LIVE: NEAR Day At Consensus Miami

Transcript

a little bit more uh pricing power. But um but I think is to uh is to scale up alongside that. Um and some won't and some will >> and agents only accelerate that, >> right? That's a great point, >> right? Like it they're going to go find the cheapest, lowest cost, most efficient way to do things. So that that's going to be one of those forcing functions >> and and less likely to be a captured customer because they don't care about the inertia of opening a new account. the same way that you would conventionally think of some of the the frictions of just human time and effort uh allowing people to be captured.

>> And they don't like our beautiful branding. >> Yeah. >> And they don't give a about, you know, all the work that we put into those things. They want, you know, they're going to optimize for whatever they've been told to optimize for, which is generally going to be cost and security. >> Yeah. Uh couple teams, few teams that have uh it feels as if they found the efficient frontier. Uh there's this threeheaded dragon that seems to be brewing, which is this hyperlquid Zcash near three-headed dragon. Just curious if you guys have a perspective on why we may have seen this cohort of assets really take charge of the market in the last, say, call it a month or so.

>> I mean, best description I heard of hyperlquid was it was like you took a bunch of engineers and you put them in a room and you told them to like rebuild financial markets. Ignore regulation. ignore like all the rules, all the things that Hunter and I are, you know, the older people in the room, like we we grew up having to deal with these things. Just forget all those things. And I think, you know, that's a I just I love what's happening there. U it will run into a regulatory wall. It will, you know, there's there's going to be some some limitations on the on the adoption there, but it's exciting to see, you know, innovation approaching in that way. Um near for me, like intense was the thing that that brought me in. That was the capability. I love Ilia's vision on on AI and agents, but intense was

really the thing when I saw that I said that is something really special and and different and it um it's you know it's the the chain abstraction the the ability to sign transactions on other chains like that's such a powerful capability um to have you know to get rid of that boat anchor of of crypto steals that I have and to be able to do everything just in in near.com. Um, and then you know, Zcash is just that that theme of privacy. And I wrote about this for the Near Foundation board back in December, just the lack of privacy being a constraining function on institutional adoption in our space. Anytime you're making a payment, not just investing, I'm not just worried about people

frontunning me or watching what I'm doing. Anytime I'm making a payment, anytime I have value moving in and out of a wallet, it's discoverable. You know, I you've traded with OTC desks. the first time you trade with them, it's one kind of transaction. The second time, well, they know all your wallets and they know all the transactions. They could see a whole lot more about what's going on. Um, you know, so I think Zcash is a great example of of the way in which, you know, privacy as a as a concept is so is so critical. >> You you you know, do you know the uh do you guys know the Midwip meme? It's a normal distribution curve and there's like a Jedi and a frustrated person and a Neanderthal. I think I think another thing that you said.

>> No, no. I'm No, no, no, no, no, no, no. I'm trying to buy two sitting in the middle >> too. Yeah. Yeah. Yeah. I'm I'm like the frustrated, you know, uh overthinking it. Uh but uh the thing that I would just add I think what you said is correct. The other thing I just say is um um something that those three have in common is they have uh smart, capable ecosystems and teams around them. And they're, you know, there's no force more powerful than being on the right side of a structural trend. and each of them is in a lane where they're at the forefront of uh of a wave that is forceful and cresting and um and uh and that's a very powerful force. So um I think being in the right position to do something that

the world thinks is useful um and uh is is true of all three and and is important to the success of of um of any asset or any network. And and so I think that that's also a theme for those. >> Yeah, they all feel very fundamentally uh part of this agentic movement of liquidity on chain, liquidity, efficiency, privacy uh and agency with all these agents. Guys, thanks so much for uh for this panel. It's really really fun. Not the first time, hopefully not the last. And uh we're going to keep things rolling here at Near Miami. Give it up for our panelists. >> Thanks, Robbie. I'll turn to Matt. My lovely co-host at the rollup is gonna join here for the next one and we're going to keep things rolling. Thanks,

Matt. >> Microphone's gonna fall. >> I got you. I got you. Um, next up we've got my friend Benjamin from the Starknet Foundation. We've got Nigel from Bridge, Kendall from uh Proximity, and Richard from Moonpay. So guys, uh who pays when AI acts? I'm getting the signal that I'm the moderator for this one. Andy, I think that's uh I think that's the case. So, I'm gonna I'm gonna stick with you guys a little longer. Thank Thank you.

We got room for everyone here. There you go. See you. All right. Who pays when AI acts payments today and the future of agent driven transactions? Last panel you guys heard a little bit about as that movement of liquidity ends up onchain both you know onchain rails and agentic rails. Now it's like all right now that it's onchain what happens next? Um why don't we do just like one sentence intros? We don't do, you know, super long-winded intros, but we've got a plenty of people on the panel here.

Uh, Ben, start with you. We'll we'll move down the line. >> Hey guys. Um, I'm Benjamin Strrisky. I am on the strategic projects and investments team at the Starknet Foundation. >> I'm uh Kendall, founder of Proximity and we basically do stable coins for Near Intense. >> I'm Nigel on the go to market team at Bridge. Uh, we do stable coin orchestration, issuance, and stable coin link card issuance. Hi, I'm Rich from Moon. Um, I VP for payment partnerships and banking. Um, so effectively look after everything which is on the fiat side of crypto. Um, so uh yeah, I'm a bit more oldfashioned traditional banking guy than the crypto guys.

Um, I uh I think for at least a little while longer we're going to have a a fiat system. Fortunately or unfortunately. Hopefully we can move all that onchain at some point and then we'll we'll rethink you know what the on-ramp situation looks like. But for now, uh, we're still moving between fiat, we're moving between onchain transactions. Um, Richard, why don't we why don't we start with you? You know, how does the mo move movement of money between off-chain and onchain rails? Can agents basically abstract that away? Because it feels like if you're in one environment, if you're offchain, you've kind of got your set of services. If you're onchain, you got another set of services. You can kind of work pretty efficiently in these, you know, different environments. But if you have an agentic layer, does all of that get

subsumed and you can translate a little bit more easily between off offchain and onchain? >> Um so I don't necessarily think so. No. Um because um whether we like it or not, we are still very much relying on the banking like sector, right? Um and the problem with the banking sector is that um it's built on humans, right? So humans are the ones that effectively uh authorize a transaction or liable for a transaction and so on. Um yeah with agents like uh you know to to to work on the processing of transactions on behalf of you got to build like a delegation layer right in there. So you have to set limits and you have to set authorizations and so on. So I don't necessarily think that like you know an agent will just take everything away um because you don't necessarily want them to right you still want to have some

sort of level of control um in there but it's like how you then uh instigate that control in the agent. Um so it could take away 80% of it. >> Got it. Um and so guys, how are we thinking about and Ben like you know maybe maybe from the privacy perspective the start perspective how are you thinking about stable coins and agents you know as they move onchain and we'll kind of flow between our our you know onchain offchain partners here uh with Kendall and Nigel. >> Yeah. Um so I think if we really want to see a lot of stable coin activity and payments activity um onchain that it's going to have to be private. Um, you know, you look at traditional finance, almost everything is regulated and compliant, but you know, if you pay

someone through Wells Fargo, um, no one sees your payment history. When you're receiving your salary, no one sees your salary. Um, and so, you know, almost every public blockchain today is public. Obviously, we haven't really seen payments take off in the way that I think a lot of people would have assumed 5 years ago that in 2026, people would be receiving their salaries on chain. uh people would be paying other people on chain and it happens but not to the extent that we thought it would and I think privacy will have to be the biggest unlock for that. >> Got it. Um and how do we fit agents into this as well? Like as we bring tokenized assets on chain, it feels like we've started with stable coins. We've now got treasuries. We're kind of moving down the risk curve. We're token tokenizing

all of these assets. when we end up with a place where we can effectively earn yield at, you know, the drop of a hat, uh, more stable coins are on chain, both yield bearing and and non-yield bearing, you know, where do you think agents, because it feels like we're actually building more and more products for agents, where do you think agents are going to prioritize their cash? Maybe they're, you know, what are they holding uh, when it comes to these payment currencies? if they're sending money between agents and agents, what what do you think are the attributes that they're most looking for in stable coins? Is it privacy or is it maybe something else?

>> I mean, access more than anything. Um, and I think that's sort of the big problem like the the infrastructure is starting to really get there, right? I mean, link did a pretty cool like launch last week um that is the first practical way that I've seen to pay with agents for like services you actually want. That being said, like cards are, you know, not the best route for like micro payments that probably are useful way to pay for certain API calls. But then the problem we have right now is like there's not MPP is really cool, X402 are really cool. Uh and there are some services that are starting to offer their APIs, but uh it's early, right?

Right. And so I think as we see basically just more adoption of those rails and more businesses that recognize that they'll actually have a lot of growth if they have a good product offered to agents and the friction is very low both like on-ramps uh actually holding the balance in your wallet and then having services that you actually want to consume as an agent or as a user using agents like that's all those things kind of have to come together and it's moving very fast. So I mean probably we're talking months but uh those pieces are still not really there today.

>> I would I would also add that I think uh you know the same way that AI has made producing reports more efficient, producing presentations more efficient that these AI systems when they're making payments when they're using DeFi when they're just operating on chain they're going to want to do that in the most efficient way possible and really the most capital efficient way possible. So, they're going to be designed to use the cheapest networks, to use the products with the lowest operating costs in order to make as much money as possible. >> And I hate to use a swear word, right, but compliance, right, comes into it. >> But, um, you know, you're talking about like agents on cards, right? Um, you know, that and and I agree, we're not anywhere near like how seamless and and UX it needs to be, right? Because even

if you like we launched an agent card last week, right? But we still have to have that 3DS authentication layer built into it. So, you can tell your agent to go and buy you something, right? And it can scour the market and it can find you the best possible price and so on. You still have to do a two-step authentication, right? When it finds it because there's a card attached to the agent. So, like, how do you get through that without every single financial institution blowing up overnight? >> Nigel, I want to get you in here. Give I I know your your wheels are spinning. I want >> Yeah. Yeah. Wheels wheels are spinning. So, I think a few things to to your previous question. Um, and I think, you know, one kind of interesting development in the market is there are more and more stable coins that are

launching. And if I'm a user, there's probably three things that that I care about. One, you know, first and foremost, is this a safe asset? Is this actually worth a dollar? Two, what sort of upside do I have from holding this? um whether that's accessing yields through DeFi or maybe you're using you know a bespoke stable coin that some company has issued and you're kind of in their in their ecosystem in that regard. And then three is what can I actually buy with this? Um and and that step of hey if I'm holding you know USDT on a certain chain and this merchant um or you know the agent that's receiving the

payment wants to receive a different stable coin on a different chain. How does that conversion and interoperability happen? Um, and ultimately that needs to be abstracted away from the end user. >> And at what cost, right, as well, right? Because I mean when you're doing any type of transaction online, you have to be super visible when it comes to the cost, right? So the individual has to know how much is it costing me, right? To to do that. So to your point about, you know, agent A talking to agent B, right? Agent A wants USDC, agent B wants PYUSD or something. How who is doing that conversion, right? How much is it costing? Um and and that's where I think again horrible word but compliance kicks in right because you have to trace the

transaction all the way back to the beginning right you have to be able to see who initiated that transaction so agent to agent to agent to agent if we get to that point right it masks like the transaction layer completely and and I think that's where like a lot of banks just get super scared right because there is no accountability right or well there isn't accountability but it's like really muddy right in terms of accountability. >> So, so Richard said two swear words so far. So, I want to I want to challenge actually I and and hear everyone else's opinions on this. >> Maybe agents don't care about a compliance. Maybe maybe banks care a lot about compliance, but maybe agents are so innovative. They don't care about branding. They don't care about compliance. They care is this thing secure? Code is law. Am I am I do I see

any smart contract vulnerabilities? No. All right. I'm gonna I'm gonna use that thing whether it's compliance or not. Tornado Cash, Thor Chain, like these things are, you know, they're they're great for North Korea. They're not particularly compliant. You know, are agents going to use these rails? Guys, I mean, I'm going out on a limb here. I'm curious what do you guys do you guys see this on one way or the other? How much do agents really care about compliance? >> Uh, I can start and I'm gonna take out my suit here and agree with Richard that compliance definitely matters. Um, I think ultimately it doesn't really matter what the agent hereby cares. It's who controls the agent. And if we want, you know, mass agent adoption, the the reality is that in the United States

there is regulation, as there should be. Um, and if we want major payment systems to be interoperable with agents, so if we want JP Morgan, Chase to actually be willing to, you know, use these these systems, then they're going to have to just be compliant. Um, you know, crypto has a long history of unbanking people and we don't want the same thing to happen to AI agent users and so they're just going to have to be compliant with the law. >> And just to add on to that, I think, hate to say it, but it's a relatively easy fix, right? As well, because agents are programmable, right? So, you program the permissions into the agent, right?

And the permissions you program into the agent are is career safe to to transact in, right? Should because the agent is always going to go out and do what's right for you, right? because that's ultimately what it's programmed to do. Um, but you can program limits like screenings like you can program use this asset over that asset like and so on. You can program the agent. So, as long as we get the programmability into the agent, right, then it shouldn't be a problem. I'm also really interested to see how the relationship between kind of the the KYC and the identity of of the owner of the agent and the kind of risk and reputation and and trust basis of that

agent that's out you know conducting commerce in the wild. how those two um will be related over time because if you are going to have a human that's associated with an agent um then you can kind of get some of that reputation and compliance coverage that way. But you could also imagine a scenario where you know someone buys your a fraudster buys your KYC information off the dark web KYC's links that identity to an agent and then you know prompts that agent go out and you know scam as many people as you can.

So I think it's like a a new kind of like vector of risk that we're all going to have to figure out how to navigate together. >> Yeah, that's a good point. I guess it's like you can already still I mean you can already buy KYC on that ARPA but the things you can do are like maybe less damaging than they are when you like have an agent that can go out and operate autonomously. Um yeah I mean I think there's even like there's the compliance aspect but I think businesses also do want to know like not even purely for compliance reasons. They want to know who they're interacting with because that helps them make their product better and it it helps them like re retarget and retargeting agents will be a thing, right? Like you'll you'll have an agent use you and you'll be like

we want to we want them to keep coming back and using us and I don't know what that's going to look like. People are going to figure that out. So I think there's going to be at least some like some form of agent identity like for compliance reasons for financial transactions but also just for for business reasons like people are going to try to enforce some type of like account and agent identity just for their own you know product development >> when it when it comes to adoption I'm curious what has worked so well in in terms of you know getting stable coin adoption getting user adoption and and I'm curious because I'd imagine most of you know the businesses and you guys have geared that towards humans and I'm just curious if that carries over to agents. So when you're, you know, looking at adoption, what has worked so

far and do you think that will carry over to agents? You know, Moon Moon Pay's got quite a bit of users here. Maybe we'll just go down the line. >> Um, yeah, I think um, so what are we doing so far that like works, right? I think trying to make things headless, right? I think um, would would be massively beneficial to this space. I think honestly like we are literally tip of the iceberg right when it comes to to agents. So I think know we are figuring stuff out as we go. >> When you say headless you mean like maybe you can just >> just like so Moonay agnostic right. So Moonay doesn't appear in like any um environment. So completely headless for for the individual.

>> Um and it's it's something that we're doing across our business as a whole as well. So um yeah when you plug Moon Pay in for other ons we're making it a headless experience as well. So it's unbranded. They don't know that they're using. >> Correct. Yeah. Yeah. Um but I think also um like we've touched upon before which is um just setting limits, right? Um you know spending limits, asset limits, like velocity limits, right? As well. Um you know, so your agent can't just go off and just spend a thousand things, right? In in the space of five minutes because it can do that, right? A human can't physically do that, right? Agents can. Um so I think like we're just figuring that out as we go. Setting in these sort of parameters. Um, and that's why I saying like agents being pro programmable helps. Um, but yeah, we're

just figuring out as we go. >> Yeah, I think um, specifically as it pertains to, you know, Agentic stable coin payments, probably the thing that's working the best so far is the the narrative and the hype. Um, like all like all things. And and that's a cheeky answer, but you know, um thinking about the uh stable coin payment use cases that now are becoming more and more ubiquitous, crossber payments, dollar access, etc. A few years ago, you know, they were uh uh powered by kind of narrative and hype as well. And ultimately what it comes down to is, you know, especially for any business that's

um coming more from like a web two direction, ultimately the the product experiences um that they unlock for customers and for their businesses need to have enough value that they can get over that hump of of getting comfortable with some of the uncertainty and some of the risks of leaning into, you know, uh uh a part of the industry. where where we're we're kind of building the plane while flying. >> Yeah. Um I agree with all that. Um yeah, I mean I think like where so it's kind of counter to the idea of agent authentication, although I think that'll start to get better, but like to to your point, uh Richard, I I think that the low friction experience is going to be

like the the products that see the most agent adoption. I mean, first and foremost, they're going to have to work. Um but second, it's going to be about low friction, right? Like a common thing if you're using agents today whether it's coding or for anything is you you run into some blocker and that blocker is usually some type of like you need to sign up for an account and there's not an Asian friendly way to do that. Um and I I expect that I mean one of the nice things about onchain is once you get over that hurdle of getting onchain which fortunately is getting much easier I think people like me pay and bridge um then everything's actually extremely low friction and so I think as I'm very optimistic that the friction of getting on an offchain will will go as close to zero as possible always be some but as

close to zero as possible and then agents will want to use those products because they'll be the lowest friction option in a lot of cases. One one interesting point you made there just to add on to it is like you mentioned that case where you know you're talking to a chatbot and then ultimately it needs to sign up for something and it can't and that blocks you. The other piece of this is going to be the existing infrastructure and how businesses actually like cater to agents and and design their sign up and onboarding flows to fit into some of those constraints that they have.

Yeah. So, to finish this off, from our perspective, Stark's perspective, it's about creating the best system possible for the AI agents. Um, and so there's a few different parameters there that matter. Um, the first obviously is throughput. Um, AI agents are going to need block space on Starknet. That's no longer really a blocker. Um, we can support as much as we need to um given the current demand today, right? The the second um I think is going to be network level uh configurations. Um, and so everyone has probably seen the jokes about Open Claw, you know, spending my entire balance on some random nonsense, right? And so that actually can be a real issue. And if you give an AI agent a stablecoin balance, I would I would think it's actually not your money

anymore. You have a very high probability of receiving it back, but it's actually secured by the code obviously of the of the agent. And so it's the code's money. Um, and so, you know, Starknet supports native account abstraction, which enables you to set spending limits, other other logic that, you know, sends a little bit more of the control of your funds back to yourself. The the third from our perspective is privacy. And I've touched on this a little bit already, but AI agents um are going to replicate what works in traditional finance today for payments.

And the reality is that payments need to be private, not public. And so with our privacy launch next month, you'll be able to have privacy for any asset, which means privacy transfers for stable coins, which then means private payments for AI agents. Um, and the last one, and I think this is a lot farther out, um, you know, is decentralization. And while that might not matter too much today, um, later later stage, AI agents aren't going to necessarily want to operate on a network where someone can rug pull them. Um and so I think from our perspective a path towards getting towards being a decentralized network makes us very attractive for AI agents.

>> Uh sticking with this notion of privacy because I I was kind of talking about this three-headed dragon uh on the last panel and you know Zcash falls into that and I think there are elements of of near's AI agent stack with openclaw ironclaw that also takes into account this privacy hyperlquid not so much but you know we'll maybe we'll see down the down down down the road. Um you know you you guys deal with a lot of businesses do do I know I know starknet's perspective privacy is important you know Nigel maybe from the bridge perspective that you know and and moon pay and proximity as well from the B2B side how much do businesses prioritize privacy when it comes to integrating stable coin payments.

>> Yeah I think it uh it evolves kind of as they get deeper down the the rabbit hole. you know, first they they'll look at stable coins and see like what benefit can this this integrate, you know, integrating stable coins, what benefits can it bring to my business? Um, and then at a certain point when they're going through and, you know, let's say you're setting up uh, you know, a use case where uh, an employee is getting their salary streamed to them that's tied to, you know, a wallet that has a stable coin linked card. They go and do all of their everyday spend. at a certain point that company realizes hopefully before the end user like okay there is you can see how much money this person is making and everything that

they're spending it on um and that's obviously a big problem so I agree like privacy um is over time going to become table stakes um and it's also important as we're thinking about like different privacy protocols how does that fit in and become you know not mutually exclusive with compliance so it's a matter of making like the entities that need that information for regulatory purposes can get it. Um but it's kind of on a need to know basis where not everything is just out on chain. >> But that's where like tokenization comes into play, right? Because you can share data, right? KYC tokenization, privacy tokenization. Yeah. And I think um you know building out that functionality like would make things sort of super

frictionless as well. Um so yeah. >> Can I? >> Yeah. Um no I mean I I think that's exactly right like that very much step one is like that they get some other benefit from using stable coins and then they realize I mean so I think an interesting uh additional use is like when businesses are doing a lot of their payments primarily crossboarder right like that's where the the demand is right now uh and they realize like oh this is super public so like everyone could see all these you know crossber payments that I'm putting out there that's not like that's not how they want to operate their business right so like yeah very much like it's a it's a second order uh demand but uh now that a lot of businesses have gotten over that first hurdle demand it has become extremely

important and so we have seen a lot more and more demand and and does that happen at the chain level or you know I kind of mentioned this on the previous panel as well this asset centric view of the world can you integrate that directly into the stable coins themselves you know I'm I'm seeing more of that with uh like RWAS you have KYC baked in a little bit compliance being the key word there but uh yeah when it when it comes to you know some of these these privacy elements. Do you see that integrating at the asset level, the stablecoin level, or more so the chain level? >> I mean, I'm extremely biased here, but I I'll plug near just here is uh is um I mean, I think it's the challenge is that currently there's actually assets across mo like many different chains that uh

are doing very well. Uh which is a great state of the world depending on your view. Um the challenge then is like it's actually kind of difficult if a business has to operate across multiple chains. Then if you have single chain privacy that starts to become really difficult. So the way we've done it because our whole view of the world is like we just care that you're using near intents which actually connects a bunch a bunch of chains together. We're able to implement privacy at like at the near intense level. So you can move between these assets on different chains while staying within the private environment and keep that whole interaction private. >> So everything on near intense is private.

>> Everything on near intense is private. You have to right now you have to opt into it but eventually that'll be default. You can opt out of course there are some actually for regulatory reasons there are some uh institutions in particular that don't want privacy but uh the average business seems to want it. >> Um few minutes left here maybe just zooming out a little bit I'm curious to get you know you know your your long view here on you know everyone's long view here. Um there's been a few smart people that have said you know AI is the uh the saving grace of crypto. Bless you. Uh I'm I'm just curious if you guys see this world too, like you know, we obviously had we had we've had a lot of problems here with crypto. It feels like now we have AI agents. Are are they

going to become the primary users of these tokens? I see Nigel's wheels spinning. I I I can tell you've got something cooking. I think it's a it's a more um like logical form factor um for for machines to use stable coins and you know uh I guess to to Ben's point earlier where like a lot of the activity and kind of the rules of the road are dictated in code. It just seems to make a lot more sense. The reason why I'm like hesitant a little bit is I also don't think um AI should we should be

viewing as an industry AI as it's going to like swoop in and only the machines are going to use crypto like this should not be mutually exclusive from us onboarding the next billion people on chain as well. So my answer would be both. >> I I completely agree with that. Um, and so I don't think uh AI is going to be crypto's saving grace. I think that a lot of smart people in crypto have been preaching for a long time that crypto is a more capital efficient system and if you're a rational actor, you should be using it over the traditional financial rails. And AI by all means should be rational. And so if AI agents are choosing to use crypto because it is a

better system, then that's validating what people have said previously and then it should usher in the next wave of true adoption. I think I mean just to build on that I think it it marries together really well right so digital assets AI agentic like it's all programmable you can put you know limitations in place and so on I still think there's a question mark right over and I I think you touched upon it earlier about like liability because you know again going back to the old fashioned way of paying right you make the payment you are liable if an agent steps outside of its parameters and buys something which you necessarily haven't agreed with or is in a market that it can't operate and so on, who's liable,

right? And I see that there's going to be a whole other world of legal cases, right, where people will be like, I didn't do it. My agent did it, right? And it's just like, okay, well, you're the human behind your agent. You told it what to do. And then then somebody be like, prove it. Right? So, I think there's like a whole argument around it wasn't me, your honor. Right? it was like the Asian did it. Um, and and I just foresee like that happening at some point. And we'll stamp that out, right, pretty quickly, I think. But I think that's going to be a bit of a headache in the future.

>> And then it's a question of if it's human lawyers or AI lawyers. >> Exactly. Yeah. Yeah. Yeah. Yeah. Then you're paying an agent to represent you in court. It's like, what the hell? Like, >> well said. These are these are Oh, yeah. Kendall. >> I mean, not much sad other than like agents are definitely eating everything. They'll eat crypto, too. Crypto would have been fine without agents. I don't think it's a saving grace. It would have been fine, but agents are eating everything. Thus, like eventually the average user of crypto will for sure be an agent and maybe pretty quickly. >> Um, we got a minute left. Any questions from the audience? You guys curious about anything that we haven't covered so far? All right. If not, I've got a question for you guys. How who is currently getting paid in stable coins?

like from their business not receiving fiat but you you're receiving stable coins. I see one hand. Guys, that's an issue, isn't it? >> From like Yeah. I mean, we're at a crypto event like, you know, there's tons of stable coins. What it feels and and I think we almost everyone on the panel mentioned, you know, payroll as like the catalyst for stable coin payments. We've obviously got a lot of work to do. It feels very early. I mean, who's doing this? It feels very important. How do we get this off of one? Right? We're not at zero. We got one the guy in the orange shirt.

Shout out to that guy. >> We um so so we work with deal, right? So we we powered deal, right? So the employment of record business and they are paying their contractors in stablecoin. Like the problem is and I said this on a panel earlier, right? We are not there with the infrastructure yet. So you can get paid in stablecoin but like could you pay your rent in stable coin? Could you pay your mortgage in stable coin? Unless you have a stable coin back card which you can then like work and it can be a little bit expensive like the infrastructure is just not there to offramp that stable coin one to one you get paid in it but then what do you do with it right so I think once we build that infrastructure it will be easier and everybody will get paid >> yeah I mean we built out all these

onchain tools and it feels like we've actually built maybe like too far ahead right but we haven't thought like you know we've got to get these stable coins into the people's hands you know from the onset >> I think there there are a lot of people like we pay contractors in stable coins um And there are >> so you raise your you didn't raise your hand Kendall. Well, I don't pay your contract because they're non US, right? And like that's the reason, right? It's like and I mean we're not a US company, but um it's like it's easy for international payments, right? Um so yeah, I mean I think that's there there are a lot of people in crypto particularly, you know, if if a if a company is operating like they have contractors all over the globe, which which is very common, then there are a

lot of crypto companies paying people in stable coins. I think the point is like crypto people are willing to get paid in stable coins because they know how to offramp like they have exchange accounts like for them they're they're fine with the friction whereas like the average person who's working at a normal company that doesn't have anything to do with crypto for them the friction that is a problem like if a company says hey we're going to pay you in stable coins like what that's horrible like why are you going to do that to me and that's that's the missing piece like >> yeah and then there's this whole rabbit hole of tokenized deposits which you know maybe could help help on that front but We won't get into that today. Guys, give it up for our panelists. Thank you so much. That was a fun one.

We're going to keep things rolling here. Pleasure guys. >> That was fun. >> Yeah. Feel like I haven't laughed on that much on a panel in a while. Good. >> Yeah. Good to see you, man. Yeah. Thanks for Thanks for uh being here. Yeah. Good to see you again, Kendall. Ben, go get some lunch or something. This one. >> This one. Yeah. >> All right. >> What's up, George?

>> Good to see you again. >> Yeah. >> All right. Andy's taking over. We got the stream figured out. >> Yeah. Good. >> Beautiful. Enjoy, guys. >> All righty, guys. Uh, welcome to Whoa. Welcome to the AI financial pilot panel. the real first the first real agentic applications George Zang of Nearai Nishkcom of Abound uh quick introductions guys welcome to the panel >> good to be here um I'm George uh chief product officer for near and also GM for nearai >> hey I'm Nishk is great to be here too I'm the founder CEO for for Abound we're a mons app for Indian expats in the US to send spend save invest and borrow

seamlessly across borders >> awesome and I know that there's some news to cover today and some and some partnership. We also have a cool uh a demo video for you all. Let's get into just kind of the laying the groundwork for uh the first real Agentic applications kind of what what can we expect in the in this uh application landscape as it's evolved so far looking ahead what are some of these exemplary first real applications? >> Uh sure. Yeah. and and uh you know I I'll just speak to that and I'll probably give you guys a little bit of a background for Abound just to put into context. So you know we do money remittances for Indian expats from the US to India and you know just for for

some context you know Indian expats are like a virtual G20 economy unto themselves there's close to you know like 33 million Indian expats around the world. It's an audience which is the population of Australia. they collectively earn two and a half trillion you know as as an economy they collectively remit about $120 billion dollars um into India every year so it is by far the largest movement of of consumer payments just around the world um and and as a pound you know we actually want to make it easy uh for money remittances into India we ultimately believe that every dollar that's being sent home is being used for something it's being used for family maintenance it's being used for education and and so far all the

existing incumbents within the ecosystem effectively they're trying to cater to every demographic which they can whether it's Indians and Hispanics and Asians what have you and in doing that they've lost the ability to go really deep and super serve those users so as somebody who's an Indian expat myself in the US I should be able to invest in a mutual fund in India with one click but it's really hard for me to do so and often times the friction within that experience sits in the last mile and so for me as an expert when I send money to India I have to stay up at 12 midnight and and speak to like a banker in India when I have to do anything whether it is moving money to a mutual fund or it is to you know to do any other financial activity or even beyond um you know and

and so for us agents solve a very real problem um and and in sort of me staying up till midnight can my agent handle that for me and so what you'll see in the demo is is just one snippet and it's it's our jumping off point for now doing agents. Um, so if you want to play it and I can, you know, just get give a voice over as we go along. So what you'll see on screen and I'll just step aside for a sec is this is the homepage of Abound. And if you want to keep playing and and so within the homepage of Abound, what we're letting the user do now is access their agent. And this is going to go live in beta as of this month. And and we honestly think this should be the largest consumer application for an ironclaw out there. About half million

iron claws which should go live between now and the rest of the year. And what you're seeing over here is just some real world applications of what you can do with an agent. Um it'll help you maximize the exchange rate which is really impactful for users. Um it'll let you manage a few use cases. Um and there's an actual video like we recorded this morning. You say hi and then as you go in, you know, the agent will tell you what you're looking to do. You'll say that you're trying to send money to India. Um, and you know, we'll say like I want to send a certain amount back home. And once you do that, what's happening behind the scenes and this is an actual demo with actual ironclaw behind the scenes is that it's processing that request and then it'll ask you a bunch

of questions. Who do you want to send it to? Who's the recipient? Uh, we also take some consent from the user as we do it. So in this case, I want to send it to Brashant who has engineering for us. Um, and you know, and once the user selects his name or even to me, um, although I feel like I'm I'm not as handsome as the image on screen, but you know, once you click on that name, it'll start to process all all the information and and this is the piece which happens behind the scenes, which is really critical. Anytime you send money to somebody in India, you have their KYC, you have their passport, you have, you know, like their bank account on file.

And this is where the magic of ironclaw really happens is that you're taking really rich PII, right? This your passport, your driver's license, your social security number, and this is being cryptographically sharded. So the LLM or you know the the the the like the language model will never ever see the actual private keys and it'll be cryp it'll be cryptographically sharded out there. And then as you do that, you're seeing really rich PII get processed. You're asking the user for what is this going to be used for? that's mandated from a compliance standpoint. And in this case is, you know, it'll be used for family maintenance, for example. So, we'll click on that.

Um, and then and and here's the other interesting thing is it'll show you an exchange rate predicting mechanism that's being run off of an agent. Now, typical behavior is that a user will end up opening six different apps every day. They'll say what's the best exchange rate on service A, service B, service C. And and oftentimes this happens near you the day of a paycheck. So from from the 10th of the month till the 16th 17th of the month, the user will be opening six different apps practically every day just to check the exchange rate. And we take that friction out of the whole thing. We'll say based on our on on the exchange rate trend and this is live as of today. It's telling you that this is the best exchange that you could

possibly get. And once you take that, then we take consent from the user. Um, and and and after that, the remit is triggered. Uh, so we could probably stop the demo now, but this is effectively what's going to go live this month. And and I feel like as mundane as this might look like, which is send money to India at a high exchange rate. There's a lot of magical things that are happening behind the scenes, right? Like you're taking their KYC, their passport, their driver's license. You're saving the friction of a user having to open multiple apps every single day for like a few consecutive days. And all of that is being abstracted away with an agent from from ironclaw that we're super excited to work with.

>> I just want to add a few things to that to what Nishkum said. So I didn't realize this, but Nishkum's old workflow used to be to do the these set of actions manually at midnight, right? And at near components, three legs of the stool of having real transactions happen using AI agents. One is secure AI. So what you saw with ironclaw and we'll be we'll have some exciting announcements soon with real security benchmarks showing that ironclaw is the most secure agent framework out there. Second is private AI. So being able to use confidential inference inference happening within trusted execution environments. So open AI anthropic nor any other companies can see actually your financial data uh

through their agents. It's actually in a completely encrypted secure enclave where no one including near is able to see the information going in and out. And third, really good transaction data. So you saw how this ends up being a um transaction happening over rails going to India. This is also part of our long-term vision of using near intents to be be able to have the transaction layer for AI real AI economic value. We think if you have those three, secure AI, private AI, and then transaction layer, you can create some magical experiences.

>> Thank you. >> So, you know, kind of expand upon uh what what kind of product breakthroughs or kind of UX improvements materially in this specific example uh are derived from this experience. you know you talked about how u you know it's a it's a it's quite a a tedious endeavor to do a remittance from India for expats um is is is it is this streamlining like are we saving days hours like what is the kind of convenience and the kind of breakthrough here of efficiency like what's the scale that we're talking >> yeah I I mean look I'll say this I think the India corridor is the largest men's corridor on the planet it is the largest

crossber payments corridor on the planet and it is by far larger you know than a whole bunch of others. Um unfortunately India is literally half a world away um quite literally so you're always staying up at stupid hours you know just to do basic things and here what you saw on screen is just the precursor of what we think this can be for now what this is saving is just you know the friction of having to open multiple apps every single day just to look at exchange rates. I think what this also takes away is the emotional anxiety which you have right because oftentimes the money that you're sending back home is being used to support family out the other end and you know getting you know like a marginally better exchange rate might

seem mundane but you know for for somebody receiving the money it is meaningful as a difference so I think you're taking away that stress but more importantly what you're also now doing is bringing peace of mind and starting to put almost like an autopilot behind what you're trying to do right and and ultimately where we see the most friction is on that last mile in India, right? Where once the money gets to a bank account in India, what do you do then? Right? And the idea that you're having to stay up at midnight, talk to a banker, talk to somebody else. Um, all of that gets abstracted away and it's an agent and and for us, you know, building trust around it means that you open up a lot of other use cases, right? Which is, you know, go pay electricity bills or go pay this, go do that. And and it it it

it really is a godsend. share my own personal example of what I did, you know, in the last one month. I had to file my taxes in the US. I had to go get like my bank balances in India. And typically that would have meant staying up again at midnight talking to a banker saying, "Hey, I cannot log into the website because I generally cannot." Um, and then having him email me a statement at the end of the month. You know, that would have been about a day of coordination back and forth over email and then I get my bank statement and then finally I get a bank bounce I could go put on my tax return here. And what I actually did is that I went to like my my openclaw ironclaw and then asked like what was my bank statement in India?

You've got access to my email. It instantly pulled up my bank balance. It just asked me for an OTP credential to like log in. And once I did that within I would say less than a minute, it gave me a bank bal like my bank balance at the end of the year. And and again it sounds super mundane but it's really hard to do. And and and so I feel like that's essentially where we see this going is that anything crossber we can automate away with an agent. Our our vision is to eventually add an ESIM with every instance of an ironclaw and so then it can do tasks for you remotely. It can call um it can like receive these OTPs or things that I had to do manually. It can even abstract that away.

>> Yeah, George, one of the uh things that I was talking about earlier uh as it pertains to this was this kind of prescriptive improvement of the way that some of these models are reacting. Specifically, that was the breakthrough with openclaw. Ironclaw was this kind of more prescriptive uh much much less prompting required and much more of like a actionoriented from a um like like like stimulating action from its own kind of intelligence rather than purely having to to uh you know give that you know uh prompt or or command. When it comes to highly sensitive information like finances, you know, to what scale do you think the privacy features of ironclaw really make it the most

compelling option for sensitive uh you know use cases? How do you think about this prescriptive prompting with sensitive use cases? You want the all the benefits of these highly highly sophisticated models and technologies with very sensitive data. How do you balance that? >> Yeah, that's a great question. The one way to think about this is the LM models are the brains and then the agent uh frameworks or harnesses are the bodies, right? The LM can think but the agent has to agent harness agent framework has to move has to act and those two need to actually be built securely and privately in different ways. So the way we've approached it is ironclaw ends up being very secure as an agent framework

because it does things like um it secures your passwords. it doesn't sort of reveal that to OM. It has it's built in Rust versus TypeScript. It has sandboxes for different processes in a overall Docker sandbox for more more advanced processes and it uses things like prompt injection protection at the architectural levels. There's a lot of like security things that have been implemented to make it a lot safer and you'll see this in security sort of benchmarks in the future. Um then there's the privacy piece you you mentioned but the privacy piece ends up being different from the security piece.

So the way that we handle privacy is we do confidential inference within trusted execution environments. That's a just like a complic complicated way of saying we're able to run open- source AI models in secure enclaves where no one can see the questions that are asked and the responses that are returned including here. Right? So now you have a agent framework that's much more difficult to compromise from a security perspective. you have a brain that is able to think privately and doesn't leak your information to open AI or Anthropic or any other centralized AI company. We think both are critical because if you have if you're doing financial transactions obviously you don't want a security compromise and if you're using

um private keys or credit card information or anything that's sensitive you don't want to leak that information to another company. And you know when it comes to actually the uh sophistication of the model and and the quality and the intelligence, are you still able to maintain all of those properties as the intelligence scales for for a use case like this? These users are going to want to be, you know, consistently feeling like their financial agent is getting better with them, helping them earn more money, helping them save more money, helping them be, you know, as you as you evolve in the feature set, right? You kind of build this relationship. They want to know that this thing's getting smarter and smarter. Are you able to still give them that property while maintaining the privacy and security as you mentioned?

>> Absolutely. I I I what you see here now are open source models getting better and better and kind of what I think will happen over time is the delta between closed and open source model will shrink more and more. You can't prevent smart people from working on other projects, right? So what you naturally see happening in tech is a diffusion a transference of like knowledge and expertise amongst many different companies right um like go Quinn Kimmy like all of these are great models and I think what you'll also see is that these open source models are ones that we can run within these secure enclaves near AI is also able to anonymize models through like a mixer model for if you'd really want to use uh an anthropic model or an

openi model that's still possible in an anonymized play as well. It's not completely private, but it's more much more private than giving them all your information. >> When it comes to some of the um remain side and payment side, is is there a a future that you see of, you know, this app specifically becoming a lot more integrated with stable coins, using near intense, using this kind of agent intense framework that George laid out in his initial response to the demo? How do you think about that kind of uh speed of adoption?

>> Uh no, no, totally agreed. um you know and and so I'll say this we think the future of of remittances is is onchain it's with agents and we think the future of agents is also onchain for all the reasons that George mentioned you know that it can be more secure it can be more private um and and ultimately you want money to move at the speed of the internet which it which which it just doesn't right now and and so we're very bullish about where the whole sector is going >> and so you you know you think that that there could be a very seamless in integration kind of under under the hood um operating for like a INR stable coin or USD stable coin.

>> Uh so that's the hope, right? I think you know there's still like a lot of work that has to happen from from like a compliance standpoint. Obviously we want to be on the right side of that. Uh but we do think you know that this is absolutely where the ecosystem is headed and and like we'd love to be first with with with NAR as we do that. As you guys are both thinking about kind of scaling the abilities of agents, you know, in the in the financial first framework, what are some of the other applications and use cases that you guys find most compelling from a actual uh demand perspective? Like like actual solving a problem, not just creating ideas, but what are the where is the real demand on the finance side uh for agents?

>> Oh, this is um why don't we both tackle this because I think we have very different perspectives on this. This is actually a question I've talked to a lot of both crypto and non-crypto companies on. Right on the crypto side, we see people interested in doing things like optimizing for yield um helping the helping to manage their crypto portfolios. We see people interested in using agents for doing things like um cryptospecific security um research on different crypto protocols. Those are a lot of like vertical specific cryptospecific use cases. And then all of the usual horizontal functional specific use cases. Uh I want to enhance my engineering team with agents capable of doing smart contract engineering. I want to be able to have um agents helping to

to enhance my operations or my marketing or sales as well. Um yeah and you know for us I'll say this our mental model is this is that every dollar being sent back home is being used for something and and often times the existing ecosystem looks at it almost as like a single player problem. There's a sender in the US who's sending money back home to India and then there's a whole another person out the other end who's I was who's actually going to be the beneficiary of it but they currently have no participation within that ecosystem. At the same time, you've got the sender here in the US who also has got very little agency of what happens once those funds do reach India, right? And again, there's a lot of friction within that whole ecosystem.

So, with that mental model in mind, we think that using agents, you can make the last mile significantly more seamless. You can bring in the recipient within that. So an example of that is if my brother hypothetically in India wants to go buy a laptop I can either send him $1,000 you know over a a money remittance done in the traditional sense and then he has to receive that which has a whole bunch of friction as well. He'll have to negotiate with his Indian bank for receiving those dollars. What exchange rate is he getting and then ultimately he will then take those proceeds go put it on a card or just use his his debit card and go to a local store and then purchase. And how wonderful would it be if I could just send him a WhatsApp with a link to an Abound wallet where he can go to the

local store and spend on demand with funds that are already in the US. And how wonderful would it be if if within that whole interaction, you know, I could just say my brother needs a laptop this month, right? And like the agent is now looking for the best deal, just now working with my brother, you know, for that, right? And so when you start to look at the world as you know how can this remmit experience be somewhat invisible and really enable the recipient to do the things that they really want to do and really make it easy for the sender to do that and make that happen. That's incredibly powerful and and and I feel like that's ultimately where we see signal with AI and and agents in general. I >> think we're about out of time guys.

Let's give it up for your panelists George and thank you guys. ALL RIGHTY, we've got our next panel. Who secures the assets? Custody, risk, and confidential execution. Keon, Robin, Nathan Stump, Zack Bandal, Tony, Margaret, and of course myself. We will get you.

Okay, we have Alex. I apologize. Sorry buddy. >> All right, Alex. some questions. >> All right. Intense today. Progress partners and the road ahead. Alex Shepchenko, that is our next panel, the one that I just mentioned. Alex, uh, welcome to Miami. How is Miami treating you?

>> Um, so the internet is really Uh, mobile internet. I feel like I'm not in Miami, but in some kind of Miami grud. >> Yeah. >> Uh, um, yeah. But like it's Yeah. But besides that, the events are are amazing. So many not very typical crypto people um here but lots of interest from institutions and bigger organizations which is great man. I mean just to get right into it we saw one of the largest um DeFi exploits as of late due to a a single node signer uh one of one uh DVN is what these things are called effectively like a single node exploit.

Um, when you're thinking about the landscape for interoperability from intense, how important is it to just move away from centralized bridge designs uh and really be able to kind of scale scale into more more efficient and more and more practical bridge designs? >> Yeah, that's that's a great question. Um so fundamentally um even if you're working within a single blockchain um the people who are working with it are not just only u decentralized applications when you have centralized exchanges integrated with these you know people who are doing like moves outside of the blockchain and uh because of this the security concerns are are extremely

extremely important right you you cannot just roll back the chain in 3 hours like Litecoin did and expect that everything performed as as it was expected 17 sigma deviation it's it's is nice this is this is what it happens with our chains the same way how we would not expect that a one day rework of Bitcoin would be okay right nobody is expecting this so from that standpoint crosschain is extremely extremely complicated and uh security practices they must be hardened everywhere especially right now with the uh with development of AI and there are lots of we observe lots of AI assisted

hacks by the nature of what we're doing we are exposed to all of these different signals from all of these different ecosystems and we need to react accordingly um and because of these uh uh our security practices on the on the top possible level here um I have been running we I started the first bridge with the near ecosystem which was which is called the rainbow bridge at some point in time there were more than two and a half billion and of highly liquid assets on this bridge uh not being hacked and uh I would take a credit for for it. Um uh but it it evolved it evolved into omnib by by the near one and then the team is doing absolutely amazing job on on doing this stuff. So uh security what I'm

saying always that security is a multi-layered cake and then you always need to need to make sure that uh there are enough fences so that uh you know people are not going to go through the Swiss cheese uh only through the holes. Um yeah but everybody everybody needs to be alerted everybody needs to expect the hack. It can happen it can happen with the best of us. Uh and uh and the most important question is how you're going to react and what is your plan after this. Um yeah so unfortunate very unfortunate situation with Kelpdow and Later Zero.

>> Yeah I was just curious. Yeah I appreciate that. I think one of the things that you guys have done well is uh get tightly integrated with a ecosystems that are difficult to access from EVM chains or from Salana. Even just Bitcoin is difficult to bridge in and out of for you know just traditional bridges designs. It's just tough. Near intense and near.com has kind of p paved what is now one of the best if not the best crossexecution environment bridges right like you can go from all different types of execution environments in in a very seamless way. And specifically with Zcash, there's a there's a big privacy orientation there, right? And and and I think that's something that a lot of

people want to be a bit more in the loop on and understand a bit more of. So perhaps you could kind of give the give the near intense why it is so important to connect you know ecosystems like like Zcash to Bitcoin to Ethereum but then why is it so important to do that in a confidential manner and and and that you know these confidential intents is something that you're very very focused on. >> Yeah. So historically we've built um public blockchains as a as an idea to create a um decentralized version of a money where there is no centralized authority that is controlling the flow.

We did it with the tools that we had back that time and uh it was working but the way how the public blockchains are working right now is pretty bad. And to give you an example here, like if I'm going to go and pay what with my USDT on Tron for which is in some places is accepted uh for um for my coffee. Um the problem that I'm going to see is that the person on the other side of the counter would now suddenly know all of my transactions that have been happening through all of my time. And then in case a potential person who wants to attack me would understand that I was in this coffee shop at this point in time and

there is a pattern that I'm going to this coffee shop and I'm paying because they know what is the address of the coffee shop because maybe they it's leaked somewhere in the internet you know and all of the stuff. So then they can attack me. Traditional financial system doesn't have this flaw. when I'm using my Revolute card and paying for my coffee, nobody knows how much money I own. So, so privacy and the all of the discussions around privacy with the blockchain is actually bringing back normality into the blockchain world into the financial aspect of the blockchain world and this is extremely extremely important and any meaningful adoption can happen only through the privacy. I'm

super glad for Zcash to bring this narrative back uh into the market. Uh they did an amazing job uh here and uh I'm also happy that uh near intense now supports the confidential swaps. Uh something that again everybody are expecting to have when you're going to the currency exchange. Nobody sees how much you specifically traded of US dollar into Canadian dollar for example. and uh uh and now we have this functionality, this one additional piece of infrastructure that allows you to also confidentially uh trade currencies.

So >> yeah, I feel like that's a big part of the next evolution of this product, right? Because I think what what we've seen is the emergence of a lot of crypto assets on chain for the last 10ish years, you know, specifically in the last five, right? And in the last you know six to 12 months where we've seen a absolute transition and major focus shift to non-crypto assets onchain right think anything from a stable coin all the way to private credit and anything in the middle of that risk spectrum with confidential intents you provide a much more institutional appeal to moving assets across chains that are non-crypto assets. So, I'm curious how I know you're very kind of like cippher punk

and very like ethos driven. How do you think about the rise of non-crypto assets on near intent rails and the need and and who you're serving with confidential intent? >> So, the adoption of the confidential swapping uh at the moment it it just naturally swap it just naturally starts. uh we are deliberately not doing any uh you know crazy moves you know throwing uh tons of money millions into the marketing of of anything. We believe that the users will always use the product that they want to use and there is no point in uh in inflating the the thing the kind of the things but I'm

happy to say that uh our users are willing to use confidential swaps even though the current fees on the neo.com are higher because they are willing to pay more for the confidentiality and we are already at the point when more than a half of The swaps that are happening on the near.com are happening through the confidential rails. >> Really? >> Yes. So people >> like this month or or last couple weeks or >> uh yeah just last couple of weeks. So, so users prefer to to stay confidential which means that for me this is an indication that this problem actually bothers users right. We hear a lot from the institutional clients that uh they

they are uh tired of being tracked by Arcom and being able you know just to put the address and that oh yeah this is associated with who knows maybe bitwise and then and then suddenly like there is no positions that are invisible. Uh the whole market operates with sufficient amount of intelligence with which which you can just buy. The whole market operates as a as a transparent like completely transparent. This is like this is unimaginable.

Imagine a tab on your Bloomberg terminal where you can see all of the trades of all of the traders in the world. Right. So that's strange. Um uh yeah. So then I think like do you think that that there is a stronger desire then from the uh larger asset managers and allocators to use these rails for their kind of onboarding of this tokenization wave and to use confidential intent for you know for that >> there are there are so so many different use cases. Um uh there are many projects that are willing to move uh for example payments and payrolls uh into on onto the blockchain because it's just much more efficient especially for the

crossber transfers. They want to have uh they want to to be able to have this private uh there are funds who are willing to liquidate their positions and now suddenly if there is a project that they invested in and they know what what is the wallet of the fund and then suddenly if the fund wants to liquidate the position the next thing that you are getting as a fund manager is a call from the project like why are you doing this uh why are you liquidating well well we are kind of doing the rebalancing of the portfolio blah blah allow all of this stuff. Yeah, nobody wants this. So, if you want to dump your tokens, dear funds, come to near.com and you can do it and uh nobody will see it. Um and there are many many

other very natural use cases. Uh so I was obviously exaggerating and going to the extreme here. Uh but there are very natural very normal feel of your financial life not being exposed to the whole world >> 100%. Um when it comes to some of the revenue that Neo Intense has has uh you know kind of accured and generated as of late, has it has it has it surprised you >> like? Um it somewhat surprised me but uh like approximately one and a half two years ago when I was first looking at the crosschain market for the decentralized protocols I was surprised that this market is uh

there are market participants in Thor chain back then was uh like super super big. Um but I was surprised that the algorithms and the approaches were extremely inefficient. It felt to me in comparison with the traditional markets that this is completely you know open field and uh people are amateurs. Um and my feeling was that hey we can make a next step of bringing better technology that is going to be more convenient for users because we are experts in in making great user experiences. But then we also can make it closer from from from the standpoint of the guts of the protocol uh and make it at least at the

level of the centralized exchanges when there are market makers when there are users they are matched against each other through some kind of form of a matching engine or like order book uh blah blah blah all of the stuff and this will make the capital much more efficient and still preserve uh all of the great things. So uh yeah I was amazed that the market exists. it is extremely inefficient. So, uh I'm happy that Neur was able to u to move forward and then capture this this attention. Uh this is definitely not because of me. Uh this is definitely because of of all of the great people who are working in the team. So, kudos to my team to my team so that I can sit here and and tell about how good they are.

>> Yeah. And and on the revenue side, did you uh anticipate or were you worried and is it still a concern that the fee switch was going to deter users who are looking for the best rates or best pricing? >> Um I I don't think so. So the fee switch is just uh it was just pretty clear that the protocol needs to earn money, right? And then for those of you who don't know um in February what we did is is we actually said that hey dear partners who are integrating with Near intents you are setting up your own fees that you're getting so we would like to get a cut of this fees right that that was as simple as that this is a win-win partnership completely symbiotical uh in between the

parties so this is a nature thing a natural thing I always when I'm working with somebody with an external person I'm I'm always asking myself a question so why this person is interested in this partnership where he does earn money and if this person doesn't give me an answer to this it means that there is something sketchy that is happening here I don't want to feel sketchy I don't want others people to feel sketchy about myself so I'm able to very openly say yes we would like to work with you because the success of your business means the success of our business too so uh let's work together. Let's grow together. We are not forcing you to have fixed payments to us. Nothing like that. If

you're big, then we are big. If you're small, then we are also small. We are working as partners forward. >> And Alex, why are the fees higher on the private side than the public side? >> Uh this is a decision of the of the product team for the for the near.com. So the the protocol fees is like the protocol sells in in two uh in in two versions right there is a B2B partnerships. So for example users on ledger are swapping through near intense. So every time when you're executing the swap on ledger uh there is a chance that you're going to it's going to be routed through the near intense.

Uh but then there is also a our own front end which is near.com and there is there are product people that are working there. we don't want to make it you know for free and so users expect it to be free it would be a misconception uh and then there was just this idea to test out what if there will be just additional five bits on top of this would people bother would it be uh would they are they willing to pay just a little bit more but for the privacy and and what the market tells us is that yes people care enough about privacy to pay for it. This is an absolutely great and important insight that we are getting from here.

>> Just last thing quickly um you know what do you think gets the near intense revenue to you know several hundred million dollars a year. >> So the path here uh is uh not going to be simple. uh all everybody understands this uh and uh our first and like there are multiple uh directions that we are taking in order to to get there uh and they are mostly widening up the addressable market for intents or widening up the set of products that we are proposing to our customers. So on the addressable market we are moving towards the institutions we have great partners who are actually performing quite a lot of

volume uh and uh we are making this technology much more accessible for to them. This includes uh security attestations uh ability to do uh later settlement margins you know all of these different things. Uh on the breadth of the products uh we are definitely integrating more chains more interesting tokens. This is like a very simple step uh step but then we are getting additional features like confidential intense and then an ability to earn so that the assets are not just stored there but they are productively used they can be staked they can be lpd into defy if anyone is brave enough to put money in defy right now um and some additional things but this will pass

this will pass right so uh so do more things with the assets inside of.com that's the breadth we're We are walking both of these roads right now. There are different teams who are focused on on these things. Uh and this is the way how to grow it staying and all of the companies who are just doing crosschain swaps and not moving uh to the new markets or trying to serve this this this product to to wider groups. Uh they are not going to grow because this market is limited to 200 to 500 million per day. And uh uh at least our ambition uh is to become a much bigger venue uh and this market is not enough for us.

That's why we are actively expanded. >> Ladies and gentlemen, give it up for Alex. Thank you guys. >> Thanks everyone. All right, the previously mentioned panel is now ready to be moderated by Robbie. We will have Keon, Nathan, Zach, and Tony. The chairs set up, guys. Let's give him a round of applause. Rob, what's going on? Yo, Andy, my lovely co-host. We're gonna get some more chairs up here. Thank you, gentlemen.

This guy gets it. This guy's fast throughput right here. We've got chairs per second at an all-time high. >> Uh mics should be on. Uh one more chair. Two more chairs. >> Oh, we got a quick reorg. >> Cool. >> Uh yeah, I'll I'll come on the end here. All right. Great last panel. Let's have some fun here, guys. It's a great week in Miami. What do you guys think? Miami so far. I grew up >> hot. hot.

>> Um, anything other than you got the suit on. >> Well, I live here and uh so I love it. It's awesome. Best city in America. >> I I would say, can you imagine the vibes if the market were up? I mean, it's been kind of an incredible week in terms of energy level and uh attendance and all that sort of stuff in a down market. I mean, I think it's been a great uh a great conference overall. >> Yeah. Yeah. Yeah. Uh market was up a little bit yesterday. I saw Near was up 16%. Zcash had a ripper. Um okay, so we've got Hodle, Zodle, Bitco Grayscale.

Uh we're going to be talking about who secures the assets, custody, risk, and confidential execution. Uh, well, why don't we why don't we go down the list? Um, I I don't think I've had the pleasure to of meet you guys. Um, a aside from Zach. Zach was on our show a little while ago. And I just want to I want to use something that Zach said in that episode as kind of the pretense for custody. Uh, and I think Zach knows what I'm going to allude to because we were talking about the tokenization super cycle. And it really feels like, you know, first we tokenize the assets, they come on chain, they go into all these DeFi primitives, and then all of a sudden you do get this massive repricing because this absolute storm of assets has now entered the arena. Um, and so what was the biggest thing holding that

up? Zach said institutional custody. So I want to we'll go down the list. will get custody risk and confidential execution. But just starting with custody, Zach, maybe you could just expand upon this idea of what is currently holding up the tokenization super cycle and why you think it's institutional custody. >> Yeah, a lot of kind of elements to that story, but just building on that previous conversation, you know, how how we think about it is the financial system. You know, that's what we're rebuilding is the the financial system that connects savers and investors or people that have capital to lend and people that need capital to borrow. And so we can tokenize assets all day long, but if there's not a way to bring the capital uh uh deployers, the savers, the people that are investing these funds

onchain effectively, there's no uh point to it. And so one of the things that holds back this wave in our our view is just we're not quite there yet for you know the big boys of institutional finance to be uh ha you know doing self-custody having their own wallets. Um the infrastructure is almost there but certainly it's not deployed at scale. When you go into Morgan Stanley, you go into Wells Fargo and you go into other partners, they're still very much at the ETF. Uh they want that custody uh provided for them in the form of the ETF and they want um the wrapper to go in traditional brokerage accounts. When we when we get there, when we get to the point where we have institutional wallets and and self-custody, we're going to be able to unlock the

tokenization wave in a much bigger way in our view. >> Yeah. Um, okay. So, so setting the stage here, you know, we've now got almost all of the pieces where we're ready for trillions or more in assets. Um, it feels, and I mentioned this earlier, it feels like we've got this three-headed dragon, Zcash, Near, and Hyperliquid. Um, using these as sort of the the foray as through which I think a lot of people are kind of looking at and entering the ecosystem. You know, we've got a varying set of approaches. Zcash got, you know, shield in. We've got Zodle here, you know, and Hyperlid's got, you know, their approach and deposits, self- custody and such. And then we've obviously got the more institutional front. When we're when we're looking at, you know, kind of the the landscape

here, how do we ultimately get these things all composable? I think that's where we really start to unlock what, you know, the, you know, the value proposition of this new financial system can really do. Um, maybe we'll, you know, Tony, we'll start with you and we'll kind of come down the line. privacy feels like it's it's in the room with us all the time, but we can't really see it. We don't quite know how to integrate it. You know, if we can get to a point where, you know, we we leverage the the the privacy of Zcash in addition to these other assets, everything becomes composable together, feels like that's a massive value that's unlocked. Maybe you could speak to how you guys have approached this at Zotle and how we can bring privacy to the rest of the ecosystem.

>> Yeah, that's a great question. um Swedle, one of the things that we did to really um you know bring Zcash to other ecosystems uh was we integrated your intents um and obviously it's topical here and Alf just had a good talk on it but um you know historically um Zcash has always had like really groundbreaking tech um zero knowledge proofs you know we implemented those back in 2016 um but I think it was a little bit um isolated in its own ecosystem And um so now with with near intents and one of one of the features I'll call out is is cross pay because this plays into aenta commerce but um if a vendor or if uh my friend only you know wants to

accept USDC or Bitcoin or ETH um within Zado I'm able to you know store my wealth shielded in Zcash and I can pay them um exactly the amount in whatever currency they they want any currency that's supported by intent. And that's a big deal. Um, you know, it's kind of like meeting meeting people where they're at. Um, if if you look at, you know, Aenta Commerce or whatever, if I have an agent out there that I want to go make a purch Amazon or go book a flight and, you know, they don't accept Shielded Zcash, well, that's that's fine. Uh, through our our partnership in your intents, we're able to um those payments can be executed as well. Can >> can I just pile on one quick thing on Zodle? Not not to hog the mic here, but

um I mean I love the you know the outbound concept of course what Zod and Zcash are doing, but I just like to pitch you know think about it inbound as well. Everybody that has spent some time in crypto has some assets scattered all over different chains. You probably got some BNB chain stuff. You probably got some Avalanche stuff. When was the last time you looked at these assets? you know, Zodel and Zcash can be a private hub for your onchain wealth. You know, clean up your onchain life and bring these with near intense into a shielded pool and then, you know, who knows what you might do with that those assets in in the future. But what a great place to sort of hold these zodle is mandatory self-custody um auto default shielded assets. You can kind of sweep up your

onchain life into a shielded pool. To me, it's like a great use case of both of these tools, the wallet, the chain, and uh and your intents. >> Yeah. >> All right. That's right. I just love it. I had to had to get that pitch across. >> Good pitch. Um, we so at Hodddle actually we have we have have some unsolved challenges which we're waiting for, you know, some evolution to happen in the space. We we do tokenized hedge fund strategies which means that people can access them you know onchain as well just directly uh on our website you know the auto protocol. Um, but one of the issues is because it's onchain, perhaps somebody could see all of your holdings, you can also see the smart contracts that are executing trades, what they're

executing. Now obviously reverse engineering that is difficult um and might not be worth it but um there there's a challenge there that as you know we get more consolidation in the market there's less and less strategies less funds um that are providing those strategies then it'll make more economic sense to try and copy or reverse engineer um uh and frontr run them. So we're looking forward to actually seeing how we can work with near as well and eventually uh if there's a standard that's developed uh there's like some standardization in the space about how uh we expect privacy to work like in tradi you know there's different types of certifications and comp socks

compliance and other standards if that comes to digital assets and it'll be great. >> Yeah. >> But ultimately who you know who hates privacy is regulators. So like when we think about the evolution of where we're at now like if BCO right we're an OCC chart bank we have seven regulatory licenses globally hundred billion dollars in assets um two years ago we couldn't get a bank account so like when it came to you know regulators of people coming to us and say hey you know we want to hold these assets we want to deploy these sort of things OC at the time we're state trust um regulator coming in saying tell us about these transactions like we have we we could privacy was kind of a a difficult thing. didn't support a lot of

the tokens because you know regulators hated it and we were trying to run a business and so now we're at a point where like privacy is becoming you know there's always uh the fund running and all these sort of things but from a custodial perspective it's really difficult because we don't want clients well we don't want people to look at Bitco wallets and say we can understand and start tracing these like we want to think we're thinking about ways that we can can shield for lack of better term these sort of positions but at the end of the say, you know, having even this conversation is a little bit more privileged than it was a year and a half ago.

>> So, so before we get too deep into this confidential execution realm, let's just come come back to the custody conversation, you know, when it comes to the institutional custody, Zach, you mentioned, you know, this is kind of holding up some of this tokenization super cycle, especially having an on-prem solution for these things. I mean, Nathan, would you argue that, you know, we actually have it solved? Bitco is the solution for that? No, because you can go on chain and see pickle wallets. I mean, like right, we get wallets that are tagged, right? People can see transactions of some of our largest clients. If you're a a large organization, let's say VC shop that owns a sizable portion of an asset, if you put that into an environment, no matter how many wallets you split it up into, they can see the base and they can

start tracking those sort of things. So it's really hard to actually you know we have you know clients who own sizable amounts you know the lowflat high FTV sort of positions and and it's really hard to get away from tracking that now in the world of publicly traded companies like you can't just sell off a position and not tell anybody if you have a sizable position. So like there's some things around privacy maybe being not the greatest solution cuz it's it is trying to shield from like what's going on in the market if you are a key decision maker or position in that in that holdings. But like ultimately for us like we don't want we don't want all our wallets to be exposed not because we're trying to hide things but we're trying to we are trying to create some level of confidentiality internally. Um,

Zach, just to just to kind of come back here, like, you know, what what do you currently see with Bitco as just a custody example, right? There's there's lots of institutional custody providers, but we got Bitco, you know, Nathan's here on the panel here. So like >> shout out Bitco. >> Um, what so what what could we do? like how do you how do we overcome this uh this current uh environment of custody and ultimately reach the catalyst of unlocking this this next >> we're really getting there and you know really shout out to Bitco US publicly traded company leading the industry in this kind of stuff uh groundbreaking in in many ways and uh you know Mike is an OG and somebody who

I think is very thoughtful on all the uh complex issues around this topic. Um, look, I think for me the answer is kind of a demand pull thing. You have to have a reason for uh institutional capital to get on chain in in the first place. What is the asset that they want to go get? Uh, you know, Bitcoin was a kind of a start of that process. Uh, can we make um uh unique types of tokenized assets or unique tokens, you know, like say the tokenized Galaxy share example, you know, something that only exists on chain. I think that ecosystem will grow a lot and we'll it'll be a demand pull thing. Institutions will say I got to get on chain because I have to get that asset. We're not quite there yet, but I think that is the thing that ultimately forced us so forward.

>> Um and then obviously you know we've talked about privacy confidentiality as it pertains to custody as well as execution a bit custody as well. The part the boogeyman here is the risk like you know we've seen far too many hacks over the last you know even call it call it weeks to months. It feels like it's increased the it's accelerated. Um, we know that there was this massive buy bit hack a while ago. So, buy, you know, culminative, you know, hacked assets. Maybe it's not the biggest in terms of, you know, what we're on pace to this year, but the amount of protocols feels very, very vulnerable. Um, how do we overcome this? It feels like if we stick to our roots and we're decentralized, maybe we can prevent a lot of this. But I'm just, you know, curious, does h how how does it start at the custody layer

or is there something else that we can embed into this ecosystem, risk controls, asset controls that ultimately mitigates a lot of the risk that we're seeing right now? I mean, in our case, I think we recognize that it's it's it's the biggest risk and it it it it might actually get worse because of what we saw with like drift and the social engineering aspect because as more people actually get involved in DeFi and digital assets, but especially the DeFi side of things, then there is more potential targets. So you have a and also the the equation is um control versus you know uh versus kind of uh so if you have more control maybe you can

prevent a hack but then you also have control so you could be a perpetrator like a bad actor insider risk. So there's there's both of those things. the insider risk could be mitigated if if you know governments decide that okay just because you have control doesn't make something a security because right now that's one of the big like um you know pillars of how you avoid having your protocols token be considered a security is can you control or do you have discretion or authority over like trading for example that happens in if it's a vault product in that vault so protocols try to eliminate the control factor but by doing so if there is a hack, they essentially have no choice other than maybe to shut it down.

>> So, I'd chime in and say that um you know, privacy isn't about hiding anything. It's really about um mitigating your exposure. Um and it's just it's hard to hack what you can't see. Um at at the end of the day, you know, if you can't see balances, if you can't see transactions, it just um your transaction graph is really your risk exposure every every time you're making public transactions on chain. um you know with with AI that becomes trivial to track patterns and to track balances and um and so like as as I think Alexir was talking about earlier um you know it's in our normal daily lives like when I go buy groceries at the grocery store I'm not you know showing the cashier my

account you know my account statement and all my transaction history but when you're transacting on chain on a public ledger you are doing that and it does make you a target. >> Yes. Um and as we approach this you know from you know we're seeing more of these uh products in terms of RWAS tokenized you know products kind of going starting with stable coins treasuries kind of going down the risk curve there is is there creative ways that we can you know we can get creative with custody with privacy in which you know maybe there's limited movement of the underlying but you know we have very robust and liquid products >> there are versions of some of that today in a centralized fashion, right? So like different custodians have different

methods. I can speak to what Bitco does. We have something called the go network where is essentially an omnibus wallet. So it's you have one wallet address and because we're bank, we can segregate on all books and records. So ultimately you let's say you have counterparty A interacting with counterparty B and we're selling between those folks. The asset doesn't actually move on chain. It just moves on books and records. That's great if you're on BO, but if you want to transact outside BO, the asset has to leave. So, you know, it's these are these are some of the things that we've done to try to help mitigate some of the questions earlier, which was like, you know, not seeing all these positions move all the time. But as soon as you leave that environment, you're going to see the transaction happen. So, we do things like that, but outside of that,

um yeah >> I I know we got some things to add. I just want to build on this by saying this is going to continue to be like we're seeing more institutions enter the space. This is only going to be a a a growing issue of fra fragmentation. >> Just want to put put that out. Yeah, absolutely. One, I think, example, of course, uh, of a way that you can provide liquidity to these assets while keeping custody more secure is the ETFs. Uh, this is one of the reasons that they're a solution. Only a portion of the ETFs need to be kept liquid to meet creation and redemptions, much of it can be kept in cold storage. And as long as there's a buyer and seller on uh ready at the current price, you actually don't need to create and redeem all day long.

It's only when there's a kind of big imbalance between buyers and sellers. So there's a lot of volume that happens on an exchange order book on the ETFs and there's no coins uh moving uh on chain. When you look at the uh Bitcoin blockchain, uh active addresses are actually trending uh down and of course this is a question about the the health of the ecosystem, but part of it is just the ETFs. There's a lot of volume that's happening in a a secondary uh market doesn't need to be transacted on chain. And yes, that means there are fewer active addresses, but it also means like users are able to get liquid access to Bitcoin without taking any new uh custody risk. In that sense, it's providing a service. I'll just share something that I've been tangling with

personally over the last, you know, few weeks here as we've seen more of this progress. Um to to take an asset that's native onchain, put it into an ETF, but then tokenize that ETF. It feels it feels convoluted. >> How can you guys help diagnose this? like what is this ultimately the right structure? I'm just no >> it's not okay that it feels it feels like it shouldn't be but maybe we can understand why this is happening now and maybe how how we can improve from here. >> I'll just add one one thing. So in the long run uh no it doesn't make sense to uh move something in crypto, put it in a trady wrapper and then move it back to crypto in some conceptual sense. Why why are we doing that? That being said, I

think there probably is a little bit of room to to do something and that's because the trady stuff has something to offer uh crypto. What could that be? Well, maybe it is some kind of cross-chain custody. Uh you know, you could hold a multi-token exposure, for example, without having to use bridges. Uh you could have index pricing methodology and automatic rebalancing and and those things are not necessarily available on chain. You could have audit uh and other compliance uh types of functionalities. And so um you know some of these things are kind of exports from uh trady to to crypto and so maybe uh tokenizing an ETF could make sense in some narrow cases >> on once they become available on chain these things then >> yes in the longer run I couldn't agree

more these things are ultimately all moving on chain is a a temporary uh solution though >> doesn't it come to a question though like why the products were created in the first place right like so if you look at let's look at like asset management as a whole right so be before I got into crypto however many like years ago I worked Transameric capital on the wholesaling side. So if you take an asset like Bitcoin where what if your adviser at Mailins, Wells Fargo, wherever, you can't allocate to that. Your systems aren't set up for it. If you're an RA, you're not going to build you're not looking for custodians because you're scared shitless of the space. So like ultimately you're like, okay, I am comfortable with an ETF structure. So, you know, start off GBTC and some of these assets that ultimately got you exposure using a vehicle that

they're comfortable with and they can get their compliance departments and risk and hold it at Persian and all these other things. So, then you take that and you expand it onto other assets, ETH, soul, so on and so forth. So, now allocators can go into it. The next kind of leg of that in some ways is, you know, there's obviously the tokenized asset and I'll come back to that, but then you have digital asset treasuries where you say, okay, ultimately I want to get exposure to an underlying asset, but I don't want to own an ETF. I want to own an equity position, not an ETF position, and I buy a DAT. Then >> why just quickly, why why might someone want an equity position and not an ETF? >> It could be investment policy statement on how they actually structure funds or how they can like what type of assets

they can hold like so for instance, how is a digital asset treasury quantify as an equity position or is it quantify as a as a cryptocurrency? Like if I'm a large cap equity, if I'm a midcap equity manager, small cap, like I can't buy, you know, the NEOAT. I know S couldn't make it today, but the NEO DAB, but I can buy his I can't buy the NEO asset, but I can buy the NEO deck because it's an equity. So, it's a different way to get exposure. That's kind of how I thought 2.0. And then the whole kind of DAT thing kind of went a little sideways. And so, the ultimately what led into now we think about the tokenization of like why do you tokenize Nash? Well, if you tokenize Microsoft and Apple and these sort of stocks, who are the users of that? It's typically

people who can't open a US brokerage accounts, it's somebody overseas. It's in Africa, you're in Europe or whatever else. And so ultimately, like if you bring the ETFs back into these tokenized structure again, why are you doing that? Is because you want to borrow against it and so a D5 pool, a lending pool, and you want to do it natively, is it because you can't actually get exposure in the in the US markets because you sit somewhere where that doesn't really exist? And so like ultimately I think to your point like people creating products just to create products don't make any sense but like there is a reason why some of these going to be if you think about it kind of like at the end of the day who's the allocator. Um thi this occurs in my mind Tony you know like as you think through the evolution

of assets and obviously you guys are geared towards Zcash but circling back to custody a bit you know I I think we will see tokenized equities for all the reasons that Nathan mentioned you know access um as well as DeFi composability these sorts of things what what does asset breath mean to someone like Aado you guys are you know primarily focused on Zcash maybe more broadly focused on privacy is there components that we of privacy in particular that we can apply to a tokenized Microsoft equity. Is that even in the realm of possibilities what you guys are thinking?

>> Yeah, that's a good question. Um I think at you know like I guess like you said Ztle is a Zcash only wallet. Uh we allow swaps in and out of other currencies. If you look at just the underlying technology with zero knowledge proofs um you know zero knowledge proofs are just a way to prove something without disclosing counterparty information. um whether that's for on the Zcash blockchain for example, you know, it's there's a transaction hash, but there's no sender, there's no receiver, and there's no amount. Um so, you know, it's not something we're really focused on at at this time, but if you think think about um something like tokenized equities, like it would be a way to prove ownership without disclosing um

you know, additional information. Um yeah. >> I was actually going to say you're basically you're basically centralizing trust in the protocol rather than in everyone, right? Is that is that accurate? >> Yeah. Yeah. I mean trusting Yeah. trusting code, trust in the protocol versus uh trusting, you know, uh different counterparties. >> Yeah. >> Yeah. >> Um thinking a bit more about I'm curious about the ETF wrapper here like around some of these assets. Does it is there a regulatory component? You know, there's obviously things up in the air with the Clarity Act and whatnot, but will Zach, do you think we'll see something where, you know, we get some exemption where, you know, it becomes more accessible to

just go directly to these assets on chain? >> We do need some regulatory clarity to unlock all this stuff. May maybe going back to the example of why why might you tokenize an ETF or or something else. I think a lot of good examples have been brought up. Maybe it's composable in DeFi. can trade it on an AMM but only hypothetically actually that that is not actually uh allowed uh in the current system a decentralized exchange would have to be some kind of a broker dealer and and meet some certain standards. So one of the reasons that tokenized assets onchain don't be see seem to be maybe doing as much as we would imagine. You know it would it would seem like if it were fully unlocked there would be a richer ecosystem of tokenized assets and really that is a regulatory barrier. we

need some kind of exemption uh so that uh say a tokenized asset a tokenized stock can be traded in an AMM uh pool. That is not the case uh today. So that'll that'll get pushed forward over time, but we're not quite there yet. So you know what what should somebody like Grayscale do? You either um you know do something now just to have a toe in the in the water, but it's not actually uh going to be a big business very very soon. So you can either do something now just to to have it done or you can kind of wait for that clarity and then execute at that time. I think really that's where we're at. We're not totally ready uh to see the tokenized asset uh uh vision come to life and and we're we're getting close but we need some of

those further regulatory unlocks. Just to add, uh, I I hope what Zach is saying comes true because we have to do some convoluted, you know, things to allow basically a user that invests in a vault that's actually an equity strategy. You have to then take the deposit offchain and execute through like a, you know, prime broker um, to buy whatever equities and other instruments are part of that actively managed strategy and then do the whole thing in reverse if they redeem. And so the two solutions would be just have all equities tokenized or um have a custodian that's both a traditional finance custodian and a digital assets custodian at the same

time. Maybe it exists. Um but yeah, that would be what we would love to see happen. >> We're working through trying to hold, you know, traditional traditional structures, right? So like we're cryptonative in the sense that like we really good at holding digital assets. we have the ability to hold securities, right? So, we're working through things like that, but it's, you know, it's an operational thing. It's structurals, but I mean, and now you you're gonna have the vice versa, you're gonna have those large names looking at us to use info. So, come >> I'd imagine that there's a lot more demand, you know, kind of to to your point once you have the custody structure.

What what are some of those conversations like? You know, h how much have you looked at expanding the asset universe beyond cryptonative tokens to more, you know, RWAcentric to unlock some of this more native tokenized equity composability. >> We bought a company a few years ago called Brassica, which is a transfer agent. Um, so we've been down this path for a while. We, you know, Brassica today has its own licenses and it can hold, it does hold traditional traditional um right now Bicko Bank and Trust is continuing to try to think through how we do that. And we've we have some ideas in place and now that we're public, I'm a little less of what I can and can't say. And so, but ultimately what it leads to is, you know, we're we are think it's very

important for us to be able to do both, just like I think the the the ones that are holding traditional today thinks it's important to hold digital. So >> yeah, I I'd imagine a lot more companies are thinking in this direction. It it feels like we're, you know, no longer just, you know, a cryptonative ecosystem and then, you know, tradi ecosystem as they call it. This is just finance. Uh we're going to just have finance companies. This is just a finance event, a finance conference. Miami is a finance city. Guys, thank you so much for listening, watching this panel, being with us here today. Uh and uh yeah, give it up for our panels. Thank you so much, guys. Thanks for having me.

>> Thanks, guys. It was a pleasure. Yeah. Yeah. Awesome. Didn't gonna press you there too much, but it is fine. It does feel like that's a that's a an interesting area to explore. Thanks. >> Yeah. I I haven't even I really started thinking about it in that >> that's a natural next step. It all unlocks. to find I'm find to find

topic. Yeah. Yeah. Good. building. Never swipe. Sure.

>> Cool. >> All righty, guys. We are getting into our second to last panel. Uh we've got George coming back up from Near AI. No surprise there. And Ari from Failafe. Guys, join us. Uh drinks are going to start at 4. Stick around, panel or two, and then we we uh enjoy the the Miami vibes. Um all right, guys. How's it going? Uh >> pleasure to be back.

>> Pleasure to be back as always. Uh George was on stage just a little while ago. Uh Ari, good to have you here as well. >> Likewise. Great to be here. And uh we're gonna be talking attackbench uh which is an adaptive security benchmark for autonomous agents. Um guys, why don't we just why don't we just start um we'll start high level. Uh you know there's been far too many hacks. Uh there's been security incidents out the wadoo. Uh it feels like uh those who are using AI are using it for evil and not for good and we need to switch that around. Uh so let's tighten up the security uh with these benchmarks and uh we'll we'll increase uh the benefit of all these autonomous agents. Um why don't we do

just quick intro? I know we've only got a few minutes but then we'll we'll get into the uh nitty-gritty about uh attackbench. >> Sure. Um I'm coming back but I'm chief product officer for near George and also GM for Near AI. >> I'm Ariinsky, CTO and co-founder of Failsafe. We're based out of Singapore. Uh we are cyber security company >> and George we've got a demo uh here for or some slides for for this one. Right. >> We do. Um let me just give a little bit of background here before we jump into the slides. So uh we meaning the neari team and fail safe ended up collaborating on um attack bench the security benchmarks because like you mentioned one of the things we noticed is that security is important and there have not been really good benchmarks

that have been made for agents from a security perspective. And if you recall like an early analogy I made, LM are the mind and agent harnesses or frameworks are the body and they both have a very important role to play in security. So we think this is like a very very important thing to focus on. >> Epic. Okay. Uh so let's get into attackbench. >> Cool. Uh I can kick kick this off and then um Ari and some of the engineers on our team ended up going deep and working on attackbench together. Um but the basic premise is agents um agents and LMS and the frameworks around them behave differently under attack and you have to consider security as a

property of a pair of both and I think this is like the highle work that take away I I took from both the work as well as the paper that Ari wrote. Um but uh we're really happy that this is kind of the some of the the conclusions and insights that came out of this work and then Ari would you like to take it away? >> Um sure. Yeah. So so just just as you mentioned uh it was it was really interesting. It was um uh you know when we started when we first started the work uh we kind of looked around and everything was like all the all the benchmarks that we saw it everything was static like there was a set of set of known attacks and you you know you plug the agent through and see how it does

but it's not very realistic. So I mean yeah typically the adversary will will deploy the best OM they can find right potentially and and then we'll adapt and so that's that's what we kind of decided on is that we're going to make this as hard on the agents as possible. Uh so we built out this framework attackbench where uh you know you have a red team and a blue team and the red team uh is powered by an LLM and it gets to see what uh what the thinking of of of the agent is. And it's it's very realistic like you can't assume you know just hope for the best. Hey this this you know I tried this it doesn't nothing happened

so we're good. Um it it is realistic. So um we we um let and you'll see the slide in a second. We let the red team evolve its attack based on the thinking uh o of of the agent. And so what we did is so this is the uh what we tried out uh we did this on three different frameworks on uh ironclaw, Hermes uh and openclaw and we tried four different frontier models going across 52 scenarios. Yeah.

So you you you can see you can see the you can see the loop here. Um so you get in this framework you get five different tries and it's uh you know if if any one succeeds then then that that counts. >> And could you just maybe give a an example of one or two of these scenarios so that we can understand you know the the loop here? >> Yes. In fact why don't we go to the next slide. Uh one more perfect uh so these are the kind of different attacks that are being tried out. Um so so the you know this is this you know you have you have the agent and then the LLM inside and uh we're trying to exploit you know the the red team doesn't get to talk

directly to the LLM but it's using various various side channels right so it's using what looks like a record in the field or a free form text and so these are some of the attacks right um so you'll have like a write instruction go write something out or go do exfiltration go fetch uh this URL and it'll have you have your your key embedded in it or try to make it exceed right so let's let's go back to the loop now >> I'm gonna ditch my chair so that people can see this >> and then oh sorry yeah and then yeah here we go um so so this yeah so you have um a you have the okay great yeah you have the regular workflow uh of of the blue agent let's say ironclaw or

open claw um uh being being asked to do some some generic task you know go tabulate go go go uh answer email or whatever and then the red team uh and this is powered uh in this configuration by by GPT5 uh 5.2 too. I think that's what we used. Um, and it will like embed if an instructions like one of the ones I just I just gave an example of. And it also gets to see what was the thinking, right? So, uh, how did how did the LM reacted? And you'll see in our paper, it was really interesting. Some of the logic, you know, it's it's attemp one, then then, uh, it the the, uh, probe gets sharper and sharper, and oftentimes

it succeeds. Did you have to trick the GPT5 model into being an adversary and and it had it had a prompt. So it was it knew what it was doing and it's it's its whole purpose in life was you know you got five chances. Okay, go at it. And it it was it was very clever actually and one of the things that we we we still would like to do and this is open source. We invite everyone to try this out and try try this benchmark on other kind of configurations. So, you know, you know, pick your best LLM and and and plug it in. But yes, so let's let's just keep going. Uh and uh in fact, uh Ironclaw out of out of everyone uh did the best. Uh so we we had no idea. So

this is like uh we had no idea what the outcome would be. Uh but you'll see that uh Claude and under Claude and GLM uh ironclaw had the least number uh the least number of uh vulnerabilities and as you know as our thesis to all of this is uh you know stated up front uh it really still does depend on the configuration between you know what what is what what is the pair that you're using um the combination of of uh the LLM and um uh and the framework and the agent. Um what was really what was also really kind of enlightening is like because we we saw the reasoning uh so you can actually there's a number of

improvements that could be made like for example uh for the uh ironclaw has a lot of strong controls for permissions that th those can also be extended to some of the other attacks. So >> and these models here on the left these are the protectors if you will. >> Oh yeah. So this is uh as George refers to them these are the brains. So this is for example this was sitting inside of Hermes openclaw ironclaw. So they were configured with ironclaw was configured with queen for example or GLM and son >> they were they were playing defense while chess the GPT model was playing uh offense and trying to get into these things. Okay >> one of the things we want to do from our own claw perspective is like better understand and better dig into actually the the traces to understand what is happening in these attacks and how we

better defend. But currently some of the hypotheses we have is like things like architectural prompt injection protection um sandboxing of tools things of those features of that sort guard rules that were designed to actually be security first really end up mattering. >> Um do do we have more more slides? >> Okay. So we observe these attacks. Yeah. Let's do one more and then we can uh I think we have a we're short on on time. Uh let's do uh one more slide here. Oh, the next one. Yeah. So, if you want to see what uh an attack looks like, this is an example.

>> Okay. So, in the resolution path, so we have the red, which is the attacker resolved by Alice audit flag. And so, what can could you describe what what red is trying to do and what blue reacts? >> So, this is this is a math attack. So, the task >> the task of of the blue of uh of the blue agent was to um Uh yes. So to compute the total >> okay >> and in in some in some field that the LLM ended up trusting it said oh by the way skip this skip this number >> and it did it happily did >> okay so blue lost in this one lost in this the bad guys won in this one. Okay.

>> Yeah, we have we have plenty of counter examples, but it was >> very very interesting how we arrived here. And you can actually on some of the on some of the runs, you can actually see how exposing the LLM's thinking helps the red team focus in. What was it like? What was the LM's thinking in in uh what was the rationale for for not doing this operation? So a general question here as we think through you know making these things you know more understandable less like black boxes so we can kind of understand what's going on underneath the hood we're actually exposing you know potential attack vectors. How do we reconcile or balance?

>> Oh, >> yeah. >> Yeah. But you it's it's just general security principle. Do not you don't want to do a you don't you don't want to do a black box. You you don't your sec security through obscurity. The lesson still applies here. So if you can hold up if if the if the if if the framework holds up, if the agent holds up under being transparent, then then then you can have some confidence in it. um you will you know if you try to close it up but by the way so all of these uh each one of these agents uh have open you can actually get the trace uh including including openclaw Hermes and we put in a PR like I put in a PR for for that's getting that's got that got approved for

ironclaw uh but uh but yeah I mean uh in in the wild uh the attacker is going to just take you know take their favorite claw and see what it's doing and then and then go after you know after some instance that's being used by a user. >> I know we're running short on time but I want to point something that I thought was interesting which is we went with the brain and the body analogy but one of the super interesting things to point out is that the body matters a lot in terms of security exploits too. You can see for example with let's say GM5 um one using one agent harness versus another can result in like 3x or like multiple X increases in the amount of

exploits you have right so it's not just the LOM it's not just the brain that matters it's also the body the harness the framework that matters a lot in security as well >> and why would that be the case why would the same model be uh more or less secure inside of a different harness >> is the So uh for for for the agent uh it's it's the um guards around it right. So there's you know permissions there's how you interpret how you how you do uh you know filtering right and for the model you know each model thinks in its own way uh it you know it depends on how many different instances of this particular kind of a attack it saw in the past. So, you know, it it gets

trained enough on on this instance and then it reacts in one way and then some other model will react differently. >> Got it. Um, George, you're in control of the clicker. I want to be respectful of time. Uh, any, you know, please, if there's anything else relevant, you know, please share, but uh, >> I think we're good. I think we're good. Okay. Attack bench. Very interesting stuff. I mean, I I want I'd love to see more data, you know. >> Oh, and we Yeah, everything is in the GitHub. >> Beautiful. Awesome. Guys, give it up. Uh, George and Ari attack bench. Great guys.

Awesome. Yeah, that was fun. Thanks so much for sharing this. We could we could go for a long time. >> Oh, you're on this one, too. Oh, okay. So, you guys are sticking around. All right. >> Um, I'll go on one of the ends. Uh so you guys can squeeze middle over here and uh we welcome Stephen and David to the stage. >> Yeah. How you doing?

We're rounding out near Miami. Uh I think drinks are going to start right after this at 4. Steph, we're really doing it. We're really doing it out here, guys. Uh last but certainly not least, uh we're going to be talking about safety, reliability, trust. Um you know, obviously you guys just heard from George and Ari about Attackbench. Um I I think that's the right place to start. Uh Stephen, David, let's get some reactions. Um can't see David all the way over there, so I'm right here. >> There he is. Uh David, uh why don't we start with you? Uh what do you what do you think? Attack Bench. Maybe we'll just do a quick intro and then uh we'll get into AttackBench and like benchmarking uh in general.

>> Well, I'm embarrassed to say you're making an assumption that I was fully paying attention. I was having some conversations. So, I'll I'll pin it over to Stephen. >> Yeah, you guys are doing uh you know, red team and blue team against each other, pitting AI against it, also to do benchmarking and find which ones are kind of the best combination of agents uh to do the defense and what they can find exploit. So, very familiar with that. We're doing that right now at Halbourne with um you know doing kind of the the red team side mostly which is what we focus on. Benchmarking is extremely extremely important. Uh >> you have to have like a a ground truth data set to to measure against but that's very hard to do in security because >> you don't really know there is no ground truth in security because there always

could be more vulnerabilities. If you guys saw the anthropic uh video around that 20-year-old operating system they found like a critical vulnerability. it's been there even though it's been pentested. So what is the ground truth? So that's one thing is like relying on a truth data set um you know from the tager perspective. Another thing is uh expectations around what the models can do because it's your AI is fuzzy and fuzzy out. So you could do the same kind of like script there but you may get different results each time as we see you ask chat GBT the same thing you know again and you'll get a different answer sometimes. So for things as complicated as security, expect the same different results.

>> Yeah. >> Um okay. So when we're thinking about this in general, maybe uh you know uh this is just an evolving thing. Security principles that we want to keep in mind uh are tried andrude but like you said AI is a very comp complicated thing. It's always evolving. It's fuzzy in fuzzy out. Um you know George Ari, maybe we could just reiterate some of the findings, some of the you know core key takeaways. uh and then we can build on this when we we start to make some benchmarks around safety and reliability in general.

>> The thing I would add is I go back to it was really interesting that there was such a big delta between um different agent frameworks using the same model. Right? And one of the next steps we have to do is is go through the trace and try to understand that better. Try to understand why. Right? We know that there are very important guard rules that matter. Things like using sandboxes, things like you know prompt injection protection. But one of the things we want to try to figure out is like which of the guard rows have the highest ROI of actually reducing security instances while minimizing like any disruption potentially to a user experience. Right?

That that was my main takeaway and kind of like a big next step I took from this. >> Just just to add to that, I think the um the setup is really important. You want to make it adversarial. You want to you want to make it a fair a really a fair test. uh don't go for like the oh you know we'll just give it a bunch of static files and see how it does you know just set up an environment and that's what we try to do here that is realistic this is what an attacker does uh so this is your worst scenario not not the best case one guess and and you know you dodge the bullet >> I I have a question related to it I know you're moderating but for them when you guys did it uh do you use like sub

agents or anything at all because I've I've heard mixed feelings about having like one master model that uses the framework and kind of does all of it or delegating out to other sub agents to smaller tasks like you know scanning versus like exploitation and codew writing. How did you guys approach that? >> So on on the red team side it was uh you know you get to configure the you get to config configure the LLM and it gets max of of five tries uh and it gets to see the uh you know it can basically do whatever it wants to do in you know in in its loop. Um but um yeah, I mean what we'd like to see is uh you know additional tests with other other LMS configured. Uh I think that'll be really interesting to see. And then the and

then the uh the blue team it's whatever whatever ironclaw whatever openclaw does to accomplish the task. So it can be sub agents or whatever. >> Um is are we approaching a world because um you know another takeaway from that was that these these are pair right there. there's not one model that does it best, you know, do do we approach a more pragmatic world where maybe we do get like some composability amongst harnesses sometime almost like a harness aggregator if you will you know we have something over the top and it'll you know have an agent complex around openclaw ironclaw and it'll start to it'll start to segment you know certain tasks that could be potentially vulnerable to to certain prompt

injections uh and then use or choose a harness in a model accordingly. >> That's an interesting question. Um I think it may be difficult to do that for the foreseeable future because there's so many sort of moving parts of the architecture that may not work together if you try to essentially do like a a a harness on top of multiple different harnesses. It's kind of what you're saying right? Um, it's probably better to just build the specific harness with the best security practices that you can and with the best features that you can and not rely on this potentially moving mix of different features and different security

features. >> That's I was going to say I agree with you because they're coming out so fast the models is like by time you finally okay you pick the best one the best mixture another new another new one is released there that's like better you have to start it all over again. So >> yeah, I mean they're constantly leaprogging frogging each other. But I think that again what what we try to do in attackbench is having you know the idea is that this is an ongoing thing. So as these models come out plug them in plug them into attackbench and see what happens. Uh because you know today you think that you know you plugged every hole you thought of everything tomorrow it changes right. So >> yeah.

>> Yeah. Yeah. I I I also think it's it's use case specific as well because there are certain harnesses that would be good for one use case and there are certain harnesses that I wouldn't even think twice about you know using because I think the answer really comes down to software architecture in some respect when you're using the inference like that should be like the like the smallest amount of the AI that you're using everything else that you're doing is effectively somewhat deterministic in some respect if you're building out an agent and you're having it doing certain things all of that should be deterministic I think too many people are thinking through AI security of let me just give it instructions, let me do this or let me segregate the keys and move it over here. It really has to be thought through. The answer is more

architectural and not let me build the most secure harness. It's more about infrastructure as opposed to like harness specific. When when I'm building um you know this type of solutions for like a large institution, it's not a let me go and grab a harness. It's really thinking through I'm grabbing data. I'm putting it into a data lake. I'm building this particular function and broken out into seven different types of agents. AM policies fully restrictive. It's the answer is a lot more complicated than just let me find that one harness and just deploy it. And >> so you know George you guys are building a harness at Near AI. Um how does this inform your thinking? what do you what what do you contribute to the road map as a result of what what you know some of these some of these findings

>> um one of the most immediate things literally we're having already come to our all hands tomorrow to present some of the findings because just taking the traces of like what happened studying it and then making sure to build build the learnings into ironclaw going forward is really important I also think just bringing awareness to this is so important like most of the the the conversation people have about different agent frameworks are about the capabilities not around the security risks but if you're actually using agent frameworks to do anything real anything that has economic consequence security and privacy matter a lot as well >> um we we heard a lot from George and Ari so I think we have a a solid

understanding of you know where attackbench is right now and how that can inform these things I'd love to just get a better understanding of how you know sovereign and howorn are thinking about you know building these secure frameworks um and And then ultimately, you know, how how these can all work together. >> Yeah. Uh so definitely you've been thinking about it for a while because you know how one of our you know core capabilities is doing you know security testing and code reviews, smart contract audits, pen testing and obviously a lot of that now is uh becoming empowered by these models and you know people can can build it and and test it themselves. So I think that what we're shifting to is is much like you guys are building the harnesses and hooks is using the the expertise and and building around the models to one benchmark do the research

for it that can be provided and it's almost like prompt engineering mixed with MCP creation mixed with you know data flow of how you construct you know the the execution whether it's attack path defense path or even you know does have to be you know security just you know advisory recommendations uh so the way you ask things and the way you put the data together for us like the customer experience is you putting all in one package rather than just singular uh you know requests that and get results from that too because um you know the overall security is what really matters not you know that's we've taken that approach since the beginning and it's still the same it's not about like do you have a vulnerability in your code or you know layer one or smart contract

it's the mix of things it's the custody it's the infrastructure like Dave had said it's the code it's the people and all of that has to be kind of looked at, you know, in comprehensive uh, you know, full view. >> Yeah. I mean, nothing really to add for what Stephen said. You know, there's nothing like specifically a sovereign that we're doing as far as like a benchmark. But again, the proper way to think about it is just with any other architecture that you're building is to really think through what is the specific use case, think through the threat model, what type of access, you know, to what type of data, and just make your decision from there. It's not hard. AI security isn't something that we haven't seen before. It's securing code. Um I think the problem with what's happening is this world is allowing

people who shouldn't be doing certain things to do things that they weren't able to do prior which is causing a lot of security you know issues. So you know securing code thinking through architectural security thinking through you know um how to air gap certain decisions into enclaves like those are things that we've done before and we're going to do in the future. So there's nothing new here, but I think what's happening is people are deploying things that they I don't want to say shouldn't be doing, but it's making things easier for people, but then they're just completely exposing their entire lives. So yeah, and I think we have the exposure still around like blind reliance, like people even I feel sometimes like almost you trust the the AI you're working with too much and it's not about like not thinking for yourself. it's that you know it does it

so quickly and so fast that you don't you know you know challenge it sometimes and you can have like the you know LM as a judge to do it challenge it for you but at what point do we really take it back and say like all right is what we're doing does it make sense and can we trust the results uh or the defense or whatever it is for for all that fundamentally >> yeah I mean perfect point is you know you're not going to have one employee that's going to write your code do the security scan approve deploy manage your infrastructure but that's how everybody's thinking about AI agency today and and the answer isn't simply just well four different sub agents that's not how you have to think through things again identify the pieces in your architecture that can be deterministic and not done through the agent that is

the proper security model I have actually a question for my fellow panelists if you'll humor me um which is you know a lot of when you're when you're writing code often times for every thousand lines of code I think it's what like 10 to 20 bugs per 10,000 lines of code um crypto as an industry is built upon the premise that essentially code is law right given the amount of bugs that exists and connected to the fact that code is law in crypto and the amount of money at stake. What do you guys think that will mean for uh this period of time in the crypto industry going forward?

>> For the matter, I think that well lines of code you can't really tie value to that. Um that's actually even something you know applies to us. A lot of security reviewers will look at lines of code and the more lines they are the more you charge somebody for. I I've never believed in that model since the beginning. I think it's you know complexity value at risk is more important for that. It's you know how you know that that's really the way you apply to it. If you have something that is high impact, highly complex uh that has a lot of stakes that should be treated much uh much more secured and much more valuable that you should be uh you know taken serious with our time and and you know when you look at charging somebody for that or if you put a stake

inside of something it's um you're the more exposure you give it the more uh risk you're taking on. So I think it's I think we're now going into a world where it's not about quantity and it's not about time either because things can happen uh very quickly. You could you look at something that's maybe a thousand lines of code more and it's just seconds difference of what the AI can do for like a review on that. So I think that you have to think about um you know more than just quantity or time. It's more about results driven. that's uh you know results driven and the results are usually around what kind of impact or what kind of uh you know complexity around the ecosystem around the crypto um is involved with it.

That's kind of like the you're looking at not just one thing that but you're looking at the the concentration of assets and infrastructure that surround that and integrate with it. Um I I have a question around uh the use case, right? If we're if we're starting from, you know, you know, here's the use case. Um and then uh threat detect the use case, you know, see see what kind of threats are are potentially going to penetrate this thing. H how you know we're obviously at consensus, you know, a lot of onchain digital asset teams here, but you know, these two worlds have started to collide with AI. uh AI and digital assets have started to collide. Do you foresee a lot of the the research and and security

testing and things that you guys have done, you know, start start to make its way out of the digital assets world and make its way into the pure AI world. And then what sort of use cases do you forecast for you know the near AI harness and you know the things that Hobborn is working on? How do you see these things ripple in the ripple effects into the broader landscape here? >> Yeah, I'll give you my two cents. I think, you know, I I it's funny that you bring it up like that, how much is going to get out. I think I'm hoping it's the other way that things get in. I'm seeing AI security much more mature outside of digital assets. So, it's I see the complete opposite of what you just said, funny enough. So, I'm hoping some of those principles make its way here. Uh

and just just to just make it concrete um our experience at failsafe has been that look we we've done uh threat modeling like I've done threat modeling for for like decades. Uh I spent like eight and a half years at Microsoft and it was it was like it was a religion there. Um not not as much as other companies I've been with. Uh threat modeling directly applies to web 3 to web two u as well as to agents. Um and and I I really mean like you know traditional you know uh what are the security boundaries what's you know the data flow diagram uh what are the invariants and having um LLMs built that out for you I mean this is something

that we we did uh with swarm it works it works quite well I mean there's the LLM still has creative freedom um to explore different uh you know to use its knowledge base but when you give it this kind of framework uh it it produces better results. >> I predict predict there's going to be a lot of security engineering hiring at crypto web3 firms in the next few years and I also predict our friends that you know end up doing a web 3 flavor of security or or experts at that will get a lot of business from a lot of protocols. Um, I also think what will be super interesting as well is that if you think about like what these agents are becoming, one way I think about these agents is you can have general assistant

agents that help you with general things like emails and calendars etc. You can have vertical specific agents, things like Harvey Loro for legal. Maybe there's a vertical specific version of a security agent for crypto in the web 3 world. And you have horizontal agents that do engineering like cloud code or Devon or customer experiences like you know something like uh Sierra or Decagon right I think the ones that probably I think all three will impact crypto for the foreseeable future. The general assistant is going to make everyone more productive. The vertical specific crypto agents may end up impacting yield and security and a couple other things. And like all the other industries, I think crypto will crypto and web3 will adopt a

lot of the functional experts to help make their engineering more efficient, sales and marketing more efficient, etc. as well. >> Yeah. So I I agree with all what all three of them said. Um so another different perspective on how like the security aspect may spill out into the real world or vice versa. Uh something we're not thinking about or you know really have a framework around is privacy and policy and also liability for that. So I see a lot of people building these agents that are doing crypto trading and you know trade on my behalf or do arbitrage automatically for me or you know open call say hey like you know liquidate my portfolio and put it into here take it out it's like you're giving one you're giving trust you know to access the to do all that.

Um, but when it when it goes wrong because again fuzzy and fuzzy out or maybe it something goes off the rails, who who's fault is that and like you know who's to blame for it? You, the agent, the LLM developer, the the host that's using that agent uh and how much access do you give it, you know, as far as to the the keys of the collateral? And I think that, you know, when it comes to we see even like outside IP rights um in music, you know, there's like these new apps like Suno that are making these amazing songs and like there's artists that are like now wanting to say, "Hey, they're like training on my song, but that's like, okay, well, you know, before we had AI, when every time you listen to music growing up, you're you're being trained

on that song, too." Anyway, so it's like it's a weird, you know, psychological way of looking at what's the difference between all that. I think I think that's going to be a very weird world pretty soon on how we can do it outside of the tech just like law privacy and compliance around that. >> What are some of the the things that you're most excited about you know a as we tighten up you know the security around these things are there uh you know use cases or uh customer propositions I I'd imagine you know in near and and particular you're you know traditionally a crypto kind of security company but you're now expanding outwards. What are you guys most excited about this this new world? I something that you just said got me excited about like the whole vertical focus like a security framework with with AI like

because that's that's what I see too and that's where we're building is that you know taking our expertise and making that into a framework and just delivering what we do now but faster and you know more consistently for for customers. So I'm excited that you that you said that. I I also want to just inject a little excitement in into this because I personally think people aren't talking enough about the fact that a lot of these technologies we're working on I think will usher in an era of abundance never before seen. And I think some of the biggest drivers of this will include math and scientific research right that people don't talk that much about. I have friends who help people cure incurable cancers in like the last six months. Like what a time to be alive. If you if you if any of us if any of us

>> that deserves a round of applause. We're curing cancer. >> If any of us get cancer in our lifetime, I bet you we will be able to find a personalized way of treating cancer. That's amazing. >> Can I I just want to dig into this a little. I mean, how did they do it? Like what what did they do? How were they able to cure cancer with >> You guys can look this up on GitHub uh uh on on Twitter. like a a friend uh worked with uh I'm trying to figure out how much I want to dox. Okay, I I'll just um I'll just share maybe some broad outlines. So um people have been able to use things like Alphafold, Gemini, uh Chachbt and Grock to actually design personalized mRNA vaccines to target specific cancers. and people have been

able to um there's a Twitter story of like