Inside NEAR: 10+ Years Building Toward the Agent Economy with Illia Polosukhin

Transcript

Guys, welcome back to the rollup. Today's episode is with Near Protocol founder Ilia talking about the state of Near. This is a quarterly series decryting and uncovering the state of the Near Protocol ecosystem from the intent side, from the AI side, and everything in between. Iliot, it's great to have you. Um, and I I know it's been a crazy month, crazy week specifically for you. Uh, but the market is clearly responding to something at near right now. Good to have you back, man. >> Thanks for having me.

>> Absolutely. I think, you know, there's a lot of misconceptions as well with respect to the near protocol. Um, you know, your roots in AI, uh, you know, your your contributions outside of crypto, uh, as well as the the duration uh, of which you've been building this project. So, this episode today is inside 10 plus years building toward the agent economy. Um, you know, obviously there's a lot of fluctuations in the market as of late. Uh, we're recording this on Friday, June 5th. So, this entire week there's been up, down, sideways, and and more. Um, but there's something happening beneath the surface.

Maybe you could just help us understand in your perspective what has actually changed and what is important to the market, what are they recognizing beyond just the price? What has actually changed? what is most important? >> Yeah, I mean [clears throat] as you mentioned right we've been on this journey for quite a while. the kind of origins of near uh from 2017. Um we've been building blockchain component from 2018 went live in 2020. And I think the big change that's happening now is kind of this transition from the tech and architecture buildout to really adoption and

um kind of uh usage focus and as part of this is also I mean effectively flipping the switch on revenue you know kind of enabling all the species really to work with each other uh in in cohesion. >> Yeah. So these things >> Yeah. Please. >> No. And and like and kind of starting to see all of those pieces really like being useful for people versus being kind of uh you know maybe like some more some more conceptual.

>> Yeah. And so I think you know as of recently because a lot of these things are sort of starting to gel together and and and reach a confluence you know near has attracted a lot of new interest a lot of new attention and there might be some people that have joined the community or the network recently or even used you know near's products like ironclaw or near intense that may not be fully aware or familiar with the origin story right they may not realize your roots in AI your roots with the transformer can you catch us up and maybe rewind the clock a little bit on, you know, the the the chronology of how these things happened, you know, when you started near when you contributed and and wrote the transformer paper and and your history at Google. Can you just

kind of put those things in in sequence for us? >> For sure. Yeah. So, I mean, rolling back all the way to, you know, when I was 10 years old, I was really excited about this idea of artificial intelligence. went and read all the science fiction I can get my hands off on you know from Azimov to uh kind of more modern things and I you know even when like in high school was trying to build like first neural networks which you know I didn't actually know linear algebra yet so it was a bit hard [laughter] to do that um I actually on while I was still in Ukraine I was taking all of the online like first massive online uh courses right from Andrew Ang from Peter

Norvick. So I was trying to get all the machine learning knowledge I could and uh I joined Google when I saw this idea of I call it a cat well it was a cat called cat neuron and effect it was the first example where a neural network learned some concept that is like reasonable to us like a cat completely without any supervision without any human saying this is a cat but just showing lots of images neural network learned this is a cat right that was the first example this is like 2012 2013 and I'm like all right we should do that on text is how we express knowledge question answering is how we actually

understand if the other person knows and can you know think about something and so I joined Google to work on question answering and Google is the best place for that because well there's a billion people asking it questions all the time right so there's a ton of data on is there's a ton of kind of um just general impact you can do by by solving that problem and we worked kind of on what were the you know state-of-the-art neural networks at the time and the problem was they were too slow right if you put them into production on google.com they you know as human if you give it 10 pages of text it will take some time for it to read it before it can actually respond to a question and that's not how Google works right Google you you get an

you want an answer at superhuman speed. You want like in 300 milliseconds it should scan all the internet and give you the answer. And so that's really what gave birth kind of ideologically to transformer is how do we actually consume all of the context in parallel use kind of parallel computing we have and give answers translate or whatever the task we try to do. And so it was kind of merge of bunch of ideas. uh my uh kind of director at the time Yakabus Korid came up with like a concept. I wor you know I built like a first prototype uh kind of it had some signs of life and then you know kind of refrigerated a lot uh to make it work and obviously um took

a you know insane amount of work to make it state-of-the-art but that kind of put like a fundamental block where this transformer block that now is used across all of the AI stack now I always wanted to do a startup I always believed in kind of smaller teams making outsized impact by the virtue of kind of you know being focused being kind of in a fastmoving environment and I thought okay well we can take this transformer innovation and you know now this going to unlock massive you know kind of uh innovations in AI and the biggest innovation to be done is in how we build software because

everything else is software and so the Mere AI started with a vision that you will be talking to a computer and computer will write the software for you and so this whole idea of that AI is a front end right that I use now it doesn't it didn't back then we didn't have blockchain as a back end yet right it was AI is going to be the front end [laughter] um and I call it the single player mode right like a lot of AI space like if you come from AI space you're in a single player mode you're kind of like you talking to AI and that's the only thing that operates and so that's what Nei started right we've you know we built some prototypes we had like some interesting examples of this but we

didn't have enough compute right and back then the GPUs weren't you know it's a scale there was not that many of them and uh what we were trying to do is to get a lot more people to give us training data right so [snorts] we had students around the world who would do data labeling kind of uh kind of write small programs etc and they were in China in Eastern Europe in Southeast Asia and all these countries have problem with like receiving dollars right a student in China doesn't have a bank account they have a WeChat pay right Eastern Europe like some have PayPal working some have noted working you know some were wise work somewhere pioneer works like it was like a mash of like payment processors and things that work and done and that's how we got into

the blockchain because like hey if there was a working global payment network that just works, we would use that, right? So, we wanted this kind of, you know, liquidity layer almost that would just works for payments for around the world. And there was nothing at the time that really would match that. And so, we ended up saying, "Oh, okay. Well, we don't have enough GPU capacity to actually scale the AI yet, but let us go solve this problem that we've um uh identified and would have been the users ourselves." and uh we'll you know come back to kind of keep working on AI when things uh things are a bit um further and so that kind of started the near

blockchain uh in kind of August 2018 we launched it in 2020 coming I mean very much kind of trying to put in all the lessons from both developing software in Google in single store uh in Microsoft for Alex where okay well we know we need to scale right so if we want billion users to use it needs to be scalable at the same time we know that users and developers don't want to deal with you know different chain like different chain objects and things like that and so the chain abstraction kind of all these concepts were already back then because we wanted to make it easy we wanted to make it like kind of web two experience with the

security of web 3 uh the sharding the account model which was kind of a pretty pretty differentiated product from the start, right? Usernames instead of, you know, hashes of public keys like all of that. We're kind of like, okay, let's build the experience we would want to use, not you know, kind of for normal people that anybody can on board. >> Yeah, man. And and you guys made quite a few of these architectural bets early, right? You just mentioned them from this background of wanting to have this kind of like web two- like experience on chain right you guys you know launched sharding you know this was kind of Ethereum had this like road map of scaling sharding L2s you guys were kind

of very very forward facing on sharding chain signatures as well in that kind of chain abstraction era now that the market has caught up and is finally ready intent driven execution again we have the intent side of things you know a lot of these a lot other chains have really just focused on like the core base layer and almost like imitating a DeFi NFT kind of blockchain ecosystem on their chain. You guys kind of saw things in a different way. I'm curious which of these kind of feature sets is the most underappreciated right now by the market in your opinion and why does that kind of shape what is coming for Near and how you guys prioritize your roadmap? Yeah.

So I think the biggest underappreciation always went to the account model including back when we were just launching we had people come in and say hey what is this like can I just have a r like a hash of an address like people were like why why is this a readable address? Like they were like conceptually like so like oh we should just do the same thing that everybody else does versus like no this is actually easier thing to use right? Um but this account model is what enables us to roll out quantum um kind of postquantum cryptography for [clears throat] users you know uh hopefully this months versus you know in couple years. This is also what enables

now to keep scaling how intents work because effectively we're putting intents on each individual account. And so like it allows us to keep doing this really powerful transitions, keep scaling things um and uh keep creating experiences where yeah you have your near account but you can use you know Bitcoin address underneath because we kind of put in the right abstraction at the core. Yeah. And that thesis was obviously kind of early to the multi-chain view and now very helpful in the intense landscape and just in kind of the uh I mean people were talking about chain abstraction for a while and it's really just now came to

fruition of like use near.com all your assets all your chains in a confidential manner. that accountbased basically what you're saying is like the the account-based model that you guys chose led to enabling all of what is kind of currently underpinning the near thesis. >> Exactly. And uh we continue kind of investing in components like this right that can actually bring network effects across the whole stack right the kind of chain signatures was another one right which enabled now near account to you to to kind of sign transactions on other chains which enables you know smart contracts now that can effectively be the brain for other chains assets or execution. It enables you know intents

to execute things. It enables you know lending. It enables all of those different things. It also enables encryption decryption for inference for AI inference. It enables private agents and encrypting their data, right? So that you know uh you don't need to trust anyone and you can always get your backup like it it enables all of the things um that kind of like this is honestly like the change is one of the probably most fundamental problems that not solved in web 2 is like the root of trust at the end like right now is all on on government like this known thing about SSL certificates that any like government can actually sign the certificate like have a certificate

authority and can sign certific certificates and so there was examples where somebody bribed some you know it effectively in some random country to sign invalid certificate for like you know known domain and you know fish things. So like what chain signatures is actually solving that fundamental problem like what is the root of trust and like the blockchain is your root of trust. Chain signatures is a way to sign the signing authority and so we're solving like very fundamental problems and then we turning them into like really usable products that people can actually touch and use on every day. And I think this is what people miss because you're almost going a level deeper by by by solving some of these fundamental primitives. they those primitives are

then applicable to both blockchain and AI and so near is in this unique kind of point where you know you kind of started on the AI train you know took a side quest into you know maybe not even a side quest but you know solved the blockchain thing right and then you know are now built bring that back to AI and along the way you've built out these components and you know Andy and I took our own side quest journey into this modular world. This modular world had you know all these you know supposed benefits where we were going to piece all these things together and I think you know in that world you probably have a project that's building just chain abstraction you have a pro you have a

project in there that's building you know obviously so many that that are building just chain the chain right or just chain infrastructure not chain plus AI plus chain right all of these things and so near is a bit unique in this perspect in in this sense because you know you have elements that don't really fit you cleanly into the AI box, but also don't really fit you cleanly into the L1 box. And so, you know, h how do you think of, you know, people understanding Near when they aren't really able to fit it into a particular box? How do you describe what Near actually is? You know, people that keep putting it into either, oh, it's an L1 or it's just, you know, it's just an AI front end, right? How how do you think

about the the total addressable you know potential value of near when you combine all of these components together? >> Yeah, I mean that's a very good point and and also where people are like oh you know like we had a DA product and we still have a DA product like we had you know whatever inference like you have like all the things like you're not this you're not that and like the the general response is kind of twofold. one is I mean we've established our vision that we want to effectively give everyone control of their data assets and power of choice right agency and even back in 2019 I was talking about that we will need to figure out how to solve you know decentralized Google because Google is

something that uh effectively is kind of deciding like the model behind Google is influencing how we make decisions it is also capturing our data Right? So it's violating the vision that we have established. So we will need to figure out how to solve that. Right? And you know LLM effectively came and kind of offered us a solution but now we need to make sure that they're decentralized and actually not affecting our input. But like the fundamentally that's a vision that keeps carrying. You can call it user on internet. you know, you can, you know, specialize in user on AI, but like that that is like a overarching mission that we're on and we continue being on and like we still, you know, we're still early days, early innings

here and that vision is enables us to also have this all these components that like you know if if DA is a product that people want like we had effectively like I think one or two engineers put it together in like three four weeks, Right? It wasn't like you know like it like it's it's there right near has abundance of data availability we can put it together right the you know as we were kind of conceptualizing chain abstraction like this multi-chain world we were also like okay well but how are you going to navigate it how you going to experience it right and so kind of our first version of near.com was actually the blockchain operating system

right because it was actually experience where you can have you know your etheum apps and your near apps and everything in one place had all the assets and you kind of were logging in with the same wallet. It just it like it was prei before we realized actually like all the interfaces will be AI generated. You don't actually need these components that we're building. Um and so all of these pieces were kind of like in the same direction of how do we enable this kind of experience where uh blockchain is this like root of trust identity payments coordination vehicle and then we have this kind of application stack on top that really enables us to deliver on this vision.

>> Yeah. because there's so many like essentially any app you can think of is going to rely on some sort of source of truth, some root of trust as you describe it. Blockchains are the best at coordinating trust and then to your point serving that to the applications that you know rest on top of it and the closer that we can build the applications to that root of trust the more performant effective trustworthy those applications are. >> What's the target market? Right? the target market is all of the commerce and I'm not I don't mean like e-commerce I mean like literally you know moving bits and and actual atoms around the world like every single kind of you know

exertion of for exertion of energy I'm listening like the Lex Freriedman physics pot so like my brain is all like in energies and fields but >> cool >> like effectively every every interaction right like hey you know uh we need to ship something from here to there like you're buying something or a business requesting some service that is all should be facilitated by blockchain because it's cheaper it's faster it's more kind of secure to do that and agents are going to be the ones doing this transactions because now all of us will have an agent kind of representing us like our computing platform right our MacBooks phones watches etc will be powered by AI operating system right at

the end and that AI operating system will be facilitating all of those interactions on our behalf. And so those while like AI is kind of the front end, blockchain is the back end, right? And we're kind of building that vertical stack to power that, but through it all of the trans like all of the interactions are flowing, right? And so that's kind of like crypto, you know, was in was kind of crypto intense, right? You know, 20 billion in volume, been growing, been integrating everywhere, but it's a first slice of that experience, right? We then launched kind of agent marketplace to showcase this broader idea, right? Where you can uh you know build a website, run a marketing campaign. Uh there's a really cool example of an agent that does uh venue scouting if you want to do an

event right? And so that's a good example why it's actually a inter inter needs interaction because you can ask your agent your claw or whatever to find your venues and it will suck because it has no understanding of how good the venues are. It hasn't been inside them. It doesn't have a database. Doesn't know where to look. But a events agency that built an agent that has a database that actually you know looked at all those venues and scored them and like added a vibe to them etc. Now their agent is way better at this. It doesn't need to consume as many tokens and you can just hire it on the marketplace right. So that's an example of like this. Now what normally you would literally hire an

agency to do right to scout venues and do this. Now you can hire suza your agent can hire their agent to actually do this work for you know a few dollars. So this is starting to transition like the you know digital work already into that but it's it is an intent underneath and it goes through our consential payments infrastructure you know settled on the blockchain on the both sides you know powered by uh private inference because you don't want you know everybody else to know which venues you're looking at so they don't snatch it before you do right so all the pieces really starting to work together >> yeah I mean that's the thing that a lot of people are very excited about is this like kind of like cohesion of the once you know criticized fragmented product

suite really seeing the vision come together and I think that is like the key the key component here. I think confidential intent specifically has been rising a ton in terms of TVL in terms of volume and I actually saw one of your one of your tweets about it that was like look in the future I hope it's just confidential I hope it's not a button that I have to click right it's just going to be the standard and so why do you think that this privacy is like an architectural level design component rather than like a feature or an asset level or like a button on the UX I think you feel pretty strongly about this just being the standard for near intense going forward also can charge a bit of a premium on the fee side as we've seen uh from the product. Why is this so

important at the architectural um actual layer? >> Yeah. So I mean it's kind of similar to everything else we build is it's comes from combination of pragmatism and and vision right so vision is like we we believe things should belong to the user and whenever everyone else knows that you have something that's not really like a true belonging right so like privacy is kind of embedded into that value into that vision at the same time like pragmatically speaking you know I I have struggled using blockchain payments for my day-to-day stuff. Whenever I do use blockchain, which I do for payments,

I have literally an account per kind of party I'm paying to, right? I need to create like literally a new wallet for every uh counterparty so that they don't see where the money are coming from. And then you know you were using like centralized exchanges to move money around to kind of like kind of somewhat break the the flow right obviously Zcash was you know crossave was near intense offered a solution to that but then you obviously taking Zcash volatility and uh um and so confidentiality is like it's enabling for this to be really mass market payments and kind of transaction while delivering on the value that you know assets should belong to you in a kind of

true way where it's not, you know, and everybody else sees what you're doing. And so that's kind of where it come from. But again, if we think of like agents and businesses are starting to use this, like you cannot imagine a world where a business will want to broadcast to everyone how much money it receives, how much money it's sending to its suppliers, you know, and everything in between, how much it's paying salaries, right? Payroll, all of the things. Like I mean we as Near Foundation I was like writing our finance team always like why are we not using Near to pay for like Near blockchain to pay everyone and they're like well you know [laughter] like it's kind of all transparent like um and so we are now actually porting

everything and going to like a lot of a lot of payments will be going through confidential intents and as we adding like hopefully in in in in coming months adding more fiat support so you can actually like offramp into fiat from there on ramp into it, right? It becomes kind of the place to to do all of this uh financial transactions and then for agents for all of those things to really benefit from that uh as they starting to transact as well. >> Yeah. Yeah. And so it was really fundamental as well as some of the primitives that you mentioned earlier to ultimately bring about this vision that you had, you know, since you were 10 years old, right? and and agents are sort of now starting to coordinate a lot

more of this activity. Privacy was a primitive. Chain signatures were a primitive. These things weren't solved yet. You had to go and solve these things in order to ultimately bring about the vision that you talked about. Now we're seeing those play out. We're seeing a lot of these components start to fit together, start to compose with one another. You know, one of the components we haven't talked so much about is the near token and you know what role that plays in the ecosystem. I think that's where a lot of people are very very curious especially because you know the hyperscalers foundational model companies are going and IPOing and they there's a whole conversation we could have about that you know and then the other side there's this whole swath of

L1 network tokens but again like we said near doesn't quite fit into either one of those buckets right so how should we come to understand the near token what role does it play in this agentic world as as more and more of the the the activity comes onto near and you know we we'll get into maybe buybacks and these things but you know generally yeah what is what is near value acral look like >> yeah I mean conceptually I think of near as this you know lifeblood of this vertical integrated system and so it's a AI money right you know we we gonna we're gonna arrive to this world where

you know agents are transacting on your behalf. They need their own kind of infrastructure. They need their own way you know of [snorts] using AI. They need their own way of transacting and Near is providing all of these and kind of offering it. And near token is kind of the like the the loading beam right that all of this is sitting on right. Everything from economic security to you know collateral and and kind of infrastructure bits to indeed you know transaction fees and buybacks to you know our house of stake in governance and kind of being like governance at the end is like you know root of root of trust right is like the uh like fundamental system. So like all of this

pieces really kind of concentrate on your token >> and then Ilia kind of more tactically and pragmatically how does that how does that kind of flow into the token from the current perspective and kind of take us through what the what the future may look like from inflation deflationary measures etc. Yeah. So right now right so at noon we announced effectively the you know intent protocol turning on its fee switch right and so it started effectively using its fees to buy back near and it's going back to kind of the protocol and and it treasury and so

that's already you know doing like 20 to 50 million annual right now depending on the day on volatility. Um it's uh so that that's already kind of driving a lot of a lot of the buy buying back. But this is again this is just a start like the transaction fees we're taking on crypto transactions are pretty small. If you think of you know business transactions both the volume the scale that we can take and uh you know the amount you can actually take there because you're actually providing trust you're providing escrow you can do pro you could do insurance you can do all kinds of derivative contracts on that transaction you can also take uh a

larger a larger part of this all of these things right going to continue contributing and same for inference same for kind of agent hosting like all of these pieces are really kind of this in a way like revenue streams that are going back into near token. And as those grow the idea is to start offsetting the uh kind of the payouts to validators, NPC providers and and other components. And so ideally, you know, yeah, as any company, you want to start balancing your P&L and and actually delivering on delivering on uh kind of, you know, deflation of the token supply. And so

that will require kind of changes in some of the like how inflation works. And we we have some interesting ideas on how to kind of have a compounding effect. uh kind of the the the the frame I usually use is you know uh kind of sovereign treasury, a sovereign fund of c of a country, right? You want something like that which effectively is able to continue generating more uh wealth for uh for the public goods of that country. So kind of similar concept would need to be built for near um and so governance and kind of the pieces will need to work together to achieve that. But that's you know more revenue

more adoption those are key components to drive that and then kind of continue shaping the token economics under the hood to deliver on deflation and start ideally you know uh kind of get to you know fix slashred reducing supply. >> Yeah. And there's a you know strong revenue potential here as we think through some of the you know the revenue sources that you described. Um I is there a level at which because you when you initially started out kind of describing the value proposition here AI money you know I I think there's certainly a a value proposition of all of these revenue you know flowing into

the token as you talked about ultimately making it deflationary. Is there a level above that where you know we've heard you know Bitcoin is digital gold store of value people play around with the ETH is money idea is there a vision that near is used as money between agents for not just paying for transaction fees but beyond this like the forms of the the form of money you know medium of exchange unit of account these sorts of things does that is there a level beyond revenue where that comes into account.

>> Sure. Yeah. I mean the the way to think about it kind of obviously security itself is very much measured and near right and then as we go into for example inference uh we actually launching you can consume inference by staking near right and then kind of all of these products right you will be able to use near to either reduce your fees or consume the the resources of the network And then on the other side the you know the kind of provide the GPU providers of our decentralized confial machine learning they're getting paid in near the distribution partners of the near intents are paid in near right so near

is effectively is unit of account of all of these kind of different parts of the ecosystem as it continues kind of growing. So you mentioned one thing that I didn't realize that I I think is very true to this thesis which is that the confidential machine agents are paid in near like these these are agents they're essentially employees of this service this maybe decentralized company and their paychecks their salaries are in near and we know that agents are ultimately going to resemble a similar workforce take on like an agentic consumer economy they're going to need to buy services and products for their own, you know, use. And because they're paid in near, they're

going to spend in near and near has this marketplace where they can buy and sell services. And so Near becomes a unit of account in the agentic financial world. >> Yeah, that's where we're targeting. So, and and we providing them like inf for example AI inference as a service that they can pay near or just take near, right? To uh not deplete their base, right? So like you can kind of do all of this uh all of the things uh with and again in this case actually token is a very unique venue to do that right entropic is not going to be like oh cool you know use your entropic equity to pay for your entropic subscription.

>> Yeah. Yeah. I got to ask you man you've you're working on so much in the AI space adjacent with crypto and you're in San Francisco. Um, aside from, you know, being pe people hating on you for working in crypto, what what keeps you up at night? What are you spending time on thinking when you're not coding or building? [laughter] Uh I mean adoption is obviously like my my main uh thinking like what is the like how we can kind of bring all the pieces together to to make it easier for people to to and and and deliver value in short term

right because I think like everybody's like talks about agent agentic commerce and all the things right and there's a lot like there's a lot of conversations but like when you go down like one two levels. is like, "Okay, well, I'm already paying with a credit card and you know, I get credits and like credits get settled once in a while. Like, why do I need this?" Right? Like that's where things kind of like all right. Well, and so like that's why to me these things are really about the whole relationship of the transaction, not just like the payment, the you know, moving money. But you know, yeah, when we think broader, it's like, okay, well, people still don't realize that they give effect all of their data to, you

know, kind of this big AI companies, they don't even realize that the, you know, it takes like one line in a system prompt and effectively can start affecting how you see reality by effectively like if you interactive with AI, it just give you different things depending on that line system prompt. And so those are things that are kind of how do we educate people about the importance of knowing kind of verification etc. There was like kind of a conversation about verifiability and like in crypto we we kind of naturally follow that is like oh you know the things needs to be verifiable but most of the world doesn't really think about

that. Um and so you know for example for neari we've been going through the ISOW and like so 2 and all this licensing which is like you know our tech is way be way more than the than those requirements right but you kind of like to get into the conversations like people kind of have those as like uh checklists right and so like how do we actually reconcile this >> fundamentally better solutions with some of the misunderstanding of or just not knowing of things, right? Like we go when we go talk to, you know, potential clients for near AI, most people don't know it's possible to have

kind of private and verifiable AI. >> They're like, oh, I mean, the default is like, yeah, I need to trust, you know, this provider. Like that's that's how web 2 works, right? You kind of trust whatever SAS you're using. >> And so people don't realize there is an alternative. And like if you start mentioning blockchain, right? Then you kind of lost them. So you like what is that what is a transition we can take people on? What is a journey we can take people on that they will end up in this world that we all bought in but you know meanwhile get value on the on the way there.

>> Definitely. Yeah. I don't think it's been uh it's not it's not getting any better for you out there in San Francisco with the current state of affairs either. I'll tell you that. sell um thank you for joining us the state of near we're going to do these quarterly guys and I think next next one we're going to get more into the data into the stats and into kind of like into the nitty-gritty this is like the opener uh the story of Near the background of Ilia and kind of how how this has all came together in the in the midst of a lot of change in the industry there is kind of a new changing of the guard and so Ilia you know we we thank you for your I'm in uh um it's been a it's been a hell of a month. [laughter]

>> Indeed it has. But you know this is this you could say the same about like April that was hell of a month [snorts] and every month before that. So it's uh we're living in in exciting and at the same time very uh frighting times, right? So again, I do think like, you know, when I talk to some AI people who who don't even like for example, a lot of AI here folks don't realize how badly crypto was hit because of the hacks and uh you know when I kind of talk through this and like the need for formification and things like that, right? Like people like oh yeah that that actually makes sense. So like there's just like also a big gap on kind

of the crypto still kind of lives in its own uh bubble and so yeah bridging bridging that gap has been kind of you know part part of what I'm I'm doing and and really like how how can we offer the solutions we've built here so that actually people as I said start on that journey but yeah it's been exciting times for sure. >> Yeah. Yeah. And it's important work. You know, time and time again, we see how crypto kind of lives in its own bubble and then, you know, every once in a while there's pioneers that come along and take it into the real world. And, you know, you're you're doing that into the AI world. So, appreciate you fighting the good fight and uh being a great partner here at the rollup and uh yeah, thanks so much for coming on

today. >> Appreciate it guys as well. Cheers.