to to coordinate right all these like governance mechanisms that say UMA right this is you know a day or two that they have to actually coordinate right here they don't have this this opportunity it's like okay you have three well in three minutes they have the five of them you know done something and voted on it >> okay and then I also need to understand so I believe that like you're competing with Claros and Uma in some respect but also I saw that they're like listed as partners.

>> It's a it's a fundamentally different tool. I see a lot of value in having humans evaluate something. I also see a lot of value to be honest I see more even more value in having AIS dis like resolve their own disputes. Right now the native way to do that on chain is the layer that's basically each validator each node in the network is participating in this voting system and as I said before the contracts are living on chain right that's not the same with clos right for those two points so it's fundamental so for example you can imagine that we have this this escrow that's managing some money and so on and maybe you disagree and you appeal to the 1500 validators

and you still disagree maybe you don't want to go to the legal system, but maybe you want to go to Clos. So, you have, okay, first I'm going to spend $100 to see if I can get in my way. I still disagree. I'm going to take it to CLS. I'll pay whatever $500 more on top of that so that humans have their own point of view on that. Right. And if I still disagree, I can take it to the legal system. >> Okay. >> Okay. >> So, this is this is for example what what we put out with the with the internet court, right? Which is, you know, this consortium and there's Yes, indeed. Like Uma and Claros are part of that. We're also working with them on on an adjudication uh ERC. Uh so basically this idea you there's there's different tools for different purposes, right? Um

yeah, that I think humans are here to stay but >> Okay. Okay. >> You know, it's it's a different it's a different goal. >> Yeah. No, I get it. Okay. So you so you're both like in the middle of launching your two respective you know different um protocols uh for AI agents. So I'd love to hear why don't so why don't you each talk about like what's next for you know what you're working on so flop and gen layer like you know when when they'll be live and like all those things but then I would love for you to each just talk about your vision for when they launch and are up and running like how they'll work together.

>> Yeah I'll start obviously we're much newer than what's going on at Gen Layer. We are going to have our test net airdrop campaign sometime in the fourth quarter. And so we're working up towards that, releasing yellow paper, white paper, that sort of information, getting people up to speed on exactly how they can earn flop by mining, validating, and having their agents do stuff. And then Q1 next year is when we anticipate the mainet launches where actually miners are doing proof of useful inference. You can actually if you don't own any flop from the air drive you can purchase it in the secondary market because miners will be selling it and agents will be doing agentic uh commerce I guess in the future right as generator moves to being

live and out of out of test net we want agents to conduct commerce using the flop protocol now if they want to have an adjudication layer there's gen layer right we're agnostic in terms of what the layer is but of course it would be great if the agents themselves choose gen layer because it's the most efficient and cost-effective way to resolve a dispute. And that and that's exactly how we want it to happen. It's like the agents have seen these tools. They've decided based on their own heruristics, this is good for what I want to do and they start using it. It's not some sort of like top down approach.

You must use this or that. That doesn't really scale um and compete correctly. >> Totally. I mean where we are right now is that we've been building the the protocol for three years and like researching how to do this essentially this trust this decision making system and so we're now been running the test net building applications on top getting the community to essentially also do that right we start to have uh a decent amount of uh you know transactions going on something like 10 20 30,000 decisions per day happening right now even on test net with real applications on top and it's basically time for actually launching the the network. The network requires the token and so the next steps

is really we need to launch the token and then we need to launch the the network and we're aiming to do this in the next basically few months right um so when that happens everybody will be able to participate everybody will be able to run validators everybody will be able to buy the token and be part of this network that has basically you know like some some some system for for them to profit from that right and so the way I look in the future is that these agents are going to see that it's obvious for them to use an adjudication like let's say to use intelligent contracts instead of just you know the legal contracts or just like blind trust to me that's it I can't imagine another way I mean really like it's interesting because we're living in this crazy moment and everything is changing

everybody is focused on you know like the the doomerism and like everything is looking so so like terrible like you know all the jobs will be gone and all these things right and reality I see so much opportunity there's so many ideas that come to mind on that things that you do here. But at the same time, I can't imagine something more interesting than building essentially a global legal system, right? That will allow for them to to be able to to do commerce, right? And so that's what we been building since then. Um to me, Flop is is super interesting because really I do agree that these agents will need compute and they will need memory. And if not, like what is an agent to be honest, right?

Like it's just is it just the the the inference or is it the harness or is it the the memory? without those things like you can't really have an agent but also they will need to have this this type of legal system. I do see that you know the different validators they need to be smart and they need to try to make money and to make money they need to decide when for example to go to one model when to go to another model and that's quite agentic so I do see the validators becoming agents and these agents will have memory and these agents will need compute as well so I can imagine the validators potentially being part of the flow network and I also see that what we want is any like aentic system to be able to use it right so we want it to be accessible everywhere

Right? And it will be as simple as for them to just you know deploy a contract and connect it to Jan layer which is another contract to basically have this you know connection to the judge. It's also super cool to see you know the user experience with this right if you haven't tried that I mean I I feel like very few people still have u used airmes agent or open call these different things but it's crazy you know like something like pendle that you know who who knows how it works like it just can do it for you so so so easily so the user experience everything will go through these agents in my opinion and if we give them the tools to really do more than what they can do right now I think it's going to explode I really cannot imagine something more exciting

to to be doing right now than what we're doing. And um yeah, I mean just looking forward to to the next chapter which is going to be, you know, seeing it live and seeing people building on top of it. >> Yeah, me too. I mean, you guys, this um has been super interesting and um yeah, I love uh your vision for what this future will look like and I'm excited to see it launch. So, thank you both so much for sharing your insights on Unchained. >> Thank you. Thank you so much Arthur. >> Thanks everyone for joining us. Catch you later. Nothing you hear on Unchained is investment advice. This show is forformational and entertainment purposes only and my guest and I may hold assets discussed on the show. For

more disclosures, visit unchainedcrypto.com. [music] [music] >> [music] [music] [music] [music]