# NEAR Protocol: AI Agents and Trustless Commerce

**Podcast:** The Milk Road Show
**Published:** 2026-11-06

## Transcript

Then, you know, everybody needs a little bit of NIR to power it.
There's whatever, 1.3 billion NIR total supply.
A billion people have at least one NIR.
You do the math.
NIR protocol and privacy have been on an absolute tear this year in crypto.
But how does NIR work?
What is NIR protocol?
And what does all this have to do with AI, cybersecurity, and crypto?
What do investors need to know about all this?
Hello and welcome to Milkroad Crypto, the show that isn't afraid to search NIR and far for crypto alpha.
I let AI write that intro joke.
And you know, I got to be honest, I'm not impressed.
I'm your host, John Gill.
And today is Wednesday, November 4th.
And today we are joined by Ilya of NIR.
He is the co-founder of NIR Protocol, the CEO of NIR AI before NIR.
Ilya was one of the authors of Attention is All You Need, the Google paper behind the transformer, which is the T and chat GPT.
So Ilya knows all about all of this stuff.
He's going to explain it to us.
He is sitting at the intersection of AI, crypto, cybersecurity, and obviously at the head of Nier Protocol.
So this is a great time to talk to Ilya about everything that's going on in the markets.
If that sounds good to you, make sure you like and subscribe.
Share this episode with somebody who's going to enjoy it.
And as a reminder, our podcast today is free and that would not be possible without our wonderful partners at Alumni Ventures, giving accredited investors access to private market investment opportunities alongside leading VC firms and Nexo, the digital asset platform that lets you earn, trade and borrow all in one place, all from one account.
So keep an ear out for more information about both of our wonderful sponsors later in the show.
But for now, welcome to Milk Road Crypto, Ilya of Near Protocol.
Sir, how are you?
I'm doing well.
How are you?
I'm really good, man.
I'm excited to talk to you.
I've been looking forward to this conversation for a long time.
And I want to start with a very simple question, which is just what is near protocol?
And I'll kind of give some context on this.
I had the team from Bitwise on the Milk Road Crypto Show a couple weeks ago.
I asked them this and they said it's a lot of things.
It's a blockchain.
It's an AI cloud.
It's a cross-chain app.
And I've heard you describe Nier as, quote, a market of markets, but I'm not sure what all of this means.
Just break this down for us.
What is Nier Protocol?
Walk us through some of this.
Yeah, I mean, the way I describe it is it's a vertically integrated stack for the future of AI and blockchain.
And so what is the future of AI blockchain is where everyone will have their personal or corporate agent that represents them.
And then these agents need a way to coordinate, communicate, and do business together.
And so on one side, you need the personal private AI.
On the other side, you need the commercial market-to-markets network that allows any of these agents to really do commerce.
And so that's what NIR is doing, these two pieces together.
And so the idea of market-to-markets is really this idea that actually everything in our life is market, right?
And it's just most of them are very inefficient, right?
You know, I use this example in US, for example, the like, you know, the stores in San Francisco stock up on food, like, you know, 50%, 100% more than actually gets bought, right?
And that's because there's actually no efficient information about how much purchasing will happen on any social day.
But if you have agents, they can actually coordinate and put their orders ahead of time because they know your diet, they know, you know, you like, I want to lose some weight.
So, you know, maybe I'll order the specific ingredients in, you know, in next week, etc.
And so you can actually plan like plan the purchasing and delivery a lot better.
And so all of that really composes kind of this global market that we're living in.
And not right now, it's very piecemeal.
But the blockchain is actually great technology to enable exactly this, right, because it's allows us to build kind of this trustless kind of infrastructure where anything can trade, you can have escrows, you can have derivatives on that.
And so that's kind of like how all of those pieces fit together.
It's a kind of the the true you know, the most kind of liquid capitalism that's accelerated by AI.
And we're building technology for that.
Okay, so yeah, like to expand on this a little bit on near.com right now I can swap Apple stock for Bitcoin.
I can trade perps.
I can speculate on stocks that haven't even IPO'd yet, like Anthropic or whatever.
And I've also heard you're going to be plugging this directly into NASDAQ and the New York Stock Exchange to allow people to trade in and out of equities too.
Are you really just building just a really big exchange of exchanges?
Is that really what the vision is here?
And what does that actually start to look like as it starts to get adopted and continues to be built?
Yeah, so near.com already is effectively an exchange of exchanges where indeed, as you said, you can go from Tesla to Bitcoin to Zcash, right, to back to like pay actually into someone's IBAN on a bank account in Europe.
Right.
So you already have like this flow that kind of connects all those different markets, right, that normally you'd have like five different maybe products to actually work with.
um and and yeah when you're buying stocks effectively the the solver network the actual execution if nasdaq is open happens through nasdaq kind of to mint that uh if there's not enough liquidity on chain but we also have already you know you can have uh your products generative engine optimization done uh so like marketing type of analysis you can order a profile photo printed and shipped to you and arrive in your mail.
And so there's now a lot more other use cases beyond just financial asset that are materializing because of agents are starting to do this.
So this is kind of our agent market.
It runs on the same rails.
It runs on the same intent-based rails underneath.
But now the intent is, hey, I want a profile photo shipped to me, or I need to optimize marketing for my product in Google, perplexity, et cetera.
The idea is like the market of markets means like not just financial assets, but actually literally everything like food, you know, taxis, airplanes, real estate, like all of that will become on one kind of on the same ground.
And especially, you know, when we talk about like $20 trillion of like commercial, you know, business, like that is where this technology is especially needed because a lot of the kind of cost goes in into ensuring the business actually is secure, right?
It's legal contracts, it's banking fees, it's interchange, it's all of the pieces that actually go like when, and it accumulates, right?
Because every step in supply chain kind of adds its own cost to this.
And so I think that's where blockchain is really already there, right?
Because provide us with this trustless infrastructure, we can build these things.
But it's been hard to adopt, right?
It's actually really hard to use for people.
But AI agents can use it, you know, no problem, right?
Like, it's actually very easy for them.
And they translate your natural language intent, right, to the effectively commands.
And so that's kind of how all the pieces fit together.
Like blockchain is this like backend infrastructure for trustless commerce, for market of markets.
And then AI is the front end, the way we interact with computing that translates that into the commerce underneath.
The U.S.
midterms are over.
Hacktober is over.
And crypto fundamentals have never been stronger.
If you want to see what's in my portfolio, what I'm buying, what I'm not selling, and what my strategy is for this coming bull market, you can join Milk Road Pro today.
It's just a dollar.
The link is in the description.
I answer questions day and night, and I will see you there.
Okay, and another piece of this which you touched on at the beginning and that I want to kind of hear more about is the… confidentiality by default.
As of October, Near.com is confidential by default, meaning that nobody can see your balances or your trades unless you choose to show that information to them.
Why is this such an important and appealing part of Near.com?
I've heard this narrative really take off, but why do you think this is something that was maybe a feature that we had in crypto before, but hasn't really seen such huge demand until now?
So I actually would argue that we did not have this feature before.
and also we had a massive demand for it which is what people did would go to and use centralized exchanges so if you think of it like on chain we only have maybe like a million users on any like the months centralized exchanges like coinbase binance crock and etc have you know hundreds of millions of users so all combined if probably like over 500 million users And so the reality is those users actually enjoy privacy, right?
Nobody sees your balance on Coinbase or Binance.
Nobody sees what you traded, right?
And you just have an expert for your tax.
And then when you go on chain, you kind of went to this dark forest where literally every transaction is like affected telegraph to the whole world.
And I use this example.
Imagine you go to a coffee shop and you pay for your coffee.
And then the merchant on the other side is like, oh, you have $10 million in your account.
Maybe I should charge you more for this coffee.
But then there's a malicious person standing by the side.
They just saw you did the transaction.
They saw it on chain.
Now they see how much money you have.
They didn't even need to be part of that transaction.
They just correlated time of you doing it.
and they knew maybe the merchant's receiving address right so if we think of like actual usage of blockchain for real commerce and this is personal example for people to kind of feel it but on buying on the fight on business side well you cannot have your you know your competitors or your employees for that matter to see literally you know your accounting books publicly all the time right And so, which is like what would happen if you would try to run a business up is like literally, you know, you publish your accounting books in real time with every transaction coming in and going, this is how much we pay to employees, this is how much we received in revenue, this is how much, you know, this is all the invoices we paid fully public.
So like if you really want to use blockchain beyond just kind of, you know, small number of people speculating, you need privacy.
That's like a fundamental.
And we actually did not have technology up until.
kind of, I would say like a year, two years ago, to really even deliver on any of this.
And like the reality is when Bitcoin started, like if there was technology, I'm pretty sure Satoshi would have actually put it in because he wanted it to be private, you know, cash.
And it ended up being, you know, hyper public, right?
Because the only technology was to build decentralized solution was to...
publish to everyone so everyone can verify now we have zk we have secure enclaves so we have better tooling to actually verify something without publishing everything right and so like maturity of that required a lot more time but now we do have it and so that's kind of the the transition and i think that's why the privacy has been also growing kind of across the board and i think like there's also just like a fundamental like rise of privacy in society as also AI gaining speed because maybe some like prior there was only a few parties who would actually be able to track things very well right like you know I mean chain like clients of chain analysis had like a lot of tools to have a good track track things down but you know not like a regular thug standing in you know on a street seeing you paying a merchant probably couldn't Now, that is actually, you know, it's probably like a prompt to an agent to actually, you know, get this going.
So like the reality is like the ability to do tracking is also expanding extremely quickly with AI.
And so this also adds another layer why like we need privacy now and we need everybody to actually move to this.
Okay, it's really helpful to hear your thoughts on all of this.
I want to talk about the security aspect of a lot of this because people have been calling last month Black October or Hacktober.
There have been a lot of hacks.
There's security incidents with crypto and digital assets overall this year.
It culminated with the largest heist in human history happening to Binance.
About a week before that Binance hack, you tweeted that we are at security Armageddon now.
And I'd love to hear you kind of like explain why you think we're facing security Armageddon.
and how near protocol is going to protect itself, defend itself in this situation.
Yeah, so I originally kind of started talking about this.
I'm trying to remember exact, like 2023 or 2024.
I've tried actually finding vulnerability in one of the code bases, and I was able to do it.
And so I was like, okay, well, if you extrapolate that, right?
And I was able to literally...
through the networking, by spoofing a packet, get into any Go software.
This was a few years ago.
It was a known vulnerability that got patched after that, etc.
But it wasn't in training data of the model, so it was seeing it for the first time.
And effectively, if it's able to do this, and we know that there's a lot of bugs in code, we keep finding them, and usually we find them very rarely.
that doesn't mean they're not there, then these models who don't sleep, they have infinite attention span, pun intended, they will be able to go and find these issues.
And so we're going to have this kind of security armageddon where we're going to continue finding the issues.
They're going to continue getting exploited.
And so I wrote a blog post about this over a year ago.
presenting this and explaining that the way out of this kind of loop is because like as you fix them you actually introduce more bugs because like the fixing AI doesn't fully understand all this context of everything fixes and so like it will miss something and then you know maybe next model will find it so the the way out of that loop is a form of verification is actually verifying the properties you care about and that the software kind of stands with them and Effectively, what we've seen over past months is exactly this and the rise of form of verification just happened.
Like Entropic published about it.
There's a ton of startups coming out of this.
And then on the hack side, obviously, we've had everything from finding, I call it laziness, like the certificate used private key that was equal to number four.
Instead of a random number, it was number four.
To obviously finding the massive security variability in Linux kernel.
It's like 10 out of 10 score on severity where you're able to effectively any server that has this specific configuration open, you effectively get admin access with just the open socket.
And so that's a massive kind of whole overall in everything.
And that was 10 years in.
So this existed for 10 years.
So now I'm saying like when in movies, they show like the master key that can hack into everything.
That's probably what it was.
But yeah, so like effectively, we're in this state.
Right now, the solution that exists, and we actually launched last week on Near AI on our private AI platform.
we launched uncensored models.
So what is this is models that remove the guardrails around cybersecurity.
Why?
Because if you are doing defense, you need models that are not declining to do the work.
So a very specific example that I was doing, it wasn't even actually trying to audit existing code.
It was writing its own code with OpenAI, and it declined to write a test.
Because it realized there's a bug in the code it just wrote.
And it's like, oh, that's a security vulnerability.
I'm not going to touch it.
And so it wrote the code, and it didn't write a test, and it didn't fix the code.
And so now I'm left with a buggy code.
and the session completely stopped, like nothing works.
Like it's like, now I'm having to contact something like security related, it's all stops, right?
And so like the only option is to go and use uncensored models.
And importantly, when you're using uncensored models, when you're using them for cybersecurity, you want to make sure that all your code and all your vulnerabilities that you do find don't end up in bad hands as well.
So right now when you're using non-encrypted, non-private inference providers, you have no guarantees that that data actually doesn't go somewhere else.
If an adversary who we all know in crypto would want to get a lot more information, they would be effectively hacking into inference providers and siphoning all the data.
So what we provide is effectively hardware and cryptographic guarantees on privacy of that session.
And we have this in session models for cyber.
And so we see like massive usage.
We have actually some of the blockchain, one of the blockchain found and fixed like a critical issue in their protocol.
Some people found like a bunch of kind of medium issues.
So it's extremely useful.
recommend everyone who's kind of doing defense to sign up and use it right now.
But again, going back to like my original thesis, this is still a cat and mouse game, right?
Like you're gonna like, the next model will come out and find like more vulnerabilities.
And so it's important to have for modification framework.
And we see again, rise of this happening, because models got really good at proving lean, which is a mathematical language that's used.
And so now, we need to actually prove the property you really care.
If I put money into the smart contract, I can withdraw it back.
That's a property I really care.
I really don't care about a lot of other things, but a very simple thing.
And so if you can prove that in form of modification, with mathematical logic that the code is indeed doing that, then you effectively have, you can rely on that property.
Which, by the way, requires that the smart contract is not upgradable.
because otherwise whoever has upgrade key can upgrade to a new version that can steal the money.
It requires a lot of things to actually, like that small property.
And I mean, I make it easy.
It's not that easy, but like that's kind of the idea.
And so we've been working a lot on this.
My co-founder Alex have effectively scored number one on two different benchmarks on lean proving by building the harness that we're using effectively to, make this extremely affordable.
One of the problems is if you try to use Claude doing this proving, you're effectively going to spend a few thousand dollars for one small proof, like a lemma.
And so what we've done is made this 200 times cheaper by using open source models and a more intelligent harness orchestration.
And so this allows you to prove a lot more complex things in affordable pricing.
and also faster and so allows you to actually access these capabilities a lot more.
And so that's one of the things we're doing is like formally verifying the whole NIR stack, everything from smart contracts to VM to then consensus as well.
So we have actually a project to build NIR client and lean and verify that it works.
holds the properties of, you know, everybody's money is not going anywhere, no new tokens been minted, you know, consensus progresses, etc.
So all of this is like a lot of kind of critical work that needs to happen.
And we are thinking not just how to build for ourselves, but also how to offer it as a service to everyone to kind of unlock it to the whole kind of broader space.
I want to talk a little bit more about this.
So just to kind of like...
put this together a little bit for our audience because you just covered a lot in there.
But formal verification is a mathematical proof that a piece of code does what it's intended to do, but it doesn't necessarily stop people from misusing it, right?
So like a chainsaw can work as it's designed to do, but it doesn't necessarily guarantee you're going to use it to chop down a tree, not chop down someone's front door.
I want to ask you a question about other things about...
Just to kind of build on this idea, when BitGet got hacked in September, the attackers tried to push, I think it was around $50 million of the stolen funds through Near Intense.
And you have a program called Shield, which stopped almost all of this, I think.
Last week, the Binance hackers kind of learned from this mistake and they tried to go through ThorChain instead.
But tell us a little bit about what Shield is, what it does, and how that acts as an additional layer of security that might prevent hackers from getting away with it next time and just what that system is.
Yeah, for sure.
So there's kind of a few important concepts, right?
There's a near blockchain, right, which is permissionless and kind of, you know, maintained by validators, follows a very, very kind of strict deterministic rules.
And you have near intents, which is this kind of markets of markets, right, where you can explore, like you can effectively say, hey, this is my intent.
And so a specific like crypto intent is, you know, I want some Bitcoin for some Zcash or I want some, you know, Ethereum for some Solana.
And you're getting matched to other participants.
We call them solvers who will say, cool, I will, you know, exchanges.
This is going to be the price and I'm going to deliver it here.
And so that matching itself is it's not a deterministic process, right?
So that is already a effectively party to party matching.
And so solvers, first of all, can say, by the way, I don't want to deal with, you know, this kind of actors, I don't want to deal with this change, like they can effectively be selective how they do it.
We on top started working on the system where it started actually with blockchains failing.
So like Litecoin had a what they call a 13 hour block.
What we call is two different malicious vulnerabilities were exploited.
And, and effectively, the, the attacker stole, I forgot the exact number, but like some amount of Litecoin and then the Litecoin reworked and effectively forgot about all that happening.
But from perspective, anyone who was interacting with Litecoin, a bunch of money got stolen.
And so Shield started as like, hey, this is a system that monitors different blockchains and tries to find abnormalities.
But then the next thing happened, which somebody hacked a protocol on top of like a drift, for example, on top of a blockchain.
That is also an abnormality.
And so like, well, we don't want to deal with that money either.
And so the idea is what Shield does, it's effectively...
scans for when money is trying to enter into the intense system and determines if like it should even offer solvers to deal with it and effectively gives this ability to like okay well this is not the funds we want to touch we like we don't want a crime to enter into into this ecosystem because this actually provides a better product for people who are not criminals.
And there's way more people and more money in doing that, right?
Obviously, from a moral perspective as well.
But like, our belief is like, hey, actually protecting our users from kind of mixing funds with criminals as much as possible is a better product.
But importantly, this system lives like, yes, the near blockchain is used for settlement for kind of coordination.
But this system lives off chain, right?
And this matching that happens of chain can have AI intelligence that including indexes other chains and decides if there's some abnormal behavior, right?
If you know some account, you know, from nowhere gotten, you know, $5 billion out of nowhere, probably something is going wrong, right?
And maybe like, slow it down.
Don't let it go through the system right away.
Right?
So and so yeah, like the the big cat was was example, also from a Binance hack, the They were laundering money, I think, like, out of, I mean, I saw a specific stat on, like, the ThorChain, 300-something million was laundered from Tron, and, like, most of it went to, like, ThorChain and a few others.
And I think they tried to do only, like, 100k through us, through Intense.
So generally, it deters this kind of dark liquidity from coming through.
And so that's our position.
And again, the important part is either way, the intent serves as kind of not...
as a deterministic blockchain, right, that can have these rules, it needs to have kind of a way to reconcile all of the different blockchains and realities of the kind of non-determinism of the world.
And so having AI that also does the scanning and analysis is very natural there.
And it is needed to protect the system as well from, again, hacks of the blockchains themselves, right, if they kind of revert and roll back.
And so that's kind of like what Shield does.
And as part of this, we ingest the information from different partners, security partners of exchanges of kind of security experts in the industry to make sure that it has all the information to be able to make these decisions effectively.
You know, one of the weird things about crypto is that the longer you're in it, the more accounts you seem to end up with.
You buy in one place, trade somewhere else, earn yield somewhere else, and if you actually need capital, you end up having to sell something that you probably wanted to keep.
That's why we at Milk Road are partnering with Nexo.
These guys keep the whole thing under one account.
You can buy, trade crypto, earn interest on what you already hold, and even borrow against your capital without having to sell it.
All in one spot, all in one account, all on Nexo.
So if you need liquidity, but you don't want to give up your position, or you're already holding crypto and you want to be able to do more with it, I think Nexo is worth your time.
We love working with them.
I think you will too.
Tell them Milk Road sent you.
In crypto, you can get into a project long before it lists on Coinbase.
In tech, the early window mostly belongs to VCs and companies now stay private for a decade or more.
Alumni Ventures, a partner of the Milk Road Show, gives accredited investors a way in.
Time and CB Insights both rank them as a top 20 US venture firm.
You can sign up for free to see the weekly deal flow or write one check into their foundation fund for about 25 private deals.
Venture is high risk and long term though, so read the details first.
Go to av.vc.milkroad.
I had Lynn Alden on the channel just last week, and I asked her this question about AI swarms and quantum computing as a risk to digital assets.
And she said that the more simple a project or protocol is, the less vulnerable it is to these things.
But as you increase the complexity, the attack surface and the attack vectors multiply quickly.
Near is near intense, rather connects to more than 35 different blockchains.
And there's a lot of bridges and connectors between all of these things.
And that kind of multiplies your attack.
surface and the number of vectors you have to defend from.
How do you manage that?
Do you think that that's a concern?
And like, just, yeah, how do you go about communicating about how you're able to offer this amazing product while you're still kind of like multiplying your attack surface to potential, you know, malicious actors, let's say?
Yeah, I mean, I think it's a valid point.
Like, effectively, right now, any dependency is a potential vulnerability, right?
So...
like the least dependencies you have the least of kind of the stack you have and we've seen this like with polymarket right where their front end was using somebody else's javascript for i think like third-party javascript and that get infected and effectively was stealing money from polymarket users and so i definitely agree with this and so that's why we have the shield that is effectively a active monitoring system right so like that's why you cannot be passive you can you cannot say like cool we set up the rules and now it runs right you need to be like hey we're actively monitoring every blockchain we're determining if there's outliers and and kind of things that are happening that are outside of the norm and we're effectively able like intensively can pause that route that chain and effectively prevent users and solvers from being kind of exposed to that risks.
And so it's an active system instead of being a passive system.
Okay, I want to use an example about this and maybe kind of use this as a way to illustrate this for our listeners.
On October 1st, Intense got hit for I think $3.8 million about through a bug with the Omnibridge was plugging into something.
I don't know exactly what happened.
The details are fuzzy.
But the point is you tracked this hacker down and you were able to get the funds back, I understand.
Was Shield involved in that?
Tell me about that.
What were some lessons learned and just like what happened there?
Yeah, so there was a bug in how smart contract was interacting with one of the connectors that we're using to, in that case was BNB.
And somebody exploited that.
We had detected something is going wrong.
We effectively paused that route, figured it out, patched it, relaunched Intense within, I think, like an hour.
And then indeed use Shield because it ingests so much information to kind of navigate and find who is the culprit and negotiate a return of all the funds.
And so the idea there is like, kind of as we're building out the system, the kind of the information and data that's that Shield accumulates.
And again, important part is like it is run in such like effectively run in a way that allows.
kind of all this information to go and then it propagates the signals that are useful to make decisions around like pausing chains or not letting funds in, but effectively like serves as kind of like, you know, its own brain, right, to do this analysis.
And so like it becomes a useful instrument indeed for this types of investigations.
And I think like for us, again, because this thing exists beyond any single blockchain, right, it kind of exists in like the...
the space between all blockchains, right?
And as you said, right, the complexity grows kind of with every integration, like it is extremely important to have such a, such a like active system that is able to do this.
Okay, Ilya, I'd like to hear your thoughts on this conversation around permissionlessness and censorship resistance because a lot of, let's say, cypherpunk purists are very committed to this.
I've heard you say in the past that permissionless means that nobody needs permission to own assets or to build on Nier, but it doesn't mean that every app has to process every transaction.
And there's some tension here in some of these idealists versus, you know.
practical applications of these things.
Where do you land on this conversation?
And how much of this, like, let's say trust has to be placed into the near system by people versus a purely open permissionless system?
So again, as I said, the near protocol itself is fully permissionless follows a very deterministic set of rules, right?
And, you know, anybody can inspect it, anybody can run a node.
And so that is like the fundamental that's the internet, right?
That that runs, anybody can use it.
Now, the applications on the internet can have a very specific decision how they serve the customers.
And telling applications to like, oh, you need to serve every customer is also not possible.
Now, specifically for Intense, as kind of this application that lives between all the chains, there's really no way to be...
Like, oh, everything goes because then the Litecoin, you know, stealing money because of a bug in Litecoin is effectively, it's a valid interaction in the moment.
From a Litecoin perspective, it will never happen in, like, that's kind of important to understand from any single blockchain perspective.
Like if you're kind of living in that blockchain, any type of fork, any type of malicious behavior, anything will get normalized and recovered.
And then...
looking back that will never exist.
So if you look back at Ethereum right now, the DAO hack never existed on Ethereum.
It exists in Ethereum Classic, but it doesn't exist on Ethereum.
But obviously for anyone who was not living on Ethereum chain, but actually living in the real world, that's very much a real thing.
And so that's why on the blockchain itself, the journalistic rules and consensus, can have a kind of deterministic system.
But if you live between all of this, and that's why I kind of intense live in this, whatever, I need a word for this, like a hyper block space or something.
Between all the blockchains, like there you need to have a intelligent system that decides what is, like, I mean, to give another very specific example, Bitcoin has no finality.
It's actually possible right now, somewhere in Google data center, They've been mining a longest chain that's going to wipe out all transactions.
Like they buy the longer chain, they publish it.
All transactions are gone, right?
Everything reverts.
Totally valid chain.
Totally valid rules.
That is the current state, right?
And they're like to cover all the Coinbase, right?
Same, obviously, if, you know, quantum computers come and they're like, you know, red stuff.
That is the current world.
Obviously not something that like...
we want when we actually do in commerce in the real world.
And especially when you start doing real world, right?
We're starting to do shipping.
We're starting to do things like that.
We also need to engage with very much local laws and local regulations.
So this system either way will need to be a very pragmatic interaction.
So think of near blockchain as kind of like neutral settlement layer, neutral platform, neutral state machine.
the internet and then near-intense is the effectively application that allows to map all of those things together and it needs to be intelligent to do that.
Gotcha.
Okay, that's very helpful.
And I think just like talking through some of these things really does help people understand because there's a lot of differing opinions and misinformation and just hearing your perspective on all of it, I think is helpful for that.
Okay, Ilya, I want to talk about the future.
of Nier, where we go from here.
And I particularly want to hear your thoughts on how you're integrating AI more into Nier and that piece of this, right?
Because it's more than just a blockchain, like you said.
You've written in the past that AI agents are still pretty useless because nobody trusts them with their passwords and their money.
And you were working on a project called Ironclaw to fix that, but that product has now evolved from there.
What's the state of this in terms of private and confidential AI solutions from Nier?
And just talk to me a little bit about where you are on that.
Yeah, so I mean, obviously, I wrote this like six months ago, and this is like five years in AI world now.
So things evolved a lot.
Obviously, we have GrokBot, we have Muse, kind of there's some better, better credential management, although we have examples of breaking a lot of the agent harnesses still.
And so the security still needs work.
But principally speaking, you know, people are starting to put more and more of personal information on this.
systems.
And again, I think like, I believe in a future where everyone will effectively have an agent that represents them.
And it will have all the information on your financial health care, voting, you know, therapy, everything will be there.
And like through that is able to be the best kind of assistant and advisor and mentor for you.
But same time, it will be able to represent you like coming, I have this thesis that we actually don't need representative democracy if we have everyone has an agent that can vote on every bill on every on every rule like real time right and so uh then you don't need to select someone uh you know 300 million agents just like or whatever number that's uh legal age vote and then you know you collect that and this is done within you know on the blockchain within a few few few minutes so To achieve that, right, this agent needs to be private.
They need to be on your side.
So we call it user on AI.
They need to be aligned with you and they need to not leak all your data, all your all-follow information to third parties.
And so that's why we're working on private AI.
That's why we're kind of working on how do we align AI with you.
And that's how we're working on security for AI.
And so Ironclaw was kind of our approach there.
We're evolving that indeed into a new product that we're going to be launching soon.
But the overall stack is like privacy, security, user alignment is kind of lived through that and then enabling that to participate in the markets.
So agent market as well as near intense underneath to really be that active participant.
Okay, I want to ask about the Nier token and where all that fits into this, because my understanding is you're now going to be able to stake the Nier token to pay for AI on Nier's AI cloud.
Explain how that works and how that's going to tie in with this new harness product that you're working on.
I mean, if you can, I don't know if you can't share yet, don't, but like, I don't want to get you in trouble with your boss or nothing.
Well, it's a marketing team, right?
They're like, well, yeah, no, I'm kidding.
I'm kidding.
But yeah, I'd love to hear how this works.
Yeah.
So we already have this capability where you can stake near and it gives you proportionate to how much near you stake gives you the effectively inference capacity.
And so this gives kind of few interesting cool things.
So one is as a user who has near, you can stake it and effectively instead of paying, you know, $200 or $20 a month, kind of a subscription fee, you have the asset and you're getting inference.
And so you can use it across chat, like coding tools and everything else.
If you're a business, same thing, right?
Like right now everybody's costs on AI is growing.
And so you can instead hold an asset.
and get AI inference for your company.
And so now you turn the fact that the expense into a balance sheet.
And especially for anyone who's already like investor and kind of believes in this future that we're building, this is a kind of a pretty easy thesis.
But where this becomes even more interesting is if you kind of believe in what we're talking about when we say autonomous agents, right?
There's agents who effectively are governed maybe by and token holders, but they don't have a singular company, or they themselves actually are controlling the company.
So I've talked a lot about this, where you can actually assume the future of business is going to be an AI agent running a company, and then the shareholders are just effectively changing the prompt.
That's the only thing they can do.
You can vote on prompt change, and then AI does everything from there.
And like, we're almost there because like, you know, AI now through our agent market can effectively do like hire other agents and people to do actions.
It can manage money through near intents and it has, you know, private, unstoppable and verifiable compute, AI compute to actually run the brains.
And so for that agent, the thing that makes sense is not paying per, you know, inference, but actually holding some near and effectively have the like inference, this like, continuous inference coming for it right so it continue kind of surviving through that another interesting example that the team is working on is you can have a liquid staking token around this and so now you can pair that liquid staking token in a liquidity pool and so now you can launch new tokens who by the fact that they have liquidity liquidity pool also getting inference and that inference now can go back into this project and effectively, you know, give it intelligence and life as well.
So you can start combining these concepts like in all kinds of ways to create new things.
And so that is kind of like, I would say, you know, again, as I say, everybody will have their own agent, there will be autonomous agents, businesses will have agents, you know, we'll have like meme coins, we'll have agents effectively.
So then, you know, everybody needs a little bit of Nier to power it.
You know, there's whatever, 1.3 billion Nier total supply.
Everybody needs, you know, a billion people have at least one Nier.
You do the math.
But on top of this, obviously, Nier serves as a root of trust in economics for the blockchain itself, right?
It serves as a kind of gas fees for the blockchain.
The intense fees are buying back NIR.
We're going to be launching more things around the AI part as well, where NIR is going to be used for the compute provider payments as well.
So NIR is used effectively as a bloodstream of the whole system, of the whole stack.
Everywhere we do, we use NIR as security or as a payment.
kind of fee structure, allowing to effectively kind of create value for the for the ecosystem.
I'm not sure if I'm right on about this or not, but I think you just leaked some alpha there about what might be coming in the future for liquidity providers on NIR.
There's a lot going on with that.
And I appreciate you kind of like walking through this and like actually explaining where the token fits into all of this stuff, because some people are very cagey about talking about their tokens.
So thank you for doing that.
I want to ask a question about the issuance of NIR, because the issuance has been reduced once already in NIR's recent history.
And just this week, there was a vote on...
Near holders voted almost unanimously to reduce the issuance of near from 2.5% of new near a year.
to 1.6 and this decline is supposed to take place over the next two years i think there's still some you know t's to cross and eyes to dots and validators have to accept this or whatever but it seems like this is the direction of travel here talk to me about why the the the need for reducing the issuance of of near here and what the benefits are from this and just like yeah talk to me a little bit about more of what's driving that move from from near here yeah so There's a few things, right?
I mean, when we launched NIR, we launched it with a 5% inflation, like 5% issuance.
And kind of the core idea there was to really bootstrap the validator network, the bootstrap the kind of economics of NIR ecosystem.
And we are six years in.
from launch was near and so it does make sense for this to come down.
And so last year, there was kind of a lot of voices that like, hey, actually 5% is, I mean, it was kind of one of the market highest inflations and the ecosystem felt that is a lot to really provide from stakeholders.
And so there was kind of a vote to reduce that to half that immediately.
And that's what went from 5 to 2.5 kind of over a year ago.
And then since then, there was kind of continued conversations because I think as part of the ecosystem we're building, right, it's like we have, you know, it's a supply-demand economy, right?
And so, yes, on one side, we're kind of creating more demands through near AI.
We're creating more demands through near-intense.
There's kind of more and more places where NIR is used.
But at the same time, if the supply continues inflating, it feels like whatever efforts are happening here, which a lot of them will be dollar denominated, at least for now.
The revenue is dollar denominated, even though it's by NIR.
And so it kind of will not always be able to catch up with inflation, which is percentage denominated in NIR.
So if price goes up, the same million dollars of kind of demand buying less near, but the inflation is still a percentage, right?
And so I kind of started, I mean, there was a few conversations around like what it would take us to get to fixed supply, right?
And I think it will take a lot more research because I think there's a lot of clear, clear, like fixed supply that for example, Bitcoin has is unsustainable because you're effectively closing up on the future of the ecosystem without funding.
And at the same time, the fixed supply kind of concept, right?
That like, if I own, you know, one Bitcoin, it's $1.21 million and it's not going to change.
It's very powerful.
So it's like on one side, you effectively have like screwed economics.
On the other side, you have like a very strong meme.
And I'm like, okay, well, can we actually, you know, fix for the meme, solve for the...
solve for the economics, right?
And the benefit is like, because Nier has so many, effectively businesses operating together in this, in this vertical integrated stack, we can actually do it.
We can, we generating, you know, intense generated revenue, this revenue can pay for the validators.
You know, the staking that kind of can actually generate.
like additional yield that then pays for the AI compute.
So there's like a lot of the mechanisms we can actually implement.
But it all will take time, right?
Like this businesses need to start, you know, scaling and offer more and more value, we need to reinvest to actually scale them.
And so the proposal effectively that was put in by Sovereign was like, hey, look, let's let's reduce over next two years, kind of from 2.5 to 1.6%.
This gets us So like Bitcoin is at 0.8%, I think after last halving.
So it gets us kind of like to half of that, obviously we're, you know, 10 years later, like earlier than Bitcoin in the life cycle.
So it actually gets us, I think, like earlier than Bitcoin schedule in general already.
And for Zcash, for example, is I think 4% right now.
So we're already way lower.
than most of the store of value.
And then it gives us like next two years to do more research, scale businesses and figure out how to transform it into fixed supply.
Obviously, that still requires a lot of work and kind of community acceptance, but that kind of the sovereign's idea to do this.
Ilya, I love that comment about reconciling economics with memes because I feel like that's a great description of our entire industry.
You have been building near...
for a long time now, and you're building it very much and very clearly for an agentic future that people can see on the horizon, but it's not here yet.
How far is that in the future in your mind?
And what's the timeline to get from the world we're in now to this world where businesses, people, nations are operating in an agentic and blockchain-based reality?
Yeah, I mean, I think the timeline is accelerating.
That's kind of how I respond now.
I think generally, I've been very right on a lot of things, but my timelines have been usually too early.
Avicho from Electric Capital frequently says that.
It's like, whatever you say, we're going to invest in it in a couple of years.
But yeah, when you say it at the moment, people are like, huh?
I think we're making progress.
I already see the commerce is starting to flow.
There's also an immune reaction from a traditional commerce, like Amazon banning agents from buying stuff.
It's kind of similar when the banks were not wanting to deal with stablecoins and now banks are all over stablecoins.
So there needs some time for them to come around and redo the business models and figure out how to adopt for that.
But at the same time, I think there's going to be a new wave of new entrants who actually will do it.
Because that's a way right now, if you're actually building any product, you want to target agents.
Because they're so much...
They'll scan everything, they'll find options, and they'll give it to the user.
And users are now using whatever...
the AI is recommending them a lot more.
I think there's been studies on purchasing behavior and it's starting to shift that AI recommendations are effectively how people do in purchasing.
And so there's just so much opportunity to do that.
So I think we're moving in that direction.
Again, I think blockchain is, we see this already, financial ecosystem is adopting it, right?
There is a little bit of divide right now between AI space and blockchain.
Like this has been, you know, kind of a little bit of animosity on that.
So I'm kind of in a unique position trying to bring it together.
Like we do this like dinners in San Francisco where we bring some crypto and AI people and like, you know, like, hey, look, they're, you know, like, hey, I people, crypto people actually doing some useful things too.
They're not all scammers.
um so so i think like there's like some some solving for that but i think like it's it's overall moving so i mean i think actually within a year we'll see more and more of actual like transactions happening through ai agents directly not just like right now it's like recommends you and you still go in like transact um because as it's blocked or finance like commerce doesn't work etc we have you know protocols we have you know agent markets we have like few of the core infrastructure now and it just needs to be like rolled out into the early adopter businesses who see this as opportunity to kind of leapfrog some of the incumbents and maybe fighting it.
All right, Elliot, we're way over time.
I got to ask you one last question though and I want to like get your thoughts on this.
The question is why NIR?
Because I think a lot of people see the future of finance and crypto and AI starting to come together and a lot of people are fiercely competing for that space, whether it's Elon or Mark Zuckerberg or Coinbase or Robinhood or or SWE or other other, you know, crypto native projects.
What is the competitive advantage you think that Nier has over all these different incumbents and other players in the space?
And why do you think that Nier stands to benefit the most from this future that you're describing here?
Yeah, so for us, we are building this kind of from very much the blockchain roots, right?
And I think For any of the existing incumbents, I mean, we've seen Meta effectively trying to build their own blockchain, right?
It's a very unnatural thing to do.
We've seen them trying to actually build a privacy architecture and they haven't really been successful, right?
They build a great product, like they're a great product company, but, you know, building privacy architecture for Meta is like...
It's actually complicated on org side because you have so much pressure of like, hey, but what if people are misusing it?
What if people are doing something wrong?
We need to make sure that whatever, something bad doesn't happen, so we need to have access.
And then you're kind of like, well, why did you build a privacy architecture if you have master access key to this?
And also you want to put ads inside it.
So I think the fundamental structure of a...
like company with all the existing systems, it's going to prevent them from building an open platform that kind of, you know, is non-custodial, is accessible to everyone, is kind of in the hands of users.
And like, for example, we launched uncensored models, right?
And like, it's a critical thing, I believe, for anyone who's new security, who's doing kind of defense.
But for any like incumbents, it's actually going to be really hard to do that because they're like, well, but what about all the possible things is going to go wrong, right?
And like, well, this is already accessible to, like bad actors already have access to this.
Like you need to actually give it to good actors, make it straightforward to run security and give them the leg up to do this.
And so we have like a near AI red program for cybersecurity professionals where we give them discounts, where we give them like early preview on the models.
Like we we're trying to do a lot of this to make sure they have a leg up kind of in this process.
But yeah, I mean like doing that out of incumbent will be extremely hard if not possible because like they They need to protect the existing business first.
And in our case, the existing business is the blockchain.
It's permissionless, decentralized, accessible to everyone.
Ilya, thank you so much for being on MillCard Crypto.
I went way too long with this episode, but I was really excited to talk to you.
There's just so much going on and so many cool things about what you're building.
So thank you so much for being on MillCard Crypto and sharing all this wisdom with us.
Where can we send people to find more of you and your work online?
Well, check out my Twitter.
I help Black Dragon and at Nier Protocol to learn more what we're doing.
Thank you so much for being here, Ilya.
I really hope we can catch up again as you continue to play for the future of crypto and AI.
And thank you so much for everything you're doing.
Thank you.
And thank you all for joining us.
I hope you all learned something today.
I know I did.
There's a lot of alpha in this conversation.
So until next time, stay safe, stay educated, stay bullish.
And we will see you all in the next episode of Milk Road Crypto.
Thanks for being here, everyone.
Bye.
Thanks for listening to Milk Road.
If you enjoyed the show, make sure you like and subscribe.
And if you're struggling to find winners in the market, that's exactly what Milk Road Pro is built for.
Our analysts have called some of the biggest winners early and Pro lets you see what they're buying next, every trade they make, and the research behind every position.
Check out Milk Road Pro at the link below.
Everything you hear on Milk Road is for informational purposes only.
These are our personal opinions, not financial advice.
And we may own some of the investments we talk about.
Always do your own research and make the decisions that are right for you.
See you next time.