Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
RWA Yield Meets AI Agents: Automating Onchain Income
Recorded: Sept. 3, 2025 Duration: 1:00:24Space Recording
Full Transcription
Thank you. Thank you. good morning everyone can you hear me? Harsh, Chris, welcome.
Yes, I can hear you. Exciting to be here today.
Hello, hello.
Hi, Aaron. Hello, hello.
I think we're just waiting for Dylan. I can see it on the audience And we're good, right?
Dylan, can you talk?
Hello, can you hear me?
Nice and clear.
All right.
Nice.
Yeah, then welcome everyone to episode number 12 of the R2 AMA series.
I'm Aaron, leading institutional efforts at R2, and I'll be your host today.
So we've seen RWAs like TBLs, credit, and funds explode as one of the hottest narratives
last in this year.
And at the same time, you know, AI agents are reshaping how we think about automation in DeFi.
So now put these two together and you can get a glimpse of what could be the future of passive and automated on-chain income.
So we're pretty excited about this topic today.
And to discuss it further, we brought together a killer panel.
So joining me today are Dylan from Fada Network, Chris from MC Square Finance, Sydney from Gaia Network and Harsh from the R2 protocol.
So welcome everyone. To start I'd love for each of you to quickly introduce yourself and share about how your work connects to the themes of RWA, Jill, and AI.
So maybe we can start with you, Sydney.
Hey everybody, my name is Sydney Lai,
and I've spent my career predominantly in workflow automation.
So that could be 4H and 500 companies, could be Web3 companies,
and specifically previously in the past, when it comes to hardware, I also worked in satellite companies. So be Web3 companies. And specifically, previously in the past,
when it comes to hardware, I also worked in satellite companies. So glad to be here. And
I am the CTO of Gaia. So we are pushing the boundaries of AI inference and edge computing.
Awesome. Thank you, Sydney. Exciting what you guys are building. Yeah, maybe we can move on to Chris.
building. Yeah, maybe we can move on to Chris.
Hey guys, excited to be here. I'm Chris, co-founder of MCSquare Finance,
former board member of the BBC and supported also the EU with a lot of
crypto activities and insights. And two years ago, we have been challenged with
what would need to happen to make DeFi mass market
ready and get institutions to fully move into DeFi. And that's how we created Enzyska Finance,
where we list the best DeFi strategies as ETFs directly on the stock exchange,
which means we can double the yield on the stock exchange for crypto while directly funneling institutional
capital into decentralized finance protocols and applications thank you chris excited to
know more about it uh dylan you want to go next sure hi everyone um nice to see you again uh see
some familiar faces i'm dylan kowalik. I'm leading developer relations,
and I do research in artificial intelligence,
specifically for privacy and confidentiality hosting.
So at Fala Network, which is now at Fala.com,
we're kind of like a Web 3, but also like a Web 2 team,
and we focus on all things confidentiality,
mainly with building out an open source TEE
framework or a trusted executional environment framework. And on that framework, it's called
D-Stack. D-Stack is open source. And so I'm happy to share all of the insights and how all that kind
of works because, you know, AI is not so private and it's not so confidential in the world right now.
So we're trying to make a change.
Thanks guys.
Awesome.
Looking forward to hear more.
Harsh, please go ahead.
Hi, hi everyone.
It's a pleasure to be here today.
So about R2, what we're trying to do here is we're trying to bring the Wall Street grid yields to everyone.
Basically, we're taking down that barrier that previously was with there for only big players, big investors to the Wall Street grid yield so uh without any kyc without any hassles with a clean ui with a clean
ux uh we're uh able to deliver a four to five percent of apy and 10 to 12 percent apy which
are coming from real world assets like u.s treasuries money market funds and private credit
so uh yeah that's that's basically what we're building and we're connecting uh traditional
finance we're correct connecting rwa with crypto with d5 so yeah that's that's what we're doing
awesome institutional great deals for everyone uh so moving on uh let's set the stage with the big
picture uh you know rwa is like fresh release and credit, as Harsh mentioned, have dominated the narrative so far.
Some say they're going mainstream, other things were still early.
So I want to open the conversation and ask, do you think RWAs has truly gone mainstream yet?
And compared to traditional finance, what makes on-chain RWAs yield most attractive for both users and institutions?
I think we can start on this one with you, Harsh.
Yeah, I mean, if I talk about, you know, RWEGL has become mainstream, I mean, not yet.
Rw has become mainstream. I mean, not yet. Protocols like us, projects like us are still
on the rise and we are seeing a lot of, you know, adoption by crypto towards these TradeFi partners.
A lot of this partnership, these kind of partnerships are happening between TradeFi and crypto.
So this is just the beginning of the era. So if I just give you another mention of a product like us.
So Midas has partnered with Fastenara.
They're also bringing multiple RWEs on chain.
So this is just the beginning, right?
But there are certain challenges that we see every day.
It's primarily liquidity. It's fragmented. It's not there. It's
not as easy as swapping ETH for USDC and vice versa. It's just not there. It's just not enough
today, right? Data option is still kicking in. The UX is kind of broken. But, you know,
broken but you know as I mentioned R2 is trying to focus on improving and enhancing the UX for
the retail to kick in you know so that the retail can easily onboard themselves so that the retail
can easily you know get access to the kind of yields that we are you know providing uh and and um i i feel the
the third one is if you think about it a a lot of regulation has uh fallen into place but uh
uh it's not complete yet it's it's just for stable coins and it's just for you know keeping a record of it but but i feel uh you know
the there is a huge fog within the regulatory compliance i feel which which you know kind of
does not allows a lot of players a lot of threat fine institutions to not operate into certain
countries which is kind of a bummer i
would say because a lot of these countries have a lot of liquidity lying in so uh i mean in in my
perspective these are some of the challenges that we we we face but uh these are challenges that we
also face you know as as a protocol uh and since i've been taking care mostly of the part of the
onboarding and dealing
with my partners so this is something that uh you know can be clearly seen with a lot of partners but
yeah there's there's still a long way to go uh so that before rw becoming it seems it is just 26
billion worth of uh market right now and i think this is this can easily like explode to a trillion.
And it's a possibility once we reach a trillion, 10 trillion is not that far.
And yeah, that's what I have to say.
Those are just my two things.
Thank you, Harsh.
Thank you, Harsh.
I think maybe Chris, you want to add on top of this?
Because you are more on the rwa side of things also
yeah like i i totally agree we are still early like currently we have about 60 65 billion tvl in in rwas which is next to nothing um we the the current big estimations like i just was on the phone last week with ubs
they expect the crypto market till 2030 to go um like 8 to 10x maybe in general the defi market is
20 to 30x but then the rwa market is more than 100x. So there we see this is kind of the new narrative
that will push forward the bigger scale
of the market growth.
And so we are very early in the structures.
And we see that also that even for experienced
DeFi experts, building up a good strategy
is like looking into the needle into the haystack where
can i find the yield that is stable enough that prove itself most of the protocols do not have
a standard risk assessment yet because like if you talk about rwa you need to fully review all
the risk and especially how custody is handled.
And all of these things is currently still done offline
and manually by a big team of high-level DeFi experts.
So there's a lot of friction that further can be automated.
And the more it's automated,
the higher the yield will be that
people can access it.
The huge opportunity is that if you look into traditional real estate, for example, which
is the biggest RWA market, you get 7% to 12% on average.
That's typically rather easy to get, but that is not enough to operate in decentralized finance very well, because here 12% is already something that you can get comfortably.
If you know how, you also can find some opportunities for lower caps where you get 20%.
And I think this is the direction where we are heading generally in the market. If a market would be highly optimized, things are automated, risk management is also automated and can be evaluated in real time, then I think we are heading into this direction.
And I think that would be an amazing opportunity that we totally disrupt how wealth management
is done all around the world.
And so it's a very exciting future.
Very exciting future.
And yeah, moving on, we talk about RWAs, now the string in AI.
So AI agents are becoming the execution layer in finance. So I want to start maybe on this one with Sydney
and ask what unique capabilities can AI agents bring to on-chain investing?
Things like, I don't know, automated strategies, risk management, something else.
And if you think like AI agents and RWA yield could make DeFi investing
feel like a self-driving car experience?
Yeah. I mean, I think that we can actually talk about RWA and AI under one roof, right? So if we
think about RWA and the peer, like T-bills, credits, funds, I think where AI fits in with RWA
is that one of the pain points that i see in rwa is liquidity fragmentation
so rwa yields you know they're scattered across different chains different protocols um it's
really really hard to optimize right um we have professionals back in the day all right so the
professionals you would see you know on the stock exchange or they're like active traders or anything like that. And so when you have this liquidity fragmentation, how do you create programmable AI agents
that can handle programmable assets? And so that they can actually, frankly, do a lot of these
trades and rebalancing and optimizing faster, better, more efficiently,
and on a much grander scope, probably compared to humans, for sure.
But compared to professionals, well, I guess it's the professionals that trained the models
for these AI agents.
So I actually do think that there is a Venn diagram between how AI handles a liquidity
fragmentation,
as an example.
Thank you, Sidney.
Maybe you can also add on top of this, Dylan?
A little bit, yeah.
I mean, there's the compliance of how an AI can meet these types of standards that was
just mentioned. So at least on our side, for what we see,
how AI and FALA come together is mainly because
artificial intelligent bots and private chat, for example,
is a little hard to achieve.
But when you train an MCP server
and you give the model some context about say like the regulations or
say, Hey, this is what's currently going on to the, uh, onto an exchange. Like, what are the
rules? What are the regulations? How is it that I actually list this asset? Um, and if you don't
actually have those scripts, say if like the code needs to be written, or maybe you have to have
something that's like sort of kept in check, How do you always make sure that the prompts that you're soliciting to the AI don't actually
change?
So long term, we can see the AI not only acting as sort of like a guardrail, because in the
TEE, you could see that if the model context protocol has some sort of like, you know,
compliance check that it needs to run over
and review on say of your market exchange or some component in your architecture, you
want to make sure that it's consistent, right?
And that there's no sort of like prompt injection attack.
So that's where I see a lot of this merging, because once you have a contract or really
just a program inside of a TEE, you can get what's known as a remote attestation
about that program,
which means that you can provably prove to an institution
that the program has been running
in the same exact way for years,
which is really important,
especially when you're trying to be things
like HIPAA compliant or GDPR compliant.
So when you wanna be able to create a marketplace,
you also have to remain compliant.
But if you're trying to be automated, you have to make sure that the model context protocol doesn't change.
And then that's what remote attestations provide to you.
So when you combine all three together, you can really make something pretty good.
So something I want to dovetail to Dylan's point is that, you know, he talks about TEE and he talks about regulation. So I think that TEE works really, really well, especially in the case if you're building AI on
hardware, as an example. I mean, a lot of AI is on hardware, but I guess the point that I'm trying
to make is like if you're hosting inference on a hardware. So to be explicit, we are launching
our Gaia phone. We're able to launch an AI inference locally on the device. And I think
that when it comes to all the different types of ways you can securitize and enable regulation to
follow those regulations that Dylan talked about on a device, I actually think that TE is a really,
really good example for securitizing AI that is locally on a mobile device. And a mobile device could be like a phone.
It could be, you know, we remember Google Glass,
Meta Glasses, you know, all these different types
of like IoT devices as an example.
And again, if you talk about RWA in the sense of
a physical security, it could be maybe a device like a car, a phone, a house, something like that.
I mean, we talked about smart homes as an example.
You know, the future is absolutely fucking crazy.
Like, just think about like having a smart house being an RWA, you know, 30 years from now.
That sounds amazing.
That sounds amazing.
And yeah, I mean, also moving on for retail users,
you know, the barriers to access RWA yield are still there.
Some from custody to complexity.
So what do you see as the biggest barriers today?
And with no code AI agents rising,
can we really imagine like a future where someone just types,
I don't know, I want 5% stable yield and the system executes it you know automatically we can start with this
one uh on this one with with you chris i think that's a great question um the if you say i
how the question would be is hey i, this is the assets I'm having.
I would love to have about 5% stable yield, but I try to have it as the lowest risk possible.
What can I do?
And then it can give you free offers.
What kind of backing do you want to have?
What kind of stability you want to have?
of stability you want to have.
It's like, how old is the protocol?
It's like, how old is the protocol?
In what kind of safety and security do you invest?
Or like, how is your assets insured?
And it's all three things that you then have to decide
which kind of protocol you want to select.
What probably happens much more,
and this is one of our labs currently doing a test for that,
can we answer the question hey I
have 10k available I would deploy it to stable coin yield what should be a good
basket for me probably you say okay four or five percent you put in there like
60% of your assets then nine to% to 12% probably through your curator where you closely watch how this curator changes
and how it adapts its yield, where you get 9% to 12%.
And then have like 10%, 15%, maybe maximum 20% of your assets
into something much higher where you maybe even generate 20%,
30% or even more to be more experimental,
so you understand more how DeFi works.
And I think that's a very interesting setup to generate yield from.
It is possible in DeFi, but the amount of data,
the amount of question that you need,
and the amount of custom APIs you currently need to get this information
makes it a very, very hard problem to solve.
I will open the mic here to Sydney, Dylan.
So I can quickly say that like when you're when you're trying to guarantee these certain
yield percentages like the AAR, for example, of a, maybe even of a particular corporate bond.
Let's discuss a little bit how it works, right? Because I've only recently, I've done a podcast
with a couple of these RWA folks and have even had a little interview here and there, mainly about
like what's really going on, what's really popular. And it mainly has to do with like stable coin
yield, right? And it mainly has to do with, hey, I have this yield,
I want to be able to achieve some said target, like maybe even five to 10% over the long term,
like a stretch of about three to five years. So the AI is not necessarily like engineered unless
you actually engineer it to achieve these very specific results. Once you do, and you have some
sort of predictive curve, then you place it inside
of a TEE so that you could actually say, hey, if I promise the actual server using my private key,
I want to make sure that no one else actually sees my trades or even sees my commitments to
these very specific strategies. So there's another thing when it comes to compliance is
making sure that you're actually, in terms of anonymity, that you're protected.
And so that when you're running these private keys on a server and that you're not doing it locally,
how do you do it virtually in the cloud so that it can be running at 10x the scale with thousands of other people all at once,
basically banking into a specific yield strategy?
Well, you have to basically leverage a light client.
You have to leverage your keys, right? And so how do you protect these keys? You have to put them
inside of a black box. And if these black box are intermingling with a particular AI, you want to
make sure that the AI A doesn't steal your private key and B that you can actually verify that it's
not, you know, stealing the private key, that it's actually conducting the actual orchestration of the yield strategy alone. So that's how I see it. Like if you want to
guarantee that 20%, you want to guarantee that 30% that Chris just mentioned. Well, A, you want
to be able to verify the strategy and B, you want to make sure that the entire guardrail is set in
place so that you can actually intermingle funds between you and thousands of other participants. Otherwise, you're basically working with programs that are, you know,
basically sneakily, you know, backdooring into the protocol. And you don't even know if the host of
the protocol is even kind of leveraging it to their own benefit. We see this a lot with
decentralized exchanges. Even with Binance, we always know that something is going on. They have
the stats, they have the analytics. So how do you compete against bigger players? Well, you got to make sure that the
entire protocol is anonymous. I mean, yeah, simply put TLDR, Dylan, with what you said,
it sounds like there's just a lot of wallet complexity. And, you know, I think when you
talked about earlier, just what is the user experience?
The user experience right now, we have humans trying to basically manhandle every single wallet login information.
There's different wallets for different chains.
You know, so ideally, when you are using kind of like that low code example that you talked
about earlier, which is like, hey, I want X amount of percentage and yield.
Ideally, you don't
personally have to essentially manhandle all the wallet complexities that can be automated.
And then just like a really, really high level, guys, let's go back to basics, right?
When you are prompting all of these agent toolings where it's like, hey, do this for me, do that,
make this financial decision for me, make that financial decision for me, you need to give it context in terms of like, what is your financial
situation? Like how much capital can you deploy? What is your, what are just your taxes in general?
Do you need short-term gains, long-term gains? Age is another one because like oftentimes I hear
like, hey, make this agent, do do DeFi I want to be rich.
Rich is literally too vague so you guys you guys got to remember to like just really really fine
tune your prompting so that it actually makes appropriate yields that's both realistic and
also just profitable for yourself that makes sense. Yeah absolutely absolutely. I agree on that.
It's more complicated than it sounds.
You have to consider different variables.
Harsh, I don't know if you want to add something on top of this.
To be honest, I've been hearing the conversation.
It's interesting to see.
One of the things that I would definitely want to share
is you know this space is growing of AI agents uh uh a crypto project uh founder that I know
of recently I met him like uh like three weeks ago so they're uh working on uh on on, you know, something like a chat security agent, which, you know, just you just prompt it and it will make the agent or a DAP, whatever you wanted to make it.
Right. So I think what I just said, me also said, like, it's about the efficiency of the prompt.
Right. Just telling it to be rich rich to make me rich or you know
that's that's kind of vague but to be very specific about the prompt it's it's where uh the success
lies so yeah uh i mean it's it's just interesting to uh know and uh hear these conversations so yeah
like like few months ago actually mark andreason you did something like that right
through an agent like you can like 50 000 i think you asked him to make him like as much money as
possible something like that right i don't remember the name of the agent by the way
if you guys remember please go ahead like as an example there are looping strategies where with stable coin yield you can get 300 400 percent
but the issue is if one of the stable coins or protocols that are included which are typically
three or four just deep x 0.1 percent then all your capital is lost and if you ask the ai can
you maximize my wealth can you make me rich very quick?
Then this is exactly what it chooses.
But it's something where if you're lucky, you get rich, but there is a good chance that
you lose everything.
And is that actually something that you want to choose?
And I think here also is the big challenge of the risk that you are taking, because currently
in the crypto space
everybody talks about yield where to get more yield what is kind of the great tvls that everybody's
having but there's barely a discussion about what risks are you jumping into like we have now so
many stable coins launching and some of them are actually really highly risky, like taking a secondary share of insurance risk, which means if the insurance case hits, you're the first one to get liquidated on all your assets.
And it's listed as a stable coin with 300 million TVL.
And in my opinion, this is a risk that you should not be willing to take without knowing it.
But many people do now in
crypto yeah absolutely it's gonna i think it's gonna be more difficult to have this kind of
framework as you were mentioning chris because there's a bunch of stable coins this is kind of
a commodity right now so in order for the retail people to understand their underlying assets and
the risk that you mentioned.
So yeah, I want to ask the audience, would you trust an AI agent to manage your DeFi portfolio? Please put us in the comments. And moving on, of course, as Dylan mentioned, where
GIL and AI meet, regulators won't be far behind. So I want to open the mic to all of you and
far behind. So I want to open the mic to all of you and ask if do you see AI driven execution
layers creating new compliance risks and how do we balance the black box nature of AI with
the transparency DeFi is built on. Maybe we can start on this one with you, Dylan.
So I guess the best way that I can answer
this is just speaking of again like at a high-level architectural point of view
from just like a case study that I recently done with a team called spout
finance so they were like oh you know pretty big winner with the chain link
hackathon recently and what they did in terms of achieving compliance is that they actually had to prove that
their architecture was secure. And because that's the first thing that a lot of these people will
be looking at is whether or not the guardrail or the AI pipeline is efficient, right? And so what
they did is that they said, hey, let's take this new contract called ERC-3643. And also let's take this new contract called ERC3643. And also let's take ERC720.
And let's actually mix and merge these.
But the nice thing about what these two types of contracts do on Ethereum
is that it allows for a coprocessor or like an off-chain server
in the sense this is the TEE.
What the off-chain coprocessor is going to do
is that it's going to take the TEE. What the off-chain coprocessor is going to do, it's going to take the event emission.
It's going to then substantiate the actual corporate bond.
The corporate bond in terms of settlement is usually quite complex because there's a lot of these back-end heavy, legal heavy commitments that both parties have to make in order for any kind of sufficient remittancy,
which is the transfer of some asset that comes with this behemoth of paperwork. Well, how do
you transfer it from party A to B? Well, you want to make sure that A, that both parties remain
private. That means when they actually conduct the trade, that it's a completely private transfer.
But we always knew that this was possible. We had Zash right we had monero right we had zero knowledge proofs
but when zero knowledge proofs cannot actually prove things like that are very expensive to
prove such as the behemoth amount of paperwork you have to also do that plus the asset transfer
all on chain but then how do you do it on chain without actually
revealing the underlining data and then while meeting compliance all at the same time and so
that's exactly what a programmable TEE will do like quite literally all you have to do is just
deploy the program to a docker file if anyone's familiar with this and then you can run quite
literally anything rust python typescript whatever it needs to be.
And so this program, since it's exposed, you can actually verify the git commit hash.
So that means, hey, I post some program that needs to do the remittency between two parties.
I also want to be like verifying that the contract has not actually changed.
So how do I do this?
So before I even take the action of providing a private transfer for remittancy between corporate A to corporate B bond, I want to make sure that it's
going to instantiate on chain. So on chain, it's going to conduct the transfer. Then the transfer
actually doesn't occur on chain. It actually happens in the co-processor where then the two
parties automatically do what's known as sort of like an atomic swap mine for yours and your for
mine and then at the same time the when the trade actually uh happens the commitment of that proof
the proof which is known as the remote attestation that ciphertext gets posted to the event loop
which is what the ethereum contract was awaiting for and then once the ciphertext actually gets
proven then people are doing things like fhe where they can decrypt the the ciphertext actually gets proven, then people are doing things like FHE, where they can decrypt the actual ciphertext to verify to the other prover, hey, this is in fact sent by Dylan.
Dylan was sending party A corporate bond to Sydney for corporate B.
And then the AI can actually ensure that the actual decryption occurs so that once it sees, you can almost treat the AI like an oracle.
And once the oracle actually picks up the event loop,
it can make the commitment for you using your private key.
That way you don't have to actually do anything,
but it's the thing that's going to ensure that the prompt
doesn't get injected with a separate script.
And basically that's why you have to train your model context protocol
for a specific RWA strategy.
And then you can start talking about your yield strategy. Because after
that, you want to first achieve private transfer, then decryption and commit the actual result to
party A to party B, and then make sure that it's signed by your AI. And then that's where you can
say, okay, now 10X the actual private transfer to 10,000 different commitments and throw it onto an L2 chain where I only spend
$1 but I've made a thousand to 10,000 different trades for a specific strategy that's trying to
consistently optimize to the regression curve which means like am I quantitatively easing or
tightening on my personal stablecoin account like am I buying more assets am I losing more assets
am I selling the assets like what am I doing and then that's where your AI really kind of takes control from there
oh my a lot of things to consider Dylan thank you for the whole overview detail explanation
Sydney I think I'm gonna go super high level I think you know you you asked the question
Aaron about regulation right and you know I think Chris talked about this a moment ago, right before Dylan,
which is just like, people forget to like, think about risk, like they forget to consider risk.
And so I think, what do you write? What do you regulators care about? Systematic risk.
Regulators care about systematic risk. My two cents is just like whatever AI application you're
building into, just make sure you have audit trails because regulators want audit trails to
understand how AI decision-making is handled. Not just the outcome, but how is it handled?
And you talked about self-driving cars earlier, right? At the end of the day, everyone's always
pointing fingers, you know, Tesla tesla car crash people die whose fault
is it is it the driver's fault is it the car's fault pointing fingers and then maybe different
states in the u.s maybe different countries you have different people that you blame um it's the
same with like d5 ai etc thank you sydney um, please, on the DeFi side,
something you want to add?
Like, I think another big issue for RWAs
is most of them are still KYC-gated,
and KYC is currently not interoperable.
Like, every week I get another protocol who writes me,
saying, hey, we are launching,
we're doing this and that in RWA.
All of them make the mistake, yes, you need to do KYC,
yes, it's proprietary to our platform,
so you can only trade our tokens on our platform.
And the core innovation of Web3 is interoperability, instant interoperability with all the protocols.
Nearly everything else is every revenue, every value that we created is derived from it, no matter if it's on the social networks or financial or yield.
And if you gate it with KYC and it's proprietary, then you miss out on the most important features
using DeFi protocols, using lending protocols, liquidity provisioning, token trading with
taxes.
That's where the magic is happening.
And so I think one of the biggest challenges we have in DEFA and RWA would be to make KYC finally interoperable through the whole ecosystems that the KYC providers have a standard, how to interchange and exchange the standards.
And you only have to do it once per wallet.
I would really love that.
So far, I do not see how we will get there.
Thank you, Chris. You wanted to say something dylan well um i i would say as far as kyc like there's
um there's a couple of protocols there's like zk pass for example you can prove your identity for
like your passport um you can link the actual ZK identity to the actual wallet itself. So therefore
you can prove like who the person or the entity is like, cause you're really realistically,
you're trying to create these protocols so you can reduce Sybil attacks. And so this is
so that bots don't actually are the only thing, bots basically meaning like AIs are not the only
thing that is behind the actual protocol. And so we kind of
meme about this. We call it the unruggable agent, where you kind of basically have a Rottweiler that
just kind of takes your money and goes, and then you have no idea of how to stop it. And then also
regulators want to make sure that there's actual human beings behind these assets, and it's not
just being traded by dark money traders so you prove that the asset
is controlled by an actual identity so you can use a lot of protocols there's zk pass
zk identity zk passport you can do a lot of these types of things um so they are available i just
wanted to make that mention chris because if you haven't seen them um identity is really hard to
crack down there's also um i believe uh plume Protocol is also working on this.
SUI also when it comes to ZK login.
So on-chain, you can actually have a smart wallet that is already attached to a particular login through OAuth.
And so OAuth already did the identification through the actual private and public key signature.
And then you want to make sure that both the ZK login and then the public and private key are contained in
a confidential environment so that no one can actually see that you're conducting the actual
trades themselves. You can make them exposed, but you want to make sure that no one reveals like,
hey, this private key or say this public key is literally attached to this particular login.
How do you do that? You do it inside of a TEE. You just make sure that you don't expose
that the actual identity was actually correlated to a public key, but you want a remote attestation
that proves the verifiability that, hey, this is not a bot, this is a human, but it's being
monitored and controlled by an AI agent that the human is actually authentically taking action on,
meaning that they're the ones actually pressing the yes button to trade. They're the ones who are actually signing the actual key. They're making
the actual decision. But the decision might have 25 decisions that come after it, but
at least you can verify the identity before the action even took place. This is known
as authentic actions.
Thank you, Dylan.
I think I agree that like civil attacks and so on, there are several solutions which are
now interoperable.
But for true KYC ownership that are like requirements for real-world assets, like for
real estate, it does not fit yet.
The key thing is like there are several ways of how RWAs are working.
One of the core one, which a lot of people trying to solve is like true ownership,
that you actually are proven that you are the owner of the real asset, which typically is
achieved via Swiss foundations. And here, the nice thing is you can even make the ownership
tax relevant because you can prove that you own a share of real estate or a gold or whatever.
But here you need to have true KYC because the custodian needs to prove to the government who is behind there.
And so it's not enough that it's proof that there is a human.
and so it's not enough that it's proof that there is a human. There is then secondary ownership
where basically you own a custodian or SPV and you prove the ownership in SPV which is much simpler
to handle and easier to interoperate because it needs to be proven that there is a human behind behind and in the case of sanctions, et cetera,
there needs to be an ability to limit ownership or movement,
which is the AML compliance.
And then there's the third level,
which basically is then the free moving token,
which is sort of that you get a voucher token to the ownership
and the voucher token doesn't need to have KYC at all.
And that's kind of the three levels that we have.
And the voucher tokens with Sybil, this is solved.
But for the other two levels, we still don't have an interoperable solution that scales.
Good takeaways, guys.
Kharch, something you want to add before moving on
I mean there's there's a lot of information that the panel had shared I mean if there's
just something that you know I just want to share I I mean, it's how I imagine the future of AI to be integrated with crypto.
Specifically, it's the dApps on the front that are going to be using lot of adoption in the
front because the end user is is the person who is going to be exposed to these wallets wallet
extensions so i just i just feel that this is uh this is going to be that but uh yeah uh more or More or less, it was great to understand how we can imagine a future with AI and understand how this automation can be done, how you can earn best yields while you sleep and while the agents do the work for you.
And I think it's a good financial future for us.
So, yeah.
Thank you, Harsh.
And yeah, I think the last question I have
in looking ahead is if the next couple of years
do you expect AI power LWA investing to emerge
first in wallet plugins, institutional robot advisors,
or curve chain yield aggregators and
if you have to describe the end game vision of ai and nwa gil in one sentence what would it be i
know it's a tough one but i know you guys can do it i'm open the mic for everyone who wants to start
one who wants to start first. Well, I could try to speak to this. The thing that Chris mentioned
first
before, which is like, how do you actually verify not just bought, but also the KYC direct
commitment. So once you buy these assets, you're effectively doing the KYC there with an institution.
Most of these institutions that are actually distributing any kind of yield,
they're going to do it in a centralized way. That's the whole purpose of them making sure
that the compliance is things like using a TEE and even using things like an AI.
But then once you actually purchase the asset, it becomes permissionless. You can't stop somebody
from actually utilizing a token on the blockchain. If you want to go to Uniswap and then trade gold
for stablecoin, well, you already bought the gold from an institution that already has the backed utilizing a token on the blockchain like if you want to go to uniswap and then trade gold for
stablecoin well like you already bought the gold from an institution that already has the backed
gold derivative or even like with an etf or any kind of rwa so i i think when it comes to like
institutions it's going to be them that are going to enforce this but then like what harsh said it's
really going to be up to the applications, the decentralized application developers across multiple chains to create those sort of bridges.
And then those bridges, they're already permissionless.
If they accept those types of tokens, well, then it's really up again to them to verify whether or not it's still the same identity that's basically cross-chaining.
Or you could even use ONFTs.
And so I know that protocols like zeta chain protocols like
lorenzo and a few others they've already done this and so they just want to make sure that you came
from a a specific institution you've already verified your identity they already have the
compliance rails set in stone now if you need to bridge these assets there's already protocols out
there that you can sort of do the bridging or the permissionless bridging while still verifying your identity and because at that point as long
as it's on chain you can trade these assets but it comes down to who actually uses these assets and
it's always going to be a commitment layer from layer two to layer one that's why it's always
going to be cheaper for the actual client and so once the client is able to sort of interoperate
between things like,
you know, optimism and Arbitrum, I think it's really happening mainly on Arbitrum at the moment
from what I see, then yes, like, then you can get into specific applications, like the ones that we
host, at least. I know that we just recently did a partnership with Gaia, with Fala. And so if you
want to make sure that you go into, say, like some sort of application that ensures that your yield strategy is not going to be exposed, but you want to kind of commit to a specific strategy with lots of other people, you can do so using a program that verifies using attestations that your keys and your trades are actually going to be completely confidential to you.
confidential to you. You can only achieve these sorts of things with off-chain co-processing.
That's always going to be managed by an institution. But the nice thing at least
is when it comes to TradeFi, DeFi, and these sorts of things, TEE is actually completely
in the clear right now. And it's a little bit in the Wild West moment right now. So I expect a lot
of builders to come figure out their ideas, create their strategies, create their
fine-tuned MCP services, list them on things like Gaia, and then start interoperating with these
agents using these bridging protocols to actually kind of manage these yields. So that's how I see
the future of this and where this is going. But there's really not that many protocols.
So the opportunity is actually quite broad, quite open.
So yeah, that's that's my take on that.
Thank you, Dylan.
Sydney, Chris, I'll open the mic for you guys.
Yeah, I think
for the upcoming long term, I think what's really, really fascinating
is that you're seeing network effects within the financial structures.
So network effects is now you're able to not only have DeFi and be able to programmatically
trade DeFi, but there's also the network effects of agents being indexed, agents launching
their own tokens, different businesses launching their own tokens, whatever the case may be. And I think really like the three pre probably like high level takeaways,
just to remember is that there's like fundamental shifts, which is like, one is we're looking at
products like RWA products, DeFi products. And it's not just about the product anymore. It's
about the outcome because every, everything has a stable token everything is an exchange everything you can trade but it doesn't
really matter because product becomes ubiquitous so people people returns enterprises institutions
culturally people care about outcomes um number two as i kind of began this whole conversation with
is like the physical human traders on the stock exchange. Now we're moving from like manual to automation
or the autonomous or workflows of trading,
all of these different types of assets.
And then really lastly,
I think a lot of us are in crypto
is this like shift from like traditional finance,
which is like centralized
into this like self-sovereign financial ecosystem,
either self-sovereign within yourself and your own trading portfolio self-sovereign and the agents
um owned by the community so on and so forth
thank you sydney i think you want to go next chris sure i think it's it's a very good summary, especially for the community that is here.
If you look into the future, you can be either three types.
You can be a quant builder who focuses on a very specific strategy,
probably uses some TE and some proprietary data to optimize something very specific.
data to optimize something very specific that could be Delta neutral on EVE or even Gamma
Delta neutral on Hyperscale and others and create a single strategy for a very specific
case where you generate yield.
The second category is you can be a curator using several of this specialized token and specialized cases,
switching between them.
So the low-level trading will be done by specialized quants.
The high-level curation will be done often by communities, DAOs,
or a group of specialized traders who create these curated tokens and strategies.
And then there's the consumers who basically out of 80% will choose curators to hold their assets
and 20% go into high conviction tokens or strategies that they believe that they will do 10x.
And I strongly believe that this is the future
that we're looking now on Chinville.
Thank you, Chris.
Parish, please go ahead.
I'm good, man.
I think exactly this is what I shared earlier about the wallet integrations, right? How we see it. big gap, a big, uh, a big, big gap with, you know, a lot of this traditional finance yield
coming on chain. This is still a big gap. I just, I just read a tweet that, you know,
uh, in 2020, the DeFi, uh, proof that money can move without banks. And then 2025, I think, uh,
without banks and then 2025 i think uh uh defy will prove that traditional yield can can move
uh traditional markets can move without banks too so i i think the the the this is this is
what where we are heading towards right uh making this kind of yield accessible to a lot of people
which is exactly what r2 is trying to do so
r2 is one of the one of uh one of its kind and i believe a lot of other companies are going to
emerge uh we are going to see a lot of development here uh we're going to see a lot of capital
deployment here uh but yeah uh i mean this is there's nothing but upside here.
And absolutely, it was a great session to be learning
from the best in this panel.
But yeah, those are just my conclusive two cents.
Yeah.
Thank you, Harsh.
Absolutely.
Actually, I learned a little about Sidney, Chris, and Dylan, more things to evaluate and consider.
But yeah, I think moving on, I want to thank all our guests, Dylan, Chris, Sydney, and Harsh for an incredible discussion.
And thanks to everyone who joined us.
Please follow our speakers, all our building pretty cool projects, as you all heard.
Also drop your favorite
takeaways in the thread and stay tuned for the recap I want to give also the opportunity to our
guests to give some one last words if you want to say something guys um yes so please for everybody
who you know wants to not just like support startups um you know this is
I think this is the biggest opportunity in DeFi um since kind of NFTs in my opinion I think the
the real world asset game is kind of like the hottest thing right now in crypto at least um
it's just like with TEEs it's a little bit underground and just like with ZKPs TEEs are
a little bit underground so there's always this like transgression between technology to technology over the years.
And so for the way that I see the future, you're going to see a lot of these protocols
utilizing these types of decentralized frameworks.
So if you guys are interested to know how to build these systems, you want to tell people
that are in your community, hey, like I have this really good sort of idea.
Maybe we can improve our systems and
maybe you're bullish on a particular project that is in some you know in some protocol uh definitely
check out dstack forward slash te on github star it do whatever you got to do share it uh we're
trying to make it the number one te framework in the world um it was already audited by ZK security. So it's already hands down 100% sovereign, 100% zero trust. And we've also decentralized the key management system. And this has not ever been done. It was the first ever audit that actually did a full review of a design for the operating system level security of a TEE framework. And it passed with flying colors. And that was actually years
after the system had already been developed. It's already basically been accepted by the
Linux foundation. We got yeses from meta, we got yeses from TikTok. And so when they want to start
doing real world assets, they want to start putting like, you know, I don't know, not even
like, you know, gold, but like literally videos and content and intellectual property.
These are going to be coming next, not just like cars and things like this, like typical
things, but actual IP ownership.
That's going to be coming next.
But how do you do this?
Where do you go?
Who do you trust?
Please take it, you know, take a time out of your day just to kind of look over and
review some of the documentation that we have on falla.com.
I deeply encourage it.
Especially if you're interested in the combination of making compliant AI.
We have a plethora of tools and systems and services and templates.
If you can go to awesome templates on our website, we're partnering with a lot of different DeFi protocols and just protocols in general for storage, for DeFi, for agents that do this kind of stuff on a daily basis.
You can even check out some examples like Blormy Finance, which is hosted on FALA,
Rabi Finance, which is hosted on FALA. And these today are production-ready applications that are
already doing agent-to-aided trading using DeFi tokens, but the next stage is going to be real
world assets. We see all of them asking for
it now it just comes down to the builders and you guys to figure out yeah what are these compliance
what are the guard rails what how do we do actual appropriate risk management and um yeah bring your
ideas uh to the table we're happy to have a conversation you can always hit me up and if
you're curious about actually learning about real world assets from like a first person point of view with real builders you can go and check out the black box podcast at blackboxpodcast.com
that's actually we did a whole interview with gaia already and uh the second episode was already with
spout finance so you can actually get an in-depth look at the architecture to understand yourself
like where this industry is going so everyone thank you so much for your time and having me up on here it was a lot of fun talking thank you dylan so much
for you guys free sydney thanks for the discussion i think it was very interesting was a lot of
knowledge and alpha that was dropped today um and i'm looking forward for the next one thank you chris
if nothing more to say before we wrap up i want to give a quick update everyone's been asking for
from our community as you know our r2 mainnet is coming this sept. So after 680k test net users and real feedback coming
from the community, now it's time for the real thing.
So we're ready to launch the first on-chain fund funds.
R2 Mainnet is about unlocking institutional yield
for everyone in a permissionless and borderless way.
So whether you want to start with $100 or scale bigger,
all you need is a wallet.
Also to mention that Desnet users with points
will be eligible for additional rewards
when engaging with the mainnet.
So we want to make sure to recognize and incentivize
early supporters and contributors.
Also, we'll be posting an article and more details
soon across our channels
so you can get more about the details of the mainnet.
But yeah, most importantly,
thank you to everyone who's been with us
these past months.
Your supporting feedback got us here
and mainnet is something we built together.
So if you want to be a part of the future of Yield,
this is the moment to join.
And yeah, thank you again to our amazing panelists
and everyone who tuned in.
Let's keep building together
and see you in our next conversation.
Thank you, guys.
Thanks, everybody.
Take care.
Thanks, everyone.
See you.
