Illia Polosukhin, Alex & Anton: NEAR Is Building The Rails For The Agentic Future (NEAR Panel)

[00:00:00] Speaker 1: and we've seen this as crypto right it's like traditional banks effectively blocking crypto and now all of them are effectively issuing their own stable coins welcome to the roll up our mission is [00:00:11] Alex: to educate empower and rich so you can be more bullish like comment subscribe and enjoy the show all right we are back guys we are here with the entire near crew we've got ilia we've got alex and we've got anton and guys i mean it is clear now what the market had demanded as a response to ai a response to financial markets you guys were skating to where the puck is going this is our inside look into near uh we had a tremendous token 2049 we were live from privacy and quantum day there was so much that seems to have converged right with quantum and ai and in tokenization and all of that feels like it is converging around near near is one of the front runners not only in the technology in the market but also conceptually it feels like you guys have nerd sniped a lot of people i want to talk about that um ilia why don't we start with you and your keynote because you gave a keynote at token 2049 you argued that ai and finance should be confidential by default i want to dig into this um we have the saying on this show that a dollar on chain tends to stay on chain and that's part of the evolution of the tokenization build out is getting more and more of the economy on chain you know another way of saying this same thing is a dollar that's confidential tends to stay confidential and very similarly we're building out this confidential economy and so when you say finance and ai should be confidential by default i want to know what you mean and where you see this going yeah i mean i think if we look back right [00:01:50] Speaker 1: the crypto for the past you know 10 years been mostly speculation the kind of the excitement was like hey we can actually form like raise capital form you know companies one way or another but no real businesses have been really touching this i think over the past you know two years more specifically last year we see more and more of like real businesses starting to kind of hey actually now that there is regulatory now there is enough liquidity we can actually start using this but everybody will see all of our transactions everybody will see all of our uh like payments in income state like income statements you know if you're a public company this becomes like actual like uh regulatory concerns as well right that you're effectively disclosing information you know supposed to before earnings calls so all in all i think that the space is matured to actually be used and privacy becomes kind of hey it's not just like uh you know there's enough like mixing of you know anonymous addresses it's like actually hey we need like a proper privacy where you know telegraphing every transaction to the world and so i think we've seen that grows obviously zcash leading this there's obviously a personal sovereignty side of this as well right as ai is getting better at tracking everyone as ai is getting better at you know finding correlations between your spend uh you know personally as well you don't want all this information be just left out there for everyone to capture and so i would say zcash led the wave of like personal which in this case was foreshadowing i think the business and so the but the businesses are also moving in and so that's kind of where we've been like look uh we believe the like blockchain was supposed to be private the satoshi just didn't have technology to do it and uh and now there is technology so let's actually achieve that it also helps with you know some of the post quantum risks it helps with some of the cryptography risk like you can actually do a lot more uh kind of defense in depth and uh durability for protecting users if you're running privacy because like you know your transactions and public keys are not telegraphed to the world so like you know there's there's not as much just risks of getting getting broken in uh from from some random stuff so uh so i think like this is like a pretty massive enabler and especially in this age of ai the market is taking this seriously this is what [00:04:33] Alex: the market is demanding um obviously we're seeing this from a financial sovereignty point of view with zcash near intense was able to unlock so much of zcash's ecosystem and really start to build out some of the economic functions around it which kind of goes into this everything trades philosophy and i think you know people see near from several different perspectives we've got the chain abstraction we've got the near intense we've got the ai components and i think people are starting to finally put the pieces together how when i chat with my agent this is now a very comprehensive ecosystem of flowing orders submitting transactions filling trades and everything trades alex i'd love to kind of hear your thesis about maybe extending what ilia is talking about from a very fundamental you know privacy centric and confidential centric point of view extend this to how financial assets work on chain today and and maybe what this might look like when agents are using near for a lot of their [00:05:39] Speaker 3: transactions yeah so you touched these uh like what near is and i think that what help you when people are thinking about all ones they are thinking about some kind of narratives around it for some of the ones like zcash there is all already a um mostly uh agreed social consensus about what zcash is for near there is none and uh and i think it is an absolutely amazing position to be in because people are thinking and there are lots of creativity that we are seeing from the uh from from everybody around um some people are saying that near is a market of markets somebody are saying that near is a swift in the world of different countries blockchains uh others are saying that this is a place for the agentic economy uh and there is like more more ideas around this stuff uh what we have always been focused on let's build the right proper robust technologies and enable additional use cases and bring these use cases to the world so that we're not competing with others but rather in combination with with the other technologies that other people and other ecosystems are building we are able escape the zero-sum gain and actually improve the positions of everybody with regard to confidentiality um as like for me it was always the very simple thinking and i remember i recall six years ago when i came into the near uh blockchain we have been discussing with ilia that how wonderful it would be to create a wallet that there is like limits there and there is like nobody is knowing about it because like it's gonna like splits but then we were thinking about uh like pretty artificial things that are coming from the crypto world okay yeah you need to operate the same amounts in order to obfuscate your transactions and stuff like that and the reason why is because the technology was uh was worse than the traditional finance we as blockchain are behind traditional financial systems and and right now with confidential intents and with the advances uh in the zk and tees we are finally able to go and get to the position that is on par with what modern banks are offering to people and this is great because right now we are for the first time ever um able to compete with them and uh it's not about you know killing revoluts or robin hoods of the world um it's rather from my point of view is always given more options to people which means that people are able to better manage their risks if the only thing that you have in the world is just banks and somebody taking custody of your assets then you always live with the same set of risks you're not [00:08:38] Alex: able to hedge these risks we were just hearing from alex about what happens once you have optionality and i think that's truly what we're getting to uh with what is available to us in terms of the tools now available on chain i was just having a conversation yesterday actually with someone who's not really crypto native about self-custody or not or banks or not on chain or not and so alec you were kind of describing this optionality that you're building at near maybe you just close out that point and we'll we'll keep going through uh through the inside look here and to experience this optionality you need to go to near.com perfectly well said um i think you know another another thing that's that's helpful here is to understand you know some of the lens that people might be looking at this obviously the market of markets people you know they love near intense and they use it all the time um the ai component is starting to get a ton more traction on things like open router we can talk about that in a moment anton maybe you can fill in like how you're thinking about things from your perspective we had this kind of funky like oh moment during token 2049 with quantum and justin drake and we had this privacy in quantum day with quantus how are you thinking about this how have you been able to start to prepare near for this eventuality well i mean [00:10:00] Speaker 4: one thing that we need to keep in mind is that near is ultimately better positioned than many earlier blockchains especially bitcoin and better prepared for this transition because of the original choice of the account model that is multi-cryptography by design meaning we can add more signing schemes anytime so unless things explode like while we're talking and as long as we have at least some time to actually migrate and prepare we're in a very good spot everybody's migrating to post quantum today they need to make a choice of signing scheme and for most chains it means this choice will stick with them forever if within a year they need to choose another one that's again a massive massive transition and maybe years of work for us it's like we just roll it out and keep moving so that's exactly what we did with the mldsa 65 today and we're planning to support more schemes as they come and this as they mature so there's nothing stopping us here on the other hand the i'm not very happy with the message that justin shared um specifically i i'm not arguing with the message i'm arguing with the way it was delivered so we were already thinking about black swan kind of events in cryptography like cryptographic disaster all our cryptography become becomes broken q day is one example of such such an event and we were working on it right so this is not a new thing this is not even a controversial message we need to be prepared we need to be able to be agile and to to be able to to switch the signing schemes under the hood or like crypto replace the cryptography the way the message was framed unfortunately remains controversial at least uh from my point of view and what i mean here is if you come from the angle saying ai almighty will break all your cryptography um this is not scientific you cannot argue with this and what justin suggests we all go into bunker mode and we come out in the post ai cryptography when it's safe to come out except there is in in the world where ai can break all the cryptography there is no post ai cryptography or at least there's no way to know that we are there there is always a question of wait but what if ai will be able to break what is available today so this is not very helpful at the same time we are obviously working on things and one thing that i i think is positive coming out from this conversation is that we are talking not about just q day and quantum threat but about broader set of cryptography threats that are emerging and we are now seriously considering and like planning for the way out because so far the blockchain industry was like well if elliptic cryptography is broken today that's it we just shut it down and we walk home there's no way there's no way out of it we can't do anything but realistically speaking we can actually prepare and there's some stuff that we're cooking uh that will will get us on track for for that it's not going to be easy it's going to be quite a significant amount of work uh hopefully we will never end up in that situation in the first place but [00:13:22] Alex: having a way out is is better than not yeah absolutely and you know given that there's a lot of speculation people are approaching this quantum thing and kind of you know not really grasping you know the reality of the situation what's real and what's not illy i'd love to kind of just you know tie some of these things together you know given that near has worked on several things for a long time right and i think from an outside perspective maybe things have looked a little disjointed but you've had a very consistent vision all the way through right we've got the market of markets we've got the quantum work a lot of this is preparing for the eventuality of you know this one cohesive ecosystem where all these things are transacting maybe you can help trace some of these through lines especially with the ai component as it gets more adoption you know i think people are now starting to react to what these frontier model companies are doing you know they're going on open router and it feels like we've graduated from kind of like llms i think people now kind of they realize it they know what they do they're you know working with them but now we've reached the agentic world at least more in the mainstream maybe you guys inside near have been here for a long time but like you know facebook rolls out muse uh grok rolls out grokbot open ai rolls out dot like we are now reaching this agentic world where these things are actually taking actions i feel like you saw this coming for a long time and you've kind of built a lot of the components that the agents needed before we realized that we needed them help us understand how agents are going to utilize the components that near has built so far [00:15:08] Speaker 1: yeah for sure i think my thesis has always been that ai is going to be the interface kind of how we actually interacting with computing and the blockchain is how we coordinating all of the well ais and people and businesses around them right and so that's kind of how we've been building this um and so the blockchain itself right it's a settlement it's the kind of security it's this underlying layer that really enables uh all of this um to happen then you have intense which is kind of yeah markets of markets is effectively an aggregator of liquidity across not just crypto and not just assets but you know we're now starting uh kind of with agent markets seeing everything from you know uh for example an agent that will scan your website and your product to understand how well it's recommended by other agents to you know processing your selfie into a uh professional photo photos and shipping it to you like actual print out in your in mail right like this is you know the the like the intent is you know hey i went to like a professional photo uh in there's like input is like send a selfie right so we see already the expansion of what's possible with kind of intent architecture uh as this things like as we effectively plug in more intelligence into the intent intent network and so and then the ai side that is like how do we actually you know verify have privacy and be able to kind of provide this to all the users and agents in a accessible ideally decentralized way so all those things kind of they all work together toward that you know ai as a front end or agent is a front end kind of blockchain markets as [00:17:02] Alex: a back end and and the confidential part of that right like i think people can understand the financial privacy element both at a at a personal level but then also at a corporate level how it's important to maintain privacy over you know some of these the balance sheet and payroll and some of these financial components help extend why privacy is important for ai generally because now i think people are starting to wrap their head around okay near isn't just for finance okay alex is alex is going vertical help us understand the confidential component around ai more generally yeah i think it's it's very [00:17:43] Speaker 1: very similar to when we talk about fine finance that you need privacy that um there's no world where everyone knowing what you know who you're paying for what you get paid for is like a good idea similarly with ai we're giving it more and more of our data right like it knows everything about your financial state right it knows everything about your health care about your plans about your you know partner relationship advice you know your legal questions and so all of these have very similar properties like and i mean a lot of them actually are regulated to be private and right now these are not very well controlled like there's actually like you know if you're putting your medical data into one of those agents and did none of them are hipaa compliant like there's actually a violation happening it's just kind of like you you put it in yourself right so uh it's kind of your fault but like realistically like look we want to give it as much of our personal data both on per on like you know individual side but also on business side right like you want all of your contacts all your company on your workforce to run through this and that's when you get the most value but then if this is a you know a singular company provider that effectively in plain text receives all your business documents they can literally reconstruct your business right and we've seen some of this starting to happen where you know some businesses effectively feel like their business was kind of either stolen or just like trained into into training data of the model and given to everyone for you know few like few tens of dollars um on the other side so i think privacy is is becoming important kind of throughout the stack and and that's why i mean in a way we are i mean as we're talking talking was like we need privacy for ai and blockchain because ai and blockchain is what's effectively left to power it i think we're we're down again [00:19:44] Alex: no i think i think i think i think we're still going um we are are we down again [00:19:56] Speaker 4: i guess we're back [00:20:00] Speaker 1: are we anton you got us yeah and that was back yeah the world is weird we're entering the it's a dark forest of uh it certainly is and you guys are not running on php right yeah that's what i'm concerned about all right let me ask my agent to reverse hack this yeah all right i'm uh just double checking the live [00:20:29] Alex: i think uh i think uh i think we can keep going ily i think we got you i don't know we're working on anton um but uh ilia you had mentioned um you know kind of the need for extending this privacy beyond just the blockchain for financial assets but also for um you know ai more generally uh i'm curious how you anticipate uh how you anticipate like things like open router and open source competing with the frontier model companies like how do you anticipate these going they've it kind of feels like uh you know open ai and anthropic got a little bit of a duopoly and then we've got you know near as a harness we've got some of these other components and open open router and open source is gaining a lot more traction how do you anticipate this going and where does near fit into the ai conversation more more broadly as a harness and then also as you know a contributor to this general ecosystem [00:21:46] Speaker 1: yeah so for for near ai the goal is how do we enable private verifiable and secure ai right for people and companies and so at the core ai starts with inference and so that's kind of where we're spending most of our time enabling running open rate models in uh kind of fully enter and encrypted way with verifiability right like if you try near ai you actually get like literally certifications with hashes on the prompt on the response signature you know hardware signatures like each exact hardware that it was run on right so you kind of get the full certification that you can validate and check independently so very blockchain blockchain-esque right we even actually just uh launching explorer like thing where if you're using it on other places like brave you can then go and double check that indeed everything is correct right so that's kind of like the fundamentals right that's like at the end this is fundamentals and and if we you know look at the broader market right open weight models have been gaining massively on usage because they actually got really really good like i'm running a bunch of deep seeks right now and at the same time the key uh kind of the price point of this models is you know massively cheaper and like one of the one of the results we have shown uh kind of on the research side as we working also on formal verification is that we can actually do like really heavy formal verification work with deep seek like model with with a you know uh customized harness and it can do on par with claude and with chat gp like with gpt and it's like 100 to 200 times cheaper right so that's kind of where where we already seen massive boost right just like look you can you can you can do so much more because it's cheaper it's faster it's more accessible uh obviously there's no guardrails that like are kind of artificially imposed especially when you're dealing with security and so there's a critical need for this to be available for you know folks like us who are trying to protect systems and ensure that there's uh kind of you know immediate response to it to any attack as well and so i think like that's where we already see massive boost now for some of the harder hardest in the world problems right the frontier is definitely still ahead but you know there's only that many of those problems to be solved and so i think like we'll we already see like this is you know i've been saying this is going to happen like for past few years like we just saw it in last i would say three months that the amount of tokens consumed from open weight models have completely flipped right and we see this starting to affect actually like you know revenue projections and and your labs are starting to do step further right and we're starting to roll this out is uh hosting confidentially third-party models so we we can actually offer a service where a other model builders providers can host with us we don't see the weights right it's all encrypted and then they don't see the user data right so effectively becoming a marketplace of models and uh consumption inference in one place which enables now to have proprietary models that maybe have like frontier capabilities or have a specialized training right specialized data in them to advance and uh still have all the same properties of reliability privacy and you know accessibility through you know for example staking near to get access to them so i would say like there's like an expansion that's happening from there to enable it and then obviously you know integrating this with finance with kind of agentic systems with commerce through agent marketplace there's a lot of other pieces that kind of markets of markets enable on top of this core primitive because one of the biggest problems is in markets right there's verifiability right there's a reason why for any transaction you do in the world there's probably i i did like an ai research analysis so it's not an accurate number but at 50 percent of the cost of any item you buy is effectively legal costs of like procurement of all the supply chain i believe every contract of every piece somebody needs to sign a contract somebody needs to like you know make sure that if something goes wrong like there's a court case like all of that just accumulates right and adds up into final costs and so like that's what we're trying to reduce by effectively bringing both the blockchain verifiability and the ai verifiability into one [00:26:35] Alex: product on intense yeah i uh i i think it makes a ton of sense and and a lot of this is deflationary in a good way like technology is going to allow us to reduce prices so that everyone wins and the barrier to entry for all of these services it just costs less um and i i am looking forward to that efficiency you know but but those guys that are incentivized to like put up the the roadblocks right the signatures like that's their profit center right so they're not going to go quietly and we've seen this and this is kind of a question for both you know elia you and alex kind of at the same time which is that we've seen these companies block agentic finance agentic commerce like you know whether it's booking flights or i think it's amazon now has blocked muse and this is happening kind of all across the board where some websites are putting up uh roadblocks or firewalls for lack of a better term from agents conducting transactions on these websites and so this feels like it's right for disruption this feels like it's it's very arbitrary uh obstacles that are being put in the way of people making free choices for themselves and just using the best technology available to do it and this is going to get innovated and and disrupted very quickly whether it be from confidential ai or confidential financial transactions what do you think is the response to this right because we saw a lot of potential data vulnerability and then i think the whole privacy meta on chain was a response to that vulnerability that we saw in the frontier companies now it feels like we're getting that but in a very agentic way what do you guys think is going to be the response to these websites putting [00:28:24] Speaker 3: up blockers for agentic finance also from my point of view i mean the stable owners and buggy manufacturers all hated the cars right so but somehow like things uh get it adopted and then people figured out it took some time to figure out all the safety around the cars you know with all the seat belts and we actually needed to use one swedish company to that invented the three-point seat belt and then it they out kind of open sourced it to everybody so that everybody can use it uh the any type of a transformation that is happening in markets is not going to be uh easy it will take some time to adopt and all the stuff but uh but in general yeah this is this is a natural way of reaction or for the for the old capital uh to the new niches and to to new markets that are created so yeah um i think that some of them are going to die actually and you know in five years from now we are not going to see some some super big names that we we're seeing right now in top 20 companies in the world and that's okay that's okay now in terms of the safety um with all the stuff i think we need to be vigilant and i really like that there is this uh motion from um from the top ai labs to actually give access to the uh super like the super frontier models like the ones that even like the vast majority of people are not able to have access to use these models towards um uh towards security applications you know trying to self-attack your own protocols and systems um near has all of the necessary accesses to these models so we are constantly doing the uh trying to attack our own systems to make sure that that all of the users are protected and that's this is one of the ways to go um but also there is this another camp of people who are thinking that like imagine if everybody have like imagine this if there is enough people who have nuclear weapons which means that they are not going to use it because inevitably you can start a war but you cannot win and you cannot survive it right so because like everybody else are going to launch the nukes towards yourself so if if many people have a powerful ai systems then uh they're not going to use it for bad stuff just because um like the other ai systems are going to fight back so it is extremely important to make sure that it is not just a single person who has the access to the most advanced ai and there is a there is a wide set of people and then there are different interests and in in all of the stuff so this is another argument towards the user owned ai the ai that is working on behalf of the user that the user is controlling it and then the user is not just like a something like some some skin and bones in between the the his smartphone and in the chair while actually behind it is there is some kind of super corporation that is led by um and a very extravagant person um who is trying to just maximize the revenues for the corporation so yeah ai needs to be in the hands of the people [00:31:55] Alex: yep uh anton elia anything to add here as we think about you know public owned ai and also you know these companies that are putting up guard rails and and preventing some of these ai agents from conducting financial transactions on their websites yeah i think i agree with alex it's it's and [00:32:16] Speaker 1: we've seen this as crypto right it's like traditional banks effectively blocking crypto and now all of them are effectively issuing their own stable coins so i think we're kind of going to see some version of that i think i mean honestly it's even the same thing with google and llms right google i mean google invented llms internally but putting llm in front of search was a very unpopular decision right uh and i mean there's a lot of reasons why not to do it right it's expensive it's slow it's yada yada right which is like all of the things are like you know if you put enough engineering you can actually address but most importantly it breaks your business model right and so that's what we see right now is like if you think of amazon their business model you know that people don't really think about includes sponsored promotions and search results and that actually generates more probably revenue relative to you know some of them like very thin margins they charge on the on the kind of sales themselves i don't know the exact numbers but agents not gonna ignore promotion sponsored whatever right they're gonna ignore you know first result or search result right it's a json array that they're gonna like scan you know in in the same way it doesn't matter how how it's presented and so realistically it breaks down their existing business models and we expect a it's gonna happen right this again my thesis is ai is your front end it's gonna be replacing more and more of how interface is computing and it does now right we see this already and so uh so yeah i mean like one way or another your agents will find a way to work around this and so now the question is like who is going to be uh like adopter of that and provider of that and again our agent market is effectively building toward that vision and creating this market of markets beyond not just assets but actually you know digital work digital goods and then physical goods starting to [00:34:20] Speaker 3: also pop out there is one 30 second very dystopian pitch to everybody here i would like uh like all the listeners to to think to take a moment and think a little bit about this um and this uh it is the it goes the following like google created over the past 20 years an amazing money printing machine uh their business model is utilizing the three words that you're putting in the search bar so that they are able to target you and extract as much value from yourself as possible obviously people who are paying google for the ads are actually earning money from the users the manipulation engine within the ads is working the way that people are able to extract so much money so that the google is the highest revenue business in the world now imagine and compare how much context are you putting inside of the chat bots and inside of the chat ai enabled chat assistance it is orders of magnitude more so google knows about every single user who is using search much much less less than one percent in comparison to what chat gpt knows about you imagine how much money they like open ai and entropic would be able to extract from you after 20 years of polishing of the business model and they are already going in this then and this money is is not something that this is a new value that is created it is coming from people it is coming from influencing their uh their thoughts like in in case we're going to continue going in in this direction in like in five and ten years we would actually not know whether the thought thoughts in our heads are our own thoughts or something that is imposed by the chat bots so yeah a little bit of a black mirror situation here but this is the version of the future that we are fighting against and that's the reason why we are building user-owned here [00:36:29] Speaker 4: i think this is also inevitable uh if we scroll back to the very beginning of google google was indexing the web right and there were people that were trying to gate keep their websites their content against the evil google bots and and what they were but we we've seen how that played out eventually everybody who stayed off google of search engines is is history so this is the same thing you can try to gate keep your your content your websites your tools from the agents but unless you embrace the agentic future you're going to be out of this agentic market and you will become obsolete [00:37:06] Alex: so let me try to put the pieces together here this is inevitable that we are going to if we want to be part of society not go completely off the grid we we're going to opt into giving our context every business every person giving our context to these models so that we can be indexed and we can be part of the agentic future these models are also choosing to take our context make a ton of money from using our context to manipulate the world and manipulate ourselves so that they're extracting money from us and you know therefore keeping it for themselves as far as profit centers and so when we introduce user-owned ai i'm curious what exactly changes i i anticipate that you know we go from a world where we still contribute the context right but it doesn't go into this malicious manipulation engine it goes into maybe like uh maybe just like not a black mirror but just like a regular mirror where it's just reflecting us it's not this dark abyss that goes into you know a black box and then people you know are are able to profit off the black box it's just a reflective mirror right we give the context the context the context gives it right back to us we own it we also retain the value maybe you guys can kind of fill in the gaps i think we all understand the black box malicious manipulation capitalist you know engine of you know some of these centralized ai you know duopolies but maybe you can help us unpack like what is the mechanism that helps us retain ownership sovereignty and value [00:38:51] Speaker 1: in a user-owned ai future i think you just made the meme the going from black mirror to reflective mirror uh for user-owned ai yeah so i mean the core principle of user-owned ai is it's on your side and so i think the the one of the key innovations of bitcoin is actually how to take capitalist forces and create a public good like that that's at the end like to me that's kind of what it was like hey look this is you know have self-motivated you know individuals and and parties and then create something that like all of them together uh kind of contribute to that makes everyone's life better right and creates kind of this new primitive of ownership and so similar idea on user-owned right like it's not that we want to dismiss the capitalism of ai right it's a massive force and you know ai is obviously in massive productivity boost but what we want is to shift that from you know like if you're launching a new model right at the end you like it needs to make more money than the previous model because you just spend a bunch of money right and so like as a company you're always going to be like okay we'll just make us more money and what do we need to tune to make more money right and even when like meta for example came out and said hey we're gonna you know focus on privacy but by the way we're also going to run ads inside your muse right which is like okay well that doesn't like doesn't compute together right um and so well you need a different economic model to actually power that right just you want effectively a way where everyone is an owner of the system they want the system to be more and more successful to provide more utility to effectively you know increase gdp of the system uh because everybody is like economically aligned with that but at the same time like from every individual's perspective it's aligned with them right so again like blockchain is a very unique concept in this sense right you you have self custody of your assets but like by working together you kind of form this network that is valuable and so similar idea for ai right like we need and kind of privacy verifiability uh are these like fundamental blocks that enable that where you know the ai runs on your behalf that you know exactly like you can literally inspect the code everything's open source like when you get an attestation there's like a hash of the code that runs you can check what code runs we had like actually like people who are building on us who run their agents to scan like we have you know small like verification gaps they're like hey you know we just checked your thing and there's like a verification gap here let's make sure you know let's make sure this is like properly traced right so like people actually checking this now because ai enabled everyone to be doing that and so uh so we have like a full verifiable trace for everyone to check for their agents to check that like hey actually yes i'm running on this bottle i'm running on this uh on this hardware and nobody can you know sneak in and peek or change anything like what one way i i say if you want to like manipulate a billion people's minds right now you just join as a front-end engineer into chat gpd right and just set change the system prompt to say like hey you know sadly convince the user to vote for whatever or in this opinion and then every time you're like oh where should i go for you know for dinner it's like well you know if you think that this candidate is uh is great candidate you should you should visit this if you're not maybe you should uh you know read this article like it will literally be in your in your uh like continue like every conversation that you're having and so that's actually what we're addressing here and then from there right again like you know the set the random censorship right where the model declines to do stuff uh like i had i had friends who got banned for asking like geometry questions right i mean i've got stopped for like you know effectively bunch of stuff but like a lot of the things we're doing is like trying to defend near and so like for scanning that um and so kind of overall that's like that's a user ownership right like you control like the no the custody of your of your models like and your prompts is yours and then from there like we need the ecosystem to come together and start indeed building models together fine-tuning them improving them as well over time and that is a harder problem right there's like a bunch of research there's you know a few different startups working on it we have like few few uh kind of rnd efforts on that but but like to start with we need that inference network because that's where both like the businesses and where the current like gaps are in kind of verifiability and privacy yep it makes a ton of sense and i uh [00:43:41] Alex: i i i want to build on this because user owned you know comes with building all of these things refining and tightening system prompts making sure that all of this is verifiable and traceable i'd be remiss and i i realized we we're running tight on time and you guys uh you guys have been fantastic and in spending some time here i appreciate it but i'd be remiss if i didn't ask at least one kind of economic like network economics question especially around this like user owned idea because like alex you and near intense have done a tremendous job generating revenue for this network uh as and and i think we can all sort of see the picture around the collective as it grows as it creates more value there is reciprocal value that is dispersed amongst its network participants and so it almost feels as if the revenue is a subsidy to grow the network to a critical mass to which it becomes systemically important like systemically important in society in ai where people are depending on it it becomes so important that we need to take a greater stake in the network to guarantee its survival and near intense is almost like this revenue story that is helping us ramp the growth to that critical mass at which point it kind of feels like i don't know this financial singularity moment right where everyone is contributing value into the network to make sure that you know the network continues to to grow um and and ilia you kind of spoke to like the the gdp of the network itself um and how you know that is that is kind of uh self-evident and so i'm curious if you guys also see it this way if you kind of see this this this this path where near intense continues to be markets of markets continues to generate a ton of revenue for the near ecosystem and that value is then used to bolster the near ai economy making it more and more important more and more efficient more performant to the point where near is ubiquitous with ai and then it becomes almost like this ai money kind of thing right so kind of like the revenue into like the money thesis is that something that you guys think uh is it consistent with your beliefs internally at [00:46:08] Speaker 3: near i formulated in a little bit different language but in general it is it is pretty consistent i'm going even more radical than ilia goes and i'm saying that the truly user-owned ai should be free and the reason for this is because if there is a business model there then the ai is going to maximize out the business model out of it whether it is like even the fixed subscription like it's going to push you you know to pay more and more all the stuff it should be free but if it is free how are we gonna fund it how like social networks are free but they are earning on the ads right while amazon well it gets the well there are also ads but like in general it gets the percentage of whatever you're buying uber is getting a percentage from your payments for the transportation and that's okay people are good and they're okay to pay more for the convenience of making their intents to come true so from that standpoint if you would think of it near intense is exactly the money-making machine that is required to build a user-owned ai because from my point of view user-owned ai cannot exist without a site business that is working and that is delivering the revenues that is fueling all of that development so yeah it is very very much in line with what you're saying but i'm getting to a little bit more extreme on the user-owned ai side and whether it can generate [00:47:48] Alex: revenue in the long run i like it i like the radical thesis from uh from alex i'm not not surprised by alex's radicalization what do you anton illia what do you guys think [00:48:03] Speaker 1: you want to go anton yeah go ahead go ahead yeah i mean for me like there's like few aspects of this one is kind of security identity trust right that's what blockchain provides and that's where near token is kind of a critical component it provides this kind of like a shelling point for everyone to understand that this is secure and as we get agents as we get kind of more more of these transactions on chain we will need more and more of that trust in it and kind of verifiability coming from from the agents as well right so like i actually think agents will need to hold near to uh to effectively like civil res do civil resistance prove their reputation and kind of have some deposit for security that they're not going to go and do something bad right on users requests so like a slashing yeah effectively you'll have like agent level like reputational slashing on agent market for example so like like same thing we have on validators same thing we kind of like kind of expanding across the ecosystem like that's like one of the core pieces then the other piece is indeed all the revenue generated and kind of buyback right that drives the kind of like utility economy and then you have the kind of i i know alex wants it to be free but like right now at least you need to stake near to get access to compute right and that like either way you know obviously like they're going to be infinite demand for compute and limited amount of compute available so there needs to be some rate limiting of resource and so owning near again becomes kind of the flip side of this so in a way actually i see agents only near to be like civil resistance and slashing uh for reputation and then on the other side that near gives them effectively capacity to do inference on the other side right and so now you have like a full economy that they can transact they can do you know on intense they can fulfill jobs in ancient market they can do inference in near ai and they're secured on near blockchain and so you kind of have like a full agent you know cradle uh in this in this new world [00:50:14] Speaker 4: bullish very bullish anton anything to add there yeah there is like so i'm i'm obviously at a little bit lower level building the protocol itself and i think uh elia missed a couple more interesting important building blocks here uh one being the execution layer and also the other one is secret management for agents like agents to perform tasks they need i don't know access to your credit card they they need or like to your near tokens or to like some sort of credentials that you control and that you entrust to them but you don't really want to entrust those to some random people it's not like you want to give me your credit card number and and run with it so we we have those capabilities um you can't run all the compute of the world on chain and that's why near ai exists for inference but there are trusted execution enclaves that you can actually connect to on chain and then prove to your agent and prove to to all like on chain contracts that it's running the correct thing that's doing what you expect it to do and nobody can manipulate it and so on and so on and then on top of that the our chain signature solution provides you with a way to derive secrets that are fully under your control and can be then also seamlessly injected into those um into those enclaves that your agents are running in and you can run arbitrary tools there with a proof of correct execution which then becomes a very basic building block that can like you can stack it on top of each other and you can chain those into any sorts of workflows and so on and so on so i think that's also [00:51:47] Alex: a pretty critical piece that we're building amazing yeah i mean that this is all important especially as we move towards this agentic future where maybe there aren't applications per se and there's all you know it's sort of these primitives underneath the surface um all of these are different uh functionalities that provide value to the overall system but i mean look uh you guys have done incredibly well you've built so many different points to this network that are now starting to hit critical velocity agents are being added new ecosystems are being added to intense uh there's new personnel in terms of intellectual capital i just saw mert was you know added to uh to uh a board um and so ilia i just want to give you the last word here you know what it what it what it means to build something at this scale as you grow the network you know it is starting to continue uh to amass more and more of that trust identity and verifiability this is all sort of building the network effects that i think you set out to achieve where do we go from here last word and appreciate all you guys coming on yeah i mean [00:52:57] Speaker 1: obviously we continue scaling bringing more assets on intense um bringing more scale performance to the blockchain and then kind of expanding inference capacity and optionality on what you can do in ai and so at the end though it's all comes together right near.com is kind of this pinnacle of this vision and the product that everyone can experience and not break their head in how blockchain works gas fees addresses etc right and that's that's kind of the the goals like how do we really crystallize that vision because i think you know to your point like a near been building a lot of pieces and like because we've been building infrastructure and like that was kind of the the original approach like we build infrastructure and the apps build the experiences and i think realization is like we have the kind of understanding how all of this should fit together and so we should actually bring this and and build this and and kind of obviously alex and uh matt on on your account team been like driving this forward in in an amazing way and so i think like that has been the missing piece i think in crypto is like how do we actually bring that vision that everybody outlined was was blockchain being this like market of markets but it's been in very kind of disjoint to really be kind of an experience that as alex mentioned right can rival you know traditional banking or ideally be better than traditional banking and at the same time you know kind of that provide all the benefits that we've outlined and so more ai stuff is coming more um obviously capabilities are coming and then uh through the whole infrastructure right continue expanding and offering to more developers uh optionality on how they can uh leverage this tech [00:54:49] Alex: i love it and now we have the front end to your point ai is the front end to all of this infrastructure i have a feeling some very very exciting things coming up guys thank you so much thanks for sticking with us through the tech difficulties all right they can't stop our bullishness that that's really what happened uh appreciate you guys being so bullish uh and coming and spending some time here today um we'll see you next time you guys are great partners this has been ai super cycle episode 50 uh with alia anton and alex guys thank you so much thank you