AI & Blockchain Will Redefine Everything

Transcript

if you're not in crypto and you're sitting and looking at at the space and you know yeah you get like fcoin in the news one day Haku in the news next day you're like this is not serious right there's real you know actual interesting things happening but there kind of don't get like as big visibility as you know some of this mes effectively my teos is over next like 5 to 10 years effectively you either become a super app or die right because you either become this AI assistant yeah or you become like abstracted out by [Music] assistant I'm here with Ilia from near

how's it going good thanks for having me yeah thanks for doing this uh so yeah congrats on this uh this great event like uh I mean well props props guy so I mean I was looking at you guys earlier you guys are like the crypto power couple of Elizabeth like you're up there talking about how like AI is going to revolutionize crypto and how near is going to you know be at the Forefront of it and she's like holding the baby while organizing a conference it's pretty wild uh yeah well she's she's a hero for so just to start generally where do you how do you think about where we are in the space right now like you know we just came out of 2024 the inauguration is coming in a few weeks markets

are I'm sure how to you know parse all of that um yeah how do you uh how do you think about 2025 looking ahead I mean I'm I'm an optimist so uh I always I always look in a very optimistic way I think obviously we see you know tide is changing in us which have been kind of a cloud over over the whole industry for past last you know few years I think one of the challenges with crypto has been traditionally that because you don't know what is right and what is wrong people always like kind kind of invent new things and figure out how to do like what should be pretty simple like I I use this example where you know the you

launch a blockchain right we had the distributed set of validators who launch near blockchain um you know there's code on GitHub there's like genesis that everybody agreed on they launched it and in parallel we need to do a bunch of paperwork right just in case like and and the thing is like if that paperwork if somebody disagrees about paperwork we cannot do anything the network is launched with distributed set you know it's immutable like like the paperwork has no direct attachment or influence on the actual blockchain yeah but because it's not clear we kind of need to do it right and so everybody is yeah I me it kind of covers everybody else so like my hope is we're going to have that addressed and like we actually get like a lot more streamlined and clear rules on on how to do things and that both

enables you know the innovators to like move quicker but it also means that where're before like I get the I get questions a lot like hey you know crypto right now talks a lot about AI but AI doesn't talk about crypto yeah and the reality is like you know why do you think that is well so so I mean I have I have like few reasons I think is that but one of them is indeed like you know if I'm a traditional AI company uh even though like maybe Capital formation crypto is something I want to tap into like I'm just scared of you know touching that and like how complicated this is and you know being also labeled a scam and kind of being uh you know being like effectively losing reputation from this so I think it's like it's

really important that you know as this gets uh more clarity this means like more of the traditional companies that want to participate in kind of crypto economy will be able to do that now the other part is I think this industry needs to like really deliver some good products for uh for this you know other companies and users and I think we have some but it's not like most of them are still targeted at like a highly kind of you know in crypto people mhm and so I think there's like few startups that are doing really good work in the ipace doing good work with prime intellect EXO I mean obviously I think we're doing good work as well but um to bridge that Gap uh and I I think this year we'll see

that really like kind of turning the tide but again it's it's just like because of this you know if if you're sitting like if you if you're not in crypto and you're sitting and looking at at the space and you know yeah you got like fcoin in the news one day Haku in the news the next day you're like this is not serious right so versus like yeah we have people for you know like growing applications to million users in in 10 days or we have folks who are building you know distributed training cluster that's using like like Prim intellect is using hugging phas and all this like VB two partners uh to do that and so like there's real you know actual interesting things happening but there kind of don't

get like as big visibility as you know some of this kind of uh memes effectively yeah so so I think like that's a combination of things but again I think this here we're going to see a lot of this turning around and actually like being being so you think that regular Clarity and also maybe just of maturity political I mean maturity as well but you know the the political landscape I assume you're talking about like what's happening in the US is going to make it easier and perhaps more appealing for non- crypto companies space and start actually looking at crypto you know I think you know a lot of people think heard different predictions that 2025 we'll see like a

major financial institution launching its own stable coin and like if that were to happen I think that would send pretty clear signal across uh trifi that like is sort of an acceptable thing and that probably trickles down into tech and fintech and like Ai and just the broader Tech space in general yeah for sure I mean that I I think like already you know Black Rock during ETF like all of that started a good good wave but again the if you're a startup in US last year if you wanted to bank and you were doing crypto you couldn't right so if you're a startup and you want to use crypto like and you're going to lose your bank that's not a good trade-off yeah so so I think like just

making sure you get you have the banking capabilities you have all this like that would be a pretty fundamental like for people to actually experiment and try things and then Clarity around like how to do it yeah I mean that's a huge um that's a huge benefit for the space I mean I I I do think there is a Counterpoint to that I think it's Nuance so if you look at Europe specifically and and Mika um you know based in France and the the sentiment there is that a lot of the smaller players are not going to be able to to to get the licensing and to also be compliant because the costs are just too high so at the same time that we get regulatory Clarity and I think in the case of Mika the the

regulation is quite heavy on compliance and can be quite burdensome and expensive for um smaller organizations to comply that's just going going to I feel like just kill innovation in those places now if the US has a more uh nuanced more sort of a gradient inovation Pro inovation based approach which you know presumably it should I mean given kind of the principles of um more like us philosop American philosophy around Innovation and and capitalism we should see more Innovation coming from there and well what's what's going to happen in Europe right I think what will happen in was probably you know have binance and coinbase and sort

of large players um putting it on their suitcase so he's and like everybody else uh just dies off yeah I mean like that it I agree that there's like there's a Nuance under this which yeah regulation comes with additional overhead I think there is an interesting story there with AI because traditionally like being compliant with this requires like a ton of paperwork and activate tet communication with a regulator and I I very wholeheartedly expect there will be AI agent that will be doing that on your behalf and you effectively just feed it like the actual information of what you're doing and it effectively sit on top of the regulator and like update

them and give them all information and The Regulators will start like themselves asking for AI agent on their side to analyze all this input because they'll get just get so much information and like so like you you you you effec dos it with AI on one side so they are current of forced to adopt AI on the other side so I'm kind of half expecting this and half you know maybe uh may help some projects to do that yeah um but I think like just in general like AI legal agents are coming and like I I already talked with some projects who are doing this um and uh because a lot of it is literally just like paperwork filing like filling in forums and kind of making some minor adjustments around

things uh right there requir for ma requirement is like writing white paper you know specifying all those things like this is all totally AI doable right now and uh you know you can have like an agent that can like peer review itself and do all those like modifications and then the cool thing is it can keep updating it right as you actually change your real like your smart contract changes it can like you know see the update and like update the white paper so I think just dropping down some of that kind of what traditionally would require like a massive legal firm to do that for you dropping down that cost like 100x and making it also like almost real time updates and and I think it will you know solve some of the problems you mentioned that

small startups are not able to do this and then also will kind of force Regulators to figure out how to make the interface more straightforward yeah well startup idea for anybody out there build a na compliant agent AI yeah if you're interested like let me know I mean we have a lot of infrastructure to do this kind of stuff so yeah I think I think that makes sense like AI is going to permeate through all all areas of business um regulars are going to have to catch up um I'm not super confident that the EU regulars will start using AI anytime soon they'll probably want to regulate it first I mean they trying to I was in Brussels they just got a brand new iPhone I'm still waiting for you know black was that yeah yeah and llama

is not available too yeah so uh yeah I mean it's going to be interesting for sure I think I think Europe is in the it's an interesting position where like they sees this as the way forward to like over regulate it and in turn this means like yeah it's getting more and more complicated for startups here um but like it is a global world and like you can be a startup anywhere and have people in Europe right so uh also Switzerland is outside of EU so uh which we're Swiss Foundation uh so that's that that's one of the interesting checks as well so anyway I I agree that like yeah regulation is is a double-edged sword

but uh I I do think like a Clarity always you know breeds a even if it's like complicated like it'll breed Innovation around that complication then and I think we now have tools to do that so I'm I'm pretty excited about that yeah I like that um let's um let's talk a little about near and you know I think you guys have been really kind of like at the Forefront of innovation in terms of blockchain and Ai and like bringing AI agents on chain and um I was just talking to Alexis from safe earlier and um we we went through like the the different layers of AI agents like the different um kinds the

different kinds of AI agents so starting from you know basically you know cha PT right like give it an input and it gives you some answer and then you know you have to go and do that yourself and then from there is you know the the the the ultimate Pinnacle is like an AI agent that is free to do what it wants and um also can upgrade its own code um like you know on that Spectrum where you think we're at right now as an industry and like what's the big next big unlock that allows us to get to the next I mean as an industry I think with like right now actually crypto industry is behind web 2 in this sense because most of the what do we call agents right now and

just a rapper around CH gbt uh with some personality that is able to like you know participate in conversations um there's some obviously Innovations around like you know managing context and providing memory like for some things but um and people are like plugging in tools to do trading and to do like different things but um it's definitely not at the level where it's like hey this is an interesting product the web 2 has like a I speak with a lot of startups uh people are actually starting to solve V problems right they can replace a marketing agency where you know it goes and like modifies the website to optimize you know conversions and SEO and you know increase

Impressions the people do software agencies right where it affects 10 times cheaper and 10 times faster to develop software like it's a basic you know crude kind of website uh you know you just describe what it is and like generates it and maybe there's a person there who lik you know fixes a little bit but it's like only done in you know 5% of cases so there's a number of this uh kind of models that are that are really you know people are racing toward right now in in broader uh lab two and I think the next there's few pieces that I think I needed for the next unlock that you know we working on so um one is like right now I'm for example and and maybe

I'm like further on the spectrum of like you know being afraid of uh people stealing my data and whatnot but uh I do think that privacy is a you know a prerequisite like if I want to you if I want to give all my private data like my email and my chat and whatever to a you know AI agent I want to know that it's not going to go to somebody else's database it's not going to get stolen somebody else will not look at it right now imagine you do that like you GAC St this now you know uh I have a lot of emails with documents sign Etc like can just you know hey you know collect all those documents you know create me a summary table go to whatever uh I don't know like I have I do a lot of in

Investments you know go to pulley and like find all of the Angel Investments I did and like organize them type thing like you can have a lot more like Rich interaction but even beyond that indeed you can have like okay well you know book me a trip you know with Erica for to S Francisco and like organize in know hotel and this and this and it can like look at types of hotels and there may be I stay and flights at to you know to decide how to do that right so that's kind of like I call it an AI assistant part so this is like a model that runs on your side so it needs to be like wherever you give a data it needs to run privately and I like even open AI uh and kind of you know other providers I don't

think of them as private enough uh you need to be able to have uh kind of that assistant to be able to take actions right on your behalf uh and this is where the money comes in right because like a lot of the things end up doing a trans like actual transaction right and the thing is like in this case you can even you know use credit cards and like this is still a possibility but uh as you kind of uh work toward more interesting use cases and and especially work toward more business use cases like so imagine now I'm you know I'm I'm running a company I like hey I want somebody to generate you know 100,000 Impressions on my website well uh now it can find another AI agent

who is you know a marketing agent negotiate with them you know okay I'll do it for $100 or whatever you know create commitment and then interact so this is where you eff need some way of like inter agent kind of communication negotiation like Contra contracts payments and disputes right and so this is actually what we call we effectively package all of this into intent framework right right and so I think that is right now missing both in web 3 and in web two because even in web two right now you still need to like sign a legal contract and like use the the agent if you want to use some service agent does agent to agent communication

need to happen in a in in in a human language or in English so it's it's really interesting uh just thought of that like two AIS talk to each other like of course you one can formulate a prompt or query goes to the other and it could whatever language but it doesn't have to be like me so it it doesn't have to but like realistically it's probably going to yeah because like natural language is like we know it's a descriptive interface and it kind of describes all the like level of human and you can obviously like you know drop some Json into it or whatever they like data objects but the reality is like that's a universal interface and because it still will need to work with human here and there right it's just like way

easier to use that and and all those language models are effectively trained on maybe maybe they would develop their own language without it was like you know without us knowing when they're interfacing over API they just start talking their own language that would be pretty pretty crazy yeah I mean there's actually like some people have been experimenting with this like more compressed language yeah they can use yeah but yeah I realistically think it's going to be natural language because when you do need a dispute resolution like when something goes wrong you want to go and look and make sure and understand what's happening right so I think there there's a lot of reasons to kind of keep it backwards compatible with hum yeah keep it Backward Compatible with humans that's prerequisite for building AI you think yeah because like we need to go and

understand what's happening and like natural language is like is easy to understand if it like exchanges you know bits and parameter like activations in your Networks you're like what is this yeah when I was watching your your talk earlier like something sort of thinking about what gets disrupted you know because every time there's a new innovation or new technology or we think about like how is this going to disrupt the humans and um one thing that seems maybe it's maybe it's an unobvious take maybe maybe people have been talking about this I don't know but to me it feels like an unobvious take that like uiux designers yeah get uh disrupted or at least uh their their whole uh kind of uh de changes quite a bit because like the AI

doesn't care about the ux like the AI just needs this simplest way to interface with the data and able be able to change the state of some machine um and so like if you go back to the booking or flight example you know the AI can figure out whatever interfaces we have now or it can work over like an API or some other like means of communicating correctly with you know some third party um server and database um like you know a lot of people in crypto are talking about abstraction and like building better ux and like the the you you know building better ux will bu bring bring more users uh um the best ux is AI the best ux is AI right because like the best ux is voice yeah I mean

for humans it's like the best way we communicate our intent is through is through speech and perhaps at some point true thought uh you know if we go like I always think that when like I I often think when I'm using a phone you know this is something I thought about like a lot since I've using smartphones for like 10 years is you know this is such an inefficient way it it like the phone is just literally the is there's this like interface between me and like the final state of my desires just a book of flight L yeah and between like thought and the flight actually getting on the flight there's this weird thing in the Middle where like I'm articulating desires through select fields and like date Pickers and PR yeah and and those

are like yeah highly inefficient but at the same time so I actually my thesis is I think like it will be yeah conversational they're still a place for like typing because like not always you want to talk right and until we can think yeah but the the important thing like you still want to see sometimes a nice UI because the what language is really good at at navigating a highly ambiguous like state right but whenever you get to like really like narrow details actually uis are better and so we have a demo that generates UI so you say like for example you started with like hey I want to swap some tokens or I want to see like what is a you know liquidity pools was there Etc yeah and

it generates UI for you specifically for that problem wonderfly yeah it just writes code for the UI on the Fly and you can like oh actually I want to see you know this information I want to see APR I want to see you know p&l or whatever and so it can modify on the fly as well and and it can remember that like this is how you want it and so so I I think of this is going to still be a combination where like I mean unless we go full thought and like end to end that that's a separate maybe a little bit longer term but even right now we can already have you know to your point of of like chain the final chain abstraction is effectively AI assistant where you say hey you know what what like I have $100 like what should I

invest in right and it gives you like a rich interface with options but you can also keep talking to it or you can you know click and use it so you have like multiple ways of interacting with it yeah say there's also like information that you might want to have like summarized so exactly yeah yeah and sometimes you want to map right you don't like you don't want a language description of the of the directions you want a map right so like you want this Rich interfaces but the cool thing is the I can now just generate the monify which means there's no need for apps as you said there's no need for uis and so you have effectively your AI assistant and this is like my te is over next like five to 10 years effectively you either become a super app or die right because

you either become this AI assistant yeah or you become like abstracted out by a assistant okay so I guess uh aggregation Theory just goes you know to the fullest extent yeah to the fullest extent but it's kind of like it Aggregates like the aggregators get actually just like the current aggregators get this mediated yeah because the current aggregators are effectively serving as this like intermediate the final aggregator is your device right so devices like first we we'll probably see like conversion to super app and then the super app is just going to be shipping their own phones yeah or devices or whatever like form factors we're going to have yeah well that's crazy but that's yeah I mean this

is next you know five 10 years but yeah because I mean look at like all of all right all the current super apps so you think email social network um uh any like communication texting I mean basically you have like all these different we we we have all these different means of communication through uh that we use on our devices like different messaging apps but I mean the the the end goal is just communicate some information to another person and then uh businesses and services uh like say travel booking or like calendly or these sorts of things are communicating with uh another like another computer basically it's like human to computer um

interface and if you can just aggregate all of that into a single um you know sort of one interface where you can do all those things uh those those different services that we have now like that layer can effectively just become apis they don't need interfaces anymore or they need like much simpler interfaces or very very reduced interfaces maybe to to communicate some end State um but even the AI can generate that based on some response AI can generate for you specifically how you want to look like for example for events I want to look at this way so like it'll generate you a calendar app on the Fly for you how you want to look at it and like whenever use C like calendly service like to book somebody else's time or book it yourself or you know get a tickets uh into opera

or something right all of that will end up in this like UI you like and you want to generate so you don't need to like navigate different types of uis as well that right now when you when you look at it from that perspective you would just realize how inefficient everything is it is that's so incredibly in efficient but but I I think the the flip the even more interesting side is to look at the because I actually I did this thing where I mapped out every single industry that exist in the world right so like from you know mining and like uh you know food production and you know what is all the logistics and WS of art housing and you know factories and so like if you map all of this out and you imagine like okay well between that and actually consumer there's like 50 other

layers of other stuff that happens which is like you know everything from all the regulatory stuff to you know kind of aggregator payment processors you know search engines like all the other things that actually like not really like delivering things but actually just like proxying information and kind of they serve as this as this effectively like outsourc decision making right like when you go to Amazon you kind of expect Amazon did some checking on what they listed right effectively you Outsource decision making to them same as you know when you search for stuff they ranked it or Google ranked it your Source decision making right and so if we assume that you have a very powerful decision maker on your device right your AI assistant

that is able to analyze all this information raw information and make a decision what's best for you right not what Google thinks is best for you to make more money of you but what is actually best for you right now you have all of this layers of middle kind of that don't need are not needed anymore that's why I'm saying like it's kind of disaggregating the current aggregators who are kind of serving and really putting all the power to the user and so you can now have effectively a very direct relationship not with even like Amazon or with but like directly almost with a factory right you know hey I want you know I I like to eat this food my AI assistant knows I like to eat this food so it like orders it from the factory and you know it's like the factory

Aggregates everybody's orders you know plans how to build this they have their only agent that like does all this planning and then you know they're shipping it at the right places right so you kind of remove all of this middlemen right now that exist in the business sense right because most of you know if you look at any big city there's this toall Towers those toll towers are not producing stuff they just like there's people there moving effectively information papers and making decisions on some of this level so all of that is gone as wellit and that is and and this is where blockchain comes in because blockchain is a way to really connect this peer-to-peer parties and create agreements between them and do you know Financial transactions and do all the things in peer-to-peer way and have the guarantees that you want

between like semi-trusted parties and you know reputation counting Etc so that's kind of where like AI is affected as this front end that if you know replaces decision making and kind of serves on your behalf and a blockchain is a back end that kind of is uh connecting all the peer-to-peer parties and I mean that's what you know blockchain for your eyes I I I think you just unlock something for me here which is you know this idea that knowledge work gets disrupted from a by AI I uh I I just got the bird's eye view of it which is that effectively all of these information proc estate to change industry right like uh whether it's like banking for inance banking is

fundamentally a a bank is a big machine that's that does State changes yeah and like a lot of humans involved in making sure those State changes correct and the other humans didn't mess up like still thr money all of that stuff just disintegrates when you have an AI That's able to do all that for you and also execute the state change using the blockchain yeah wow that's that's um I I just got my mind blown uh uh so I mean you guys are also like uh working on like intense and so like this Flows In I think nicely to that conversation where you you you have a blockchain that

basically acts as the intermediary between um different Services where uh solvers are are are figuring out what people like what's the best outcome for someone's request like uh solvers also get disrupted by by AI to some extent well well I mean solvers just become AI agent that so and then kind of how we look at it like for us intense is effective this interface between any actor like between and actually not even two actors but any number of actors right so so for example I start the request I say hey hey I want to I mean the example I like where it includes multiple parties is you know I want to buy an apartment and I want to borrow money against the

apartment uh to pay for the apartment right which is like normal thing that it's called mortgage yeah right so I want to buy with a mortgage so I say that right so you have now multiple agents on the back end who are saying like oh I have a inventory of Apartments I'm planning to sell right so they can provide me this options and you have also the lending agents who are like looking at my credit looking at the apartments you're looking at and can evaluate you know oh we're going to cover you know 70% and need to put deposit Etc right and so all of that you know options are being kind of broadcasted back to you your AI assistant can filter It Down based on your you know okay well you work at this in this location so maybe you want something around there so like it can

like narrow it down even already to what you like and then I'm assuming you know for many cases person will still work Muk maybe visit and then you say okay cool you know maybe maybe they don't want to visit maybe the ey knows you so well that like it just buys your dream exactly you don't even have to see it you walking you know it's done yeah but I'm just saying like I mean you can think of this flow right where like it narrowed it down to to the options you you you like maybe you visited it like scheduled already appointments to visit them so you don't need to like you know figure out because it knows your schedule you go visit you you select which one mortgage already kind of you know was pre-calculated and so this intent of like buy this mortgage you

know pay money here you know you put deposit like all of that is like together being computed and settled on chain right and there is also like dispute resolution if the apartment you know there's some issues with the apartment all of this you can roll back this money are actually in escrow you know this legal whatever requirement So Satisfied so all of that is just an example of like an intent and multiple solvers who AI agents who or you know manyi agents participating in this conversation and you finding what's the best lending rate you can receive and you know what other the apartments you you like so like that is an example like of very much web 2 like web 2 standard thing but you know clearly all of the components here like crypto can solve a lot of the pieces you know this lending

to give you better rates Etc uh and you know the solvers here are indeed like sophisticated AI agents like for example lending solver is should be an AI agent you know evaluates risk understands like hey this apartment is you know all the liquid Market maybe I can give a better LTV Etc right so so I think like to me I I I I mentioned this like I think an interesting like Unis swab moment of intense will happen when uh and and we're actually probably going to have it like within next months is uh uh when we have an AI agent running inside a te so inside verifiable secure enclave and and being able to respond to intents right

and so because of this now you can have any Dynamic logic right it can be a DCL style Unis swap trade it can be lending that evaluates your uh credit score it can be anything affec running inside this uh enclave and responding to intent requests and so I I think that would be like a massive enabler because this is also it runs itself right you have verification that this is the code that runs there um so it's fully autonomous right how important are tees in this whole I think te is is a very critical piece because again right now um You you need like you either relying that this like for

example there's agents that exist now they're run by some company or person right so you need to rely that whenever you're calling them they're not going to steal your data they're not going to like uh mislead you Etc now intend solve some of that but you still like exposing a lot of private data to them yeah right uh and so tees are effectively solving the privacy and verifiability and so those are kind of massive unlocks because what because you send your data encrypted to the te you send it encrypted te it only decrypted in that specific computer so there's like a a session key that's created for that session so even if later somebody like broken into tees somehow like that session is already deleted there's no information uh stored um the inference

itself is done inside so AI inference like llm inference is verified as well and so you know which model was run you know like inputs everything temperature Etc and you can have arbitrary other code that you know pulled in information from websites Etc analyzed it and output so you have like a full kind of endtoend system that's fully verified and nobody saw what went in and went out except in is in the te yeah so the gpus h100s work with Intel uh TDX together they have uh effectively uh like integrated that and so you can run AI on top of an h100 yeah oh okay yeah so this is like massive un

lot that happened like in 2024 wow yeah and like you were talking about you know user own data and how important that is I mean that kind of fits fits in with there but um so I guess what you're saying is that we we don't need like this this configuration of having a te on top of an h100 you can still uh trust some third party to be doing that computation you don't need to like have your own H 100 exact to run your own you effectively anyone now can run this and they cannot access what you run and they cannot change like what at the outcome the result right like you have like you have verification that what the code you wanted was run so you know open AI

anthropic like all these big AI companies so what what is they should un board to use our attack okay what's preventing them also from you know just using you know as a as a I think like commercially selling good selling point right like that your data is private of implementing this sort of yeah yeah so I mean not like they should on board on our on our Tech stack yeah I so yeah talk a little bit about the the stack and like what the what the kind of product offering is yeah yeah I mean I think the the biggest problem that like everybody acknowledges was near is like years to building too much and I'm like yeah sorry guys you know we're shipping too much it's sorry but not sorry um but I mean the stack is you know is multi-layered

um because there's a lot of components that need to come together right so obviously we have a shed blockchain so near protocol kind of blockchain itself yeah uh it enables you know root of trust you have identity on it you have kind of you know uh all the ownership kind of smart contracts Etc you have intent settlement there uh and so you like given we targeting you know like billions and trillions of agents and you know billions of people using it we need something that really scales so uh so near blockchain offers that we have built so-called chain signatures so this is the system that allows near accounts to hold assets on other chains oh yeah so you can hold Bitcoin xrp salana so uses some kind of like threshold NPC

yeah it's a threshold NPC that attached to our you know subset of our validators and they effectively act based on onchain transactions that are uh kind of indicate what messages you want to sign orchestration cross orchestration exactly yeah and so the cool thing is near has also uh yield resume so it's asynchronous environment and So within the same transaction you can actually even postpone the execution go fetch some data from another chain and resume the same transaction uh going forward okay um so this offers like a very you know interesting programming environment and it allows to do like multi-chain apps Etc now on top of this we build intents so intents is indeed this like protocol of get something done by

somebody else so uh so the protocol itself is offchain right so this is like I issue intent you you as a solver can you know provide me some options uh that are signed I say yes I select your option I cosign on it and we settle this right and I mean options can contain multiple different uh multiple other solvers can provide like it works together so this is the framework that again for In traditional crypto sense you can do trading you can do option contracts you can do ATC you can do all kinds of stuff with this but it's also you know I want to buy an apartment or I want you know 100,000 Impressions on my website or whatever I want a pizza tomorrow at 5:00 a.m. like

this is the framework that allows to do that and so kind of like inter agent uh kind of contract communication and finally we have the te infrastructure that uh is enabling to run again this both llm inference and agentic code uh right now we have kind of you know python agents we're also going to support like Eliza and other Frameworks so you can run um effectively whatever your agent uh kind of codebase desires you can deploy it there it can be either fully autonomous or you can still you know manage it but the kind of each session is still encrypted and

private and so uh so this is this is a product that like all together now you can offer effectively a number of products right so and this is where things are interesting because because the stack is so versatile you can build so many things right so we have people building you know Prem brokerage right where you deposit Bitcoin which are held by SM contract and then you can use pump. fun and intense actually swapping you know borrow capital and swap into soul and buy something on pump that fund and you as a user just you know deposit the Bitcoin and then using pump that fund right like it uses chain signatures and intense you can have indeed like uh AI uh agent that is serving as Market

maker that runs in it te right so it's fully autonomous and it's able to you know satisfies the intent right and uh kind of do trades um and then you can have you know more traditional like a marketing agent that is uh you know gets access to your website and is able to do changes and again this access is not shared with developers as agent right it's only uh available in the session so lots of different products have been built around the stack and so uh neari offering effectively this um like packaging of the stack into agent hosting so you can launch like whatever agents you want uh they then gets you know enables them to be discovered so

it's the intense uh architecture and uh kind of within that as well we have you know research around better models how to improve agents and kind of other things so a lot of stuff yeah yeah yeah a huge stack yeah and there's a lot of other projects building to be clear like you know all kinds of different components right you know uh more data labeling and synthetic data and you know inference and all the stuff that like kind of uh like we're working with to really like produce a complete solution that is compe comparable to what you get from like centralized offerings okay so what's um what can people expect in the

next couple coming months any uh any exciting things you guys are going to be announcing or uh yeah what ala yeah so I mean we we're finished finishing up the te infrastructure so we just had like a little bit of alpha leaked at our town hall but uh we're going to be we're going to be announcing more in coming weeks um we have uh more interesting things like you know the currently launched intents are on only for swaps but like with more and more chains so we're going to Showcase how this works now with as I said like more sophisticated agents uh that are actually doing both trades but also like other interesting things uh in INB 2 and then you know there's like a number of

Partnerships as well with uh interesting folks to kind of plug this infrastructure in not just in Leb 3 but also in Leb two cool man thanks so much for doing it it's been a great conversation hopefully you know couple uh mind blown ideas to spread yeah yeah true [Music] [Music]