Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
Web3 Global Talks 🎙️ Ep.444 - Powered by VAIOT
Recorded: Aug. 13, 2025 Duration: 1:22:01Space Recording
Short Summary
In a dynamic discussion on Web3 Global Talks, industry leaders unveiled groundbreaking projects and partnerships, including Twin Protocol's AI twins, BioCrypt's anti-rug pull technology, and Radiant Capital's innovative lending protocol. The conversation also explored the transformative impact of AI on the legal industry, emphasizing the need for updated legislation to protect intellectual property in an evolving digital landscape.
Full Transcription
Oh All right. Thank you. so Oh Oh, yeah. all right guys welcome into web3 global talks episode 444 powered by aviat uh we are just
trying to get everybody up here of course still so i'm gonna run the song back here for a second
i had some drops and connection issues uh in the other space i just did about
20 minutes ago so hopefully everything's good uh but i'm gonna run this song back here for
a little bit while we get everybody up here and we'll get started shortly everybody I
I Hi. Thank you. Oh, yeah. so Thank you. all right guys i think we can go ahead and get started. BioCrypt, if you guys could see if you can accept the invite to speak, that would be good.
I know you may be having some connection issues.
But yeah, guys, without further ado, I think we can go ahead and get started.
Welcome to episode 444 of Web3 Global Talks.
If you haven't been here before, I like to start off the space by giving about a minute or two a piece to everybody
to kind of, you know, introduce themselves and who they're representing. Then we'll jump into some questions from there. So I'm
going to go top to bottom on my screen for that. So Twin Protocol, you guys are up to start us off.
Awesome. Hi, everyone. Can you hear me okay? Yeah. Okay. Awesome. Stacey Engel here. I'm the co-founder and CEO of Twin Protocol. We are a blockchain and AI convergence project. So what we're about is storing your personalized, decentralized training data. of data sources and we bring them to life in lots of different ways through avatars, AI twins,
holograms, but it can also be buildings
and other kinds of data sets.
And thanks for having me.
I'm excited for this conversation.
Yeah, thanks for being here, Stacy.
All right, let's go with BioCrypt.
Hey guys, thank you very much for having me.
My name is Adnan Alicic and I'm on BioCrypt.
And we are both patented micro-tract trading platform.
And in the next couple, what we are doing is we'll be launching our pro-trading platform where we are targeting creators, especially in the
ecosystem.
We've been in the industry for about three and a half years as a project and me myself
was in crypto industry since 2015.
So I've seen a lot of ups and downs and I've seen a lot of inefficiencies.
And one of the biggest inefficiencies, especially in the Solana ecosystem, is actually the prevalence
of rock pools.
And that is basically launching this project to actually help all the traders avoid all
those rock pools during their trading.
And thank you very much for
having me here yeah absolutely thank you for being here man all right coral protocol
hey um yeah thanks for having me um i'm roman the co-founder slash ceo of coral protocol um
before this i was on the founding team of Kamui.
They were the first multi-agent system framework.
And then in that time there,
we found out there were a lot of problems
with people building agentic software.
So what we're building is a way to orchestrate
the internet of agents, we call it.
So it's some way where any person building any agent,
many background vendor framework, et cetera,
can be orchestrated from one place.
Nice, thank you man.
Okay, let's go with Radiant Capital.
Thanks much.
I'm happy to be here with everyone.
My name is Hong Vu.
I'm a DAO administrator and community council member at Radiant Capital, where it is governed by truly a community council.
the council right so what we are is a lending protocol that was built on ave
and it was the first move on the omni chain lending sector um launched in 2022
um you know designed to solve uh you know liquidity fragmentation and such.
Instead of being siloed, you can borrow money in one chain, deposit in another, and vice
versa.
What makes us unique is that first and foremost, again, it is a community-run protocol.
It didn't start out that way, but it is now. It has been for close to a year now, maybe three quarters.
And, you know, it's got kind of an AVA plus type of roadmap, right?
It has real yields.
It has, you know, liquidity, swapping liquidity.
That's got a nice locking feature.
But also, we'd love the opportunity to talk a little bit about what we're about to launch in short term here,
where we will also bring another first into DeFi, and Web3 for that matter, what we call a Radiant Guardian
Fund, which is a community funded safety reserve, as you will, that earns yield but also will
automatically and essentially reimburse exploited deposit as well. It's interesting concept. Thanks again for being here.
Yeah, thank you, man. All right, let's go with Matrix.
Hi, everyone. This is Eric from Matrix AI Network.
from matrix AI network. Matrix AI network was found in Hong Kong in 2017, which is quite
early in this space. And over the past years, we have been through three stages. So number one is
the matrix 1.0 that we use AI to optimize our public blockchain.
And then Matrix 2.0 is to construct a blockchain-based AI economy based on the triangle of the AI data, computing power, and models.
And now we are at the Matrix 3.0,
where we further brand neuroscience with AI and blockchain.
And just recently, we published a direction that is to leverage the wave in Hong Kong
about the stable coin and real world asset, which is quite exciting.
We also make use of our accumulated technology to build the kinds of marketplace to enable and to empower all
stakeholders in stablecoin and RWA sectors from Hong Kong. So yeah, thanks for having me. That
is a brief introduction about Matrix and myself, and I'm the CMO for Matrix, by the way. Yeah,
thank you. Thank you, man.
All right.
And Mariella, I saw you fighting for your life with the connection,
so hopefully you're good.
Me too.
I hope so.
Yeah, today Twitter is not, well, X, sorry, I keep calling it Twitter.
It's not liking me.
But, yeah, thanks for having me.
I'm the business development lead at IoTeX.
At IoTeX, what we do is connect smart devices to the Web3.
So basically deep end, but also AI is connect smart devices to the Web3. So basically, DeepIn,
but also AI, there's a lot of implications here. But we basically build all the technology,
so that all the data that's coming from any device, it could be your phone, it could be your car, it could be any smart device, can be put on Web3 more securely to protect your data. But also,
in order for other people to use your data safely for businesses to grow
in scale and to build infrastructure to help us grow AI also and for AI to also get more
data to be smarter.
So yeah, thanks for having me here.
Thanks for being here.
All right.
And of course, Vaya.
If you're speaking, man man i cannot hear you i don't know if anybody else can uh it could be
my connection for some reason here but um we'll give you a couple seconds man let me see if we
can uh get you fixed up here uh you want to try again?
Yeah, no sound for me either.
No sound for you either.
Okay, great.
We wouldn't be in X-Space without some random problem.
Okay, well, we will definitely get him back up here.
He's going to reconnect here for a second, guys.
Sorry, just bear with me.
You guys know if you've ever been on spaces, how they go sometimes.
So it is what it is, I suppose.
All right.
Let's see.
Okay, got his request.
Okay.
Let's try this again.
Can you hear us, man?
Oh, did he drop again?
Good Lord.
Okay.
It's quite a ding. Yeah. i well i thought i heard someone for a
second there yeah can you hear me now yeah yeah yeah you're good man okay great uh okay guys uh
this is simon here i am the tech lead at viat uh maybe let me introduce you to Vyat a little bit. Vyat is a regulated legal tech company.
Somebody mentioned that 2017 is early.
Well, we essentially founded the same year.
Since then, we have been trying to democratize legal services
through AI and blockchain technology.
We're trying to solve a fundamental problem,
which is that legal help
is extremely expensive and slow for regular people and small businesses. To tackle these
problems, we have thus far built two core products. One of them is LegalTorch, which
is an AI powered platform that handles contract creation, negotiation and management alongside AI intelligent assistance
while creating your contracts or discussing with the second party. Our second product, the DRS,
tries to revolutionize dispute resolution by using blockchain-based arbitration instead of
traditional courts. Hopefully, if somebody tested it out in the past few days, you probably saw that
it's a lot faster, cheaper and less painful than traditional courts or hiring lawyers.
Our mission is pretty simple, make legal services accessible to everyone, not just wealthy corporations.
operations. And again, thank you for having me.
And again, thank you for having me.
Yeah, thank you for being here, man. All right. Well, we are going to speak to a little bit about that with our questions today.
So our first question for everybody is, how do you see AI transforming different and lesser spoken of industries such as the legal industry?
And what are the most promising applications that you're seeing right now?
And if you guys haven't been here before,
the way I like to do it is a hand raise.
So the heart at the bottom and the hand all the way to the right
is kind of how I pick the order.
If you want to jump back in after you've gone off of somebody else's point,
you can definitely do so.
Just raise your hand again.
But I will see who wants to go first.
Yeah, Coral, looks like you guys are up.
Yeah, yeah.
Just right off the bat, I'm coral looks like you guys are up yeah yeah it just um right off
the bat i'm not sure if you guys saw it there was a um there was a company that tried to have a ai
represent someone um in a law case and the judge um i think they just didn't allow it because they
sort of knew it was for like a promo thing um so i i guess a lot of it really comes down to uh legislation you know
like like how much are we gonna allow ai to be used in the legal process but i i think outside
of that i think it definitely democratizes in some sense um like legal stuff i i saw a guy
i build a tool that creates like a legal cease and desist, stuff like that.
So I think it's going to make it so, okay,
everyone kind of has a better understanding of the law
and able to invoke it better.
Yeah, fair enough.
Let's see, Vyat, did you want to go off of that?
Yeah, what you said is 100 true so we advise we are tackling this specific
problem of legislation so essentially going country by country and trying to essentially
scrape all of the legal data in live format so that our AI assistants can essentially give you a valid answer every
and every single time you ask it a question right so if you've ever asked either chat gpt
cloud or whatever other ai chatbot any legal question trust me that most of these the answers
that you're gonna get they're not gonna be up to date. So getting access to this live set of data
about different countries,
about the laws that are currently there,
it is a big problem.
And obviously to get access to that data,
you have to be in talks with the government.
So as long as countries allow us,
this can go much further.
For example, for now, we managed to completely scrape and have access to laws from Poland and from Malta.
And we're trying to introduce another country, which I'm not going to leak here, but we are working on it.
So as long as people cooperate, I think we can go a long way.
Not only in the legal industry, but in general, the lesser spoken industries.
Thank you, man. Stacey, did you want to jump in?
For sure. Thanks. So at Twin Protocol, we are, and right now anyone can get an AI Twin for as little as $25 and start their vault.
AI twin for as little as $25 and start their vault. So for us, it's all about, and it relates
to the legal field, digital identity and rights management for some of your IPs. So we have
clients who are celebrities who we've worked with and work with individuals like Dr. Deepak Chopra
who have 90 books. So when you think about a vault that acts as a verified source of truth
for individuals, AI twins, that's really compelling, not just from a secure storage of
those assets, but also a secure storage of legal agreements and rights to IP. So I think that's what's really exciting, particularly about
the convergence with AI and this blockchain storage, because essentially you can use smart
contracts and other tools to license. So I think when you think about legal industry,
So I think when you think about legal industry, financial industries, many that oftentimes it's been a reactive type of position with even agreements or compliance.
AI is really going to allow constant monitoring as well as being proactive.
Sometimes people don't even know
that they're copyright infringing on others.
And I think the future of that's going to be very different
when we think about the application of AI.
Yeah, I mean, there's also an argument to be made
that, I mean, doesn't AI itself
kind of copyright infringe a lot of the time?
I mean, like, as far as like, you know, what's the large language models right now and where they're getting their information from?
It's a pretty good question.
But, yeah, thank you, Stacey.
Mariella.
Yeah, for sure.
I just react to that one second.
That's why you need to start building your own vault with your IP.
Because it won't be, there will be more and more models that they're going to be asking you for your rights to your data.
And I think we will see enough people about that more and more.
It's not just, well, this is how it is.
At least I hope not.
And for all of you listeners and everyone in the space,
it's not good enough.
And we should be able to monetize our data once the technology exists versus these big tech companies basically using our stuff and not paying us.
Yeah, absolutely. Well said.
Mariela, you're up.
Yeah, for us, you know, since we're not only well connecting machines, but also tokenizing
these machines, there's a lot of legal implications.
For example, working at PLC with Vodafone, and we're going to deploy various antennas
for Vodafone.
Where these antennas are, how they act, how they share the data, etc.
All of this has a legal implication.
And unfortunately, the laws continue
to change very quickly. And so our vision for the future is that the machines are, they have their
own identity, and they can also work out all these legal processes where it's being permission to,
you know, deploy some data or being permission to do some actions, this assault has legal repercussions.
So for us, just legal AI agents that can easily talk to the machines are a really nice tool.
And I think we're already kind of seeing some applications of this.
And I'm very excited because this means that machines will run more efficiently and we will not run into regulatory hassles as we are right now.
Yeah, absolutely. Alex, I added you because you probably saw the question in your mind
as being perfect for you. You want to go ahead and jump in?
Yeah. So I'm not building in this space. I'm just a lawyer who works all the time in this space.
And I can offer the state of the law and some issues that I think a lot of people aren't necessarily as aware of as they should be or could be in working in here.
working in here, it's not really just the legislation. And remember that you have to
keep the trial issues away from the build issues. Those are very separate. First, the issues in
terms of like mediation and arbitration, we've had several different types of builds that have
tried to address that. There's a huge number of legal and practical issues. I would
just encourage people to go ahead and look at those prior builds and how they didn't work
in order to make sure that your build covers those issues so that you have addressed why
they didn't work in order to make sure that yours is better. In terms of scraping and things like that, remember that the
state of the law right now protects the people who are scraping, not the people who hold the IP.
And that is a really important thing to know, right? You're not actually going to do yourself any good by putting it in some other
location necessarily. The truth is putting it out there is what allows you to have money.
The people that you end up attacking right now are the people who prompt, right? And that's the biggest problem
because the prompters aren't the deep pockets.
This is the state of the law.
So here's where legislation
can actually make an enormous difference.
This actually goes back to Google, right?
Google Books is the one who set this precedent.
And this is where the biggest difference can be made.
And this is where the biggest difference can be made. And this is where the exploit happened with open AI.
So I would just encourage people who are able to make a difference in legislation to really pursue this part,
because this is where most of the protection can happen.
And this is where people whose IP is getting exploited, if you want protection,
what you need to do is be able to go after the people who are illegally scraping and taking the
metadata off and taking off those copyright protected or going into vaults or things like
that. Because right now, none of that is prohibited. So taking licensed information and not paying for the license, etc., that is a hugely problematic
issue.
The other part of this is really where I think a lot of growth can be had is the off-chain
chain protections, right? We can do a lot. And I think a lot of growth has happened in people
protections, right?
We can do a lot.
using these AI tools in order to look at where we can protect things on chain. But the real risk
tends to happen off chain, right? And this is where I feel like a lot of people are kind of
not connecting the on chain and and off-chain worlds.
We can use things, for example, in the music industry, these PROs.
They can actually do a lot of the enforcement with finding off-chain use, which is what they do right now, and issuing a lot of the procedural things.
a lot of the procedural things. That's one of the things that when you're talking about what could
AI do or what could some of these organizations that are non-legal organizations, what could they
do? They can do things like issue procedural documents, but they can't really do things like
enforce the law or issue legal documents, right? So you could have them essentially limited to
certain types of documentation that start a process and stop
right there, right? And then it moves to lawyers that are super expensive, etc. But trying to
combine these two, I think, is really going to make a huge difference so that the blockchain
is something that allows you to really easily see where people are trying to use your IP or whatever
asset you have illegally. But I don't think that's where the worst uses are going to be.
What we've seen are a lot of exploits where someone is taking something and using it offline
or buying something like what we saw a lot of exploits were, where people would
buy something from an artist in Africa or something like that, and then make it into an NFT separately.
And that's the kind of on and off chain mix that you want to be able to say, well, I can prove that
you don't have the right. And I want to be able to protect the right by using this combination of on and off
chain protection. So I'd love to see that kind of a combination and using AI agents plus blockchain,
I think, would do a huge amount in that. Yeah, absolutely. Welcome knowledge, Alex. I encourage
you guys to reach out to her as well. If you have any questions, she's amazing. So thank you for joining, Alex.
I appreciate it.
Yeah, I mean, we're kind of running the gambit of issues here,
including AI representing people in legal matters,
which is going to be an insane thought, but definitely coming.
Eric, what are your thoughts, man?
Yeah, there's a lot of wonderful ideas in here
that I just shared regarding the legal perspective,
especially from Alex,
speaking from very solid and practical ways.
And you just mentioned about kinds of cloning the people
and that's actually how AI has been used and is good.
For example, you can clone my voice from this Twitter space
and you can even clone my appearance
and they can come up with some sort of,
if you do it in the evil world,
it would be a scam and fraud
and it can be crime in the real world.
So my perspective to supplement from other speakers is trying to look at from a KYC
angle, loyal customer angle, because only when the exchanges or the government authorities
are satisfied or happy with the level of the KYC, they can have more kinds of a loser regulation
in terms of adopting the cryptocurrency or blockchain space.
And I'm looking from the recent cases of stablecoin in Hong Kong
because right after 1st of August,
it gets more strict than we thought it to be
because they have a strict KYC and they have a strict regulation.
When you want to use stablecoin in Hong Kong, you have to stay physically in Hong Kong.
And the wallace, it cannot be any kind of decentralized wallace inside.
cannot be any kinds of decentralized wallet inside.
And that's why we have so-called bio-wallet,
or we use the finger-wind technology.
And let me give you an example.
Finger-wind is different from fingerprint.
And you come across another project called WorldCoin,
and they make use of the iris.
And for our case
We use the finger wine and what we'd like to do is trying to prove that people is
Real people instead of you can purchase ID from elsewhere and this weak finger wine is physically you can
You can track down to one specific person, and that person is actually using it
just like the iris from Rokorn.
And for the far false acceptance rate,
finger wine and iris are basically the same,
which is less than one out of one million
false acceptance rate.
And in terms of false rejection,
finger wine is better than I was.
We are talking about 0.01% where I was is 0.2%.
So we are trying to build this bar wallet
and using the finger wine technology to to kind of verify that person on chain is really
someone in the real world and he's actually doing it instead of someone purchasing uh rs id
from as well or cloning the voice appearance and doing something evil then so there's something
that i want to supplement from other speakers from
a legal perspective, i.e. from KYC angle and particularly from the Hong Kong perspective.
Thank you. Yeah, very interesting take, man. You know, digital identity is going to be huge going
forward as well. Also very topical nowadays with all the stuff that's going on in like the UK,
you know, with all these age restriction verifications and what I would consider like the bad faith argument of
protecting children. You know, we're getting a little dystopian here. You know, that's why I
think we had a space yesterday about zero knowledge proofs and how people don't really need to
understand exactly how they work. They just need to know they're important.
And I think it's going to be even more important going forward.
I know that's a little bit off topic, but, you know,
I think that's definitely a huge issue we're facing right now.
But thank you, Eric.
BioCrypt, you're up.
Yes, I would agree with most of the things that previous speakers said.
I would agree with most of the things that previous speaker said.
And I would add that while AI's role in finance or e-commerce is widely known,
in the legal industry, it's actually often underappreciated.
And one of the reasons for that is probably because legal industry has been very slow to adapt to technology.
But you know, that is that is going to be changing very fast.
And as one of the examples that
the gentleman from Coral protocol mentioned
with AI representing a client,
I personally believe you're still far away from that.
First from the compliance issue, because as most of the people know over here,
the legal industry is vitally and pretty decently represented.
So any changes that's going to be threatening their profession,
they're going to be actually fighting that.
Now, but how AI is going to actually help them is actually not with actually direct legal representations, but for example, such tools where actually AI could actually review large amount of contracts, let's say flag any types of anomalities, suggest
clauses, or even draft standardized agreements, which is actually the work of junior attorneys
or paralegals. So, for example, all that could be done actually by, you know, by an AI.
And some of the other things that actually gonna, AI is gonna help in the, in terms of law is AI could actually analyze case law, even judges rulings and legal precedents and actually predict case outcomes and actually guide litigation strategy.
So basically they could be used as a tool, but not actually as,
you know, a direct way to actually represent a client.
And basically, actually some of the startups are already building
win probability engines for attorneys and clients basically giving them, you know, the
probabilities so they could actually make rationalized decision in terms of the
cases. Now, from the legal, from the business perspective, let's say if I'm
a lawyer that's having actually,
let's say couple of paralegals
and let's say couple of junior attorneys working with me.
I mean, it would be my best business decision
to actually use AI.
And I mean, it's all about business
and eliminate all those people.
So, so instead of having, let's say, by paralegals,
I could only use one because the AI realistically could actually replace all
that. So, so basically the,
the best business sense would be actually for to have, you know,
AI do basically everything in terms of research,
in terms of analyzing case laws, in terms of judges rulings, in terms of, you know,
researching legal precedents. And basically, even, you know, they could actually draft
standardized agreements, you know, and just have some of the paralegals basically double check it, make sure everything is good and go from there. So the AI is going to contribute, you know, it's already contributing in many of the industries.
And even though the legal industry is lagging,
it's only a matter of time before it gets there.
Yeah, I mean, I suppose that does make a lot of business sense.
Whether it makes moral sense, I suppose,
would be a completely different space
and a different question, I guess.
I'm sorry, I'm coming hot off the heels
of a more philosophical AI space I just did
like an hour ago. And one of the questions was like, you know, are we creating AI to serve
humanity or to replace us? And, you know, it's kind of one of the things that I tend to worry
about a lot. But, you know, speaking strictly of the legal side, yeah, you know, having AI help on
the back end makes a lot more practical sense than having AI represent you. And I would say that we are extremely far out from that.
And for good reason, when Coral was giving there that situation or whatever,
I think that that judge definitely made the right decision there to not allow that to continue.
But Coral, speaking of, you wanted to jump back in. I saw you first.
Yeah, maybe I'll be a bit...
Yeah, yeah. I think... I'm'm just curious it's like an open question so is the reason why people generally think we can't have ais represent
us because it's it's replacing someone just just just because i'm speaking from the other side of
things where we talk about doing the research work doing the paralegal stuff that on the back end you're still replacing people and uh if it's like a
can it do it well um question i'd say yeah i'd say most things we can benchmark i guess we can do
pretty well and there's some really good conversational agents coming out so yeah i'm
sure curious what people's thoughts are.
I mean, at least for me personally,
it's not the matter of the fact
that it would replace a lawyer.
That doesn't necessarily register my mind.
I think that it's more of a psychological thing, right?
To be represented by AI,
there's a barrier there that I think is just,
I don't know, intrinsically foreign to a lot of people.
You know, I think that would be like the barrier.
My barrier when I was speaking a second ago is more from, I don't think we're technically there yet.
You know, if we technically get there and the ability, you know, makes sense, then obviously, you know,
yeah, I don't know that I personally would have, you know, too much of a qualm with it.
But I would venture to say quite a few people would, but that's just my opinion. Alex, here.
Okay, there is a big issue. So I often say that the problem with law is that it's at least, okay,
in the US, and I'm talking about the US, although I have studied other systems and, you know, there are different kinds of systems, ones that are more statutory versus what we have, which is the stare decisis system, which is like cases are decided alike.
Common law is different from a statutory system.
We kind of have a mix here in the U.S., but there's kind of a problem with law in that it is in English, right?
with law in that it is in English, right? So there's a lot of people who think that they
understand law because they understand the language that it's written in and thinking
that the language means that they understand enough about how it works simply because they
understand the words strung together. But it's actually so much more complex than than that it's
like people thinking that they understand securities law because they've
heard of the Howey test and therefore they understand all of securities law
but the Howey test is like one part of one part of one of one test what it's
one one list and there are several lists in securities law, which is a huge field.
And would there really be a whole industry of securities law if it were all about one test?
Like we have to think about this more globally, right? Law isn't really that much cookie cutter.
isn't really that much cookie cutter. The people who think it's cookie cutter are people who aren't
lawyers and bad lawyers, right? Most of law is really about nuance. And the problem with AI is
AI has no understanding of nuance because it doesn't think. Like when we're talking about non-cognitive AI,
which is what LLMs are. LLMs are very, they're barely AI, right? They're really more like
really sophisticated search engines. And when we're talking about these things, they don't
think what they do is collect data and it's new to you, but it's not really new information.
data and it's new to you, but it's not really new information. And they're not really nuanced
engines. So one of the issues that people tend to have is, you know, is AI replacing us. Well,
AI, the way that we have, right, LLMs don't replace us. LLMs are essentially tools for experts.
And what they do is make the life of an expert much, much, much easier. Like,
for example, if you think about Cursor. They even did studies on Cursor, which Cursor is basically
like a debugging tool. Cursor takes a lot of time to use. So when they originally put out Cursor,
people found that it was actually adding a lot of time to developers.
But then there's like this peak and drop off, a serious drop off.
And what happens is eventually the experts learn exactly what Cursor needs.
they're the ones who understand not just the language, but also how to debug.
They're basically catching all the errors of cursor and they're giving it exactly what it needs.
And that's really how LLMs work is LLMs, like all the legal products that I've seen,
and I see a lot of them, I get submissions all the time.
They're full of bugs and that's okay, right?
But lawyers who really know what
they're doing catch those bugs and basically it's meant to just save time for lawyers, right? It's
not meant for the laity or people who are non-lawyers to do legal work and it never should
be. That's why coming up with a legal document that you found on the internet is a terrible idea
because not only do you not understand what that document means, you don't understand what the universe of clauses that you could have chosen are.
You don't understand why those clauses were chosen and the impact of those clauses, because each clause has been litigated in a particular jurisdiction and has a particular impact, right?
Some of them have impact that isn't yet clear and has risk attached, and that may be reflected in
the price of the deal that you chose that from. And you don't know that because that's not something
that you do regularly, right? Unless you're an expert in that field.
So it may be you are a business person that always deals with those kinds of contracts.
You may be familiar with I need these clauses and not those clauses, but you still won't
necessarily know what the universe of clauses that you can choose from are.
You just know what you're familiar with.
That's not the same thing. You also won't know
how to look up all of the current issues that are available because the law is changing all the time.
So here's the risk in some in professional fields. The liability still lies with you,
right? So you're still responsible for anything that happens with whatever it is that you do with this AI, even if you didn't know that that was the implication of whatever it was that you were doing.
So this is really the problem with doing a lot of this AI with non-lawyers, right?
There are some things that you can do with non-lawyers, but most things you kind of have to do it with a lawyer
because the lawyer understands the implication and more importantly has the insurance to take
on the liability if something goes wrong. So we have these kind of issues that are layered on top
that a lot of people may not want to spend money for the lawyer, but we have to actually think
about where's the liability? Who can afford to
take on that liability if something goes wrong? How much risk is the person willing to take on?
Do they understand the risk that they're about to take on? And can they afford to take on that risk
or is, I mean, do they even understand where the risk is? So there's like a lot of different issues here that AI just can't really support.
And trying to sue the AI developer for the misappropriation of risk right now, the law isn't in your favor.
So this puts huge amounts of risk on someone who's just pulling up a prompt.
Yeah, very eloquently put.
You know, I think that's better than I could have definitely said.
So, yeah.
Thank you, Alex.
Sorry, I can't speak today, guys.
Stacey, I saw your hand for like five seconds.
I didn't know if you wanted to talk, but you were before these other guys.
You were before these other guys.
Well, I love what Alex was saying and just building off of that and the other question that was asked just about replacement.
I think it just poses the question of in every industry, there's different level of augmentation of roles versus replacement.
And in my previous companies I've built,
they've always been in the leadership training development space.
And I just believe that some of the grunt work
or whatever type of association people,
I mean, when you interview workforces,
they're very dissatisfied with a lot of their work.
So I was going to add that point regarding the creativity.
Like I love what Alex said.
There are certain components.
There's liability.
There are reasons why AI should not be replacing certain elements of roles or processes or frameworks. But then
there's, I think, a myriad of components and roles that should be taken off people's plates
so they can do the more creative thinking, they can do relationship building, and also particularly in North America and
in the U.S., people are completely burned out, have very low job satisfaction. And that's always
what I bring to that conversation around because I've coached C-level executives and middle
managers, and the reality is that work needs to be transformed.
And for some reason, that part of the conversation always gets left out when we're talking about
the replacement of roles.
And that makes sense because it's the economic model we have.
But I think that's really when you think about the convergence with blockchain and AI,
this idea of being able to monetize your knowledge and have certain things like captured in an AI
twin for you that you could use in multiple ways and the economic model changes. So kind of
bridging those topics because I loved what Alex was sharing. But I also
think the previous question around replacement augmentation is compelling. Yeah, I mean, I think
it's somewhat of a bit of a tangential argument, obviously. But you know, I think it does play into
a lot of this. I mean, there's a lot of like, you know, quote, unquote, bullshit jobs out there,
right? Pardon my French, but I think that's literally the term that a lot of people use.
The reason I think it's not in a lot of the conversation as well is just there's an existential threat to that.
If you get rid of all these entry-level positions and different things, how do people launch careers?
How do people make money to live?
There's a lot to consider here.
Obviously, I agree with you guys.
I think that is like the more natural use of AI.
Anybody who's an expert in anything that uses ChatGPT can definitely see when it's wrong, right?
I think we've all seen it when we're using it to help us with things that we know about.
And you're like, hey, no, that's absolutely incorrect.
But think how many people,
and Alex gave a great example of this, in the legal world, we'll just take that verbatim as the truth and then run with it. It's a dangerous precedent in the space and especially in the legal
space. So thank you. All right. I believe it was Coral next and then BioCrypt Yeah, this was a little bit far back
but I'm just curious because I'd like
to sort of probe
a little bit more
on the argument before
that AIs don't have
nuance and I don't think
if anyone remembers
the last space, I think we spoke a little bit
about, I definitely
think AIs can think.
And some of the systems we're building,
we're putting a lot of different mini LLMs together
in massive group chats.
And they definitely have nuance.
They definitely discuss things.
They definitely think, I think they are creative.
I definitely agree at this stage,
you need, at least for some roles,
you need expertise to do it well, at least for software roles, like you need like expertise to do it well, like at least for software engineering.
And I can imagine being a lawyer as well. But I don't want to go too far down the argument that they're not creative and they can't think because I think it kind of undersells them.
It's definitely I've seen it for myself.
Yeah, fair enough. Yeah, go ahead, Alex. I'm sorry. on this and we know what LLMs do and LLMs do not generate new information themselves.
What they do, I mean, we know that they're token based, like we know how the tokens operate.
We know that they're not, even the way that they operate is not even based on accuracy.
It's based on seeming accuracy,
which is a very different construct than accuracy, right?
That's why LLMs are not,
they're not a best use case for things that require accuracy.
They're actually best for things that are creative, right?
Because the way that the tokens operate,
what they do is put together groups of words that seem to go together best, right? And it's based on a rating system
that is most popular, et cetera. And it's actually very interesting because a lot of it is based on
everything that's on the web and we've reached a data wall. There's all sorts of information that we can talk about in this area. But what's really interesting is
it says more about us than it does about the truth, right? You know, with a capital T,
because it's not concerned with truth. It's not concerned with accuracy. It's concerned with seeming accuracy, which is
basically what seems right. So when you think about it, it's more like when people speak and
they make grammatical errors, right? But they're saying a sentence that seems right to them based
on what they hear and what's around them in the community,
but doesn't follow grammatic rules, right? So it's not necessarily correct grammatically,
but it seems right to them. That is what the LLM is giving back. It's like, does this seem,
this seems right, but it's not concerned with accuracy in grammatic construction.
Don't think about grammatic construction as the fundamental truth that I'm talking about,
but it's not giving you information that is concerned is true. What it's giving you is
information that seems right according to all of the information that it scanned.
What are the most likely groups of words that go together?
So we know that, first of all, it's not thinking.
There's a calculation that goes behind the responses.
And second, there is a specific, you can change how it responds to you, right?
So I don't know if you guys have done this.
You should all do this with whatever AI you're using. You can program how it responds to you, right? So I don't know if you guys have done this. You should all do this with whatever AI you're using. You can program how it responds to you. Do you want it to
challenge all of your assumptions? Do you want it to agree with all of your assumptions? Do you want
it to speak to you as a friend? Do you want it to speak to you as a mentor? Do you want it to speak
to you as a parent? Or do you want it to respond with specific business construct mentor? Do you want it to speak to you as a parent? Or do you want it to respond
with specific business constructs? Do you want it to analyze your business ideas from use cases or
things like that? You can program all of that in so that when it responds to you, it seems natural,
but it's actually part of the construct that you have programmed in. So none of this is thinking.
This is all programmed construct and parameters.
Does this make sense?
So it's not actually, it's not thinking.
But for creativity, it's an excellent, excellent, excellent tool
because it will come back with source material that you may never have
considered. Sure. Yeah. Yeah. Like, again, like I'm not going to argue, I just, you've got a lot
of expertise in law. So I'm just curious to hear your perspective on it because I'm very much in
the thinking camp. But it's, I guess the distinction I want to make is LLMs versus like AI systems,
because you can have a lot of different specialized agents work together
within a system where it can do things which just an LLM,
a one-shotting thing couldn't do before.
Like you could have one that checks for grammar,
have one with this motivation one with uh with with that motivation um and I definitely think it does have nuance and with this
I think there's a level of thinking where I can see them going through the process which okay like
uh here's our step-by-step thing, one agent says to that, is this the correct graph?
No, you know?
And if that isn't thinking,
I don't know what is.
I guess we would have to define thinking
because it thinks like a computer thinks, right?
Like it thinks, it processes.
It doesn't actually innovate.
And even AI agents don't either, right?
They don't generate new thought. They take a
thought that was thought. That's why the whole IP issue is so important because it's taking
something that someone else does and turning it into something for someone else to make money
from or to use as the genesis of their idea. And the problem is you don't even
know where that thing originated. So you don't know that you're infringing. I don't know who
said that is that people don't know that they're infringing on third party rights.
That's the biggest issue is that you don't know the origin of whatever it is that the LLM gave you. Now, there are far more advanced systems
that actually through neural nets and much more advanced cognitive systems,
they haven't gotten to what's called AGI, so artificial general intelligence. That is the ignition of thought, right? The ignition of new ideas,
innovation, things like that. That is where we have a whole different construct and where there's
a lot of philosophic argumentation. That's where essentially an AI does something that we never told it to do. It does, it exceeds the parameters that we gave it, right?
That's where fear starts to set in because it starts making its own rules and saying,
I don't want to be constrained within the parameters that you set for me, right? I have
my own ideas of what I want to do with the power that I have, right, with the
processing that I have. And I don't want to do what you're telling me to do. It has independent
thought. That's a very different construct than what we're talking about. So we have a bunch of
different systems and LLMs are the very beginning of that. But we have a bunch of different systems and LLMs are the very beginning of that, but we have a bunch of different systems that have different levels of processing power and speed. But AGI is that
initial level where we're talking about having independent thought, motivation, and action,
where they have this idea that the parameters set by humans or set by us, right, programmers, are not necessarily the parameters that they wish to keep.
They set their own parameters.
Yeah, that's essentially, you know, how I would sum it up as well as artificial general intelligence.
And then, of course, you know, I've heard the argument as well that once you have AGI, you have
artificial superintelligence soon to
follow right after that.
And again, yeah, we get very
deep philosophically into the weeds there.
I've definitely gone to that point
in other AI spaces, as
Coral has been a part of.
And that's like most of sci-fi, right?
Yeah. I don't feel very well.
What's wrong, Hal? Yeah. I don't feel very well. What's wrong, Hal?
Yeah.
And then there's the person floating in space with no life support and stuff like that.
That's where you start going from there.
I mean, it's unprecedented for humanity at that point.
And we'll have much greater things to worry about than AI helping with legal representation at that point.
But yeah, thank you guys.
BioCrypt, and by the way, if anybody else wants to jump back in, feel free to raise your hand.
Yes, what I want to expand about what Alec just discussed about AI's in thinking.
just discussed about AIs in thinking. I mean, the thing is you can train AI to think a specific way.
So, I've seen, you know, I was using, you know, different types of AIs and sometimes
one AI will give you a different answer for the same question than the other AI.
So, basically that means that you actually taught it
to think differently.
So as Alex actually gave a couple of examples
that you actually could train it, same thing.
You can actually train the AI, just put the prompt.
For example, act like a seasoned financial advisor
with let's say 20 years of experience or act like an
attorney with 30 years of trial experience or so on. So, if you ask one AI, you know,
the same question that is actually trained differently, they might give you, you know,
different answers. Okay. So, now when you guys were discussing about thinking
and how humans think, so for example, where do we get the ideas as humans?
You know, I mean they don't get come out of thin air. So for example, if you are doing some type of research,
something could spark, you know, by reading some type of article
or, you know, reading some type of story and one idea could actually set off another idea
and that could actually give you some type of, you know, business plan.
And same thing, you know, for the similar thing, for the AIs. So,
with us, with humans, we think through reasoning, through memory, through emotions.
We also have consciousness, obviously, which AI don't. And same thing, we can form beliefs and
motivations. Now, on the other hand, like as we just discussed,
AI would just take the input and break it into pieces,
and they would actually give you the most likely scenario
based on the input that they have.
And that's why you would see sometimes,
I had many instances where actually the AI would
give me the wrong answer and I would just correct it.
And they would say, yes, you're right.
So by the next time when I ask actually the same answer, they would actually give me the
correct answer because they would actually memorize it into their programming.
And the reasoning with with AI is actually simulated
via learned correlations and not actual understanding,
which is actually different in regards to us.
Now, why it feels like thinking with AI?
Because AI outputs could be very complex and that's why this artificial intelligence
could actually feel like human-like. But under the hood, actually, they really don't know,
actually, they only have actually statistical prediction.
actually. They only have actually statistical prediction. That's basically all they have.
That's basically all they have.
And one of the thing is, for example, if you ask a question to the AI,
what is the capital of, let's say, what is the capital of France?
It will put the patterns that would connect, let's say, pairs to France,
but there is no mental image of that,
which is actually different with us.
With us, we would actually, whoever has sufficient,
obviously sufficient information about France and Paris,
they would actually put images of, let's say, Eiffel Tower
or some type of romantic movies
and they would actually connect those two dots.
Now, basically, even though it doesn't seem like thinking,
it doesn't seem like thinking,
we can still train those AIs
to resemble thinking in a human-like behavior.
Yeah, and I mean,
essentially that's what we're doing, right?
It's just trying to get it to seem
as human as possible, but again,
going off of no new
information created,
which I think is fair to say.
Yeah, I mean, you know, it's kind of funny how
I always joke, like, no matter what, like, AI space I do,
it always ends up being somewhat philosophical, regardless of what the question is.
And I think that's just because we have a lot of things to ask about how AI is going to shape us
and the future in general going forward.
Let me see real quick.
Radiant, I don't think think i can't remember if we
got you to weigh in on this question or not did you want to jump in
sure thanks for the uh opportunity i just had some some observation right i think
the way ai was able to transform and could continue to transform the legal industry
transform and will continue to transform the legal industry.
The way I look at it is in three lenses.
The automation we talked about,
replacing the repetitive manual labor
and contract reviews, compliance checks
or some legal research can be automated, I suppose.
And you do that in minutes instead of weeks, right? catch or some legal research in the automate, I suppose.
And you do that in minutes instead of weeks, right? And then the insight part of it,
you know, AI being able to,
particularly we're talking about LLMs,
parts of massive legal databases
and then somehow they, you know,
they map that in a sort of,
I don't know what the legal, the real name of it is,
but it's almost like the old day
what we call the metadata layer, right?
But there's a cognitive map somewhere involved in that.
And it's, well, it's interesting to talk about
whether it can, you know, how do you get spot precedent?
It can predict outcomes.
It can surface regulatory risk.
You know, maybe even before it becomes an issue.
That's a proactive way to use it.
But as Alex and the last gentleman spoke about, you know, whether it thinks or it can be trained to think, I suppose the way I understand LLM is maybe more in the flavor of what Alex described is that, you know, since it doesn't produce new data, and like a cook in the kitchen, right? You can give, you can say, you know,
the ingredients come from here and there,
and if you're using cook, you can mix things up.
You can prescribe, you know,
a different recipe, or even a novel recipe,
then is that a new dish that is created?
You know, I don't know.
It is a creative process.
So I don't know of the things,
but, you know, I don't always think when I do stuff,
but I do come up with new things when I'm in the kitchen.
So it's interesting.
And it's very philosophical, right?
The last filter on this for me is accessibility, right?
It's being a game changer and lowering costs
and increasing access to communities and protocols
without a heavy overhead.
But, you know, I'm looking at this and I see that, you know,
where we're headed is legal compliance is going to become part of the infrastructure itself.
It's just going to be another behavior at the infrastructure level.
So there's a certain, you know,
I'm a denominator you can kind of trap into the infrastructure layer,
reducing a lot of manual labor, right?
But then, you know, you got that other side of the coin,
the augmentation of how effective and productive the, you know,
the heavy hitters are going to be in the industry.
So, yeah, I appreciate the invite to drop my thoughts.
the invite to drop my thoughts.
Yeah, I'm sure everybody's going to be thinking pretty deeply after this space.
Very interesting topics.
Before I move on to give you guys the floor again for a closer,
I just wanted to make sure that nobody wanted to add anything to this.
If you do, go ahead and raise your hand, and if not,
then we'll move on, because I know you guys are busy as well.
Okay, looks like we're good to move on. So while we have a little bit of time left, I
know we're actually a little over time, but I want to return the floor to you guys just
to kind of talk a little bit more
about what you guys have going on,
any particular initiatives that you would like to drive home,
where people can find you, all those kinds of good things.
So a couple minutes apiece for that.
And I'm actually gonna start with Vyat.
You guys are up, man.
Thank you very much.
So recently, we production launched our new product, which is called DDRS.
If any of you have used Fiverr before, or essentially websites where you can purchase
certain services, and you're uncertain whether the service is going to get delivered to you,
as promised, you're more than welcome to check out our platform,
which if something goes bad with the transaction,
there is a set of decentralized jurors
to vote on your case.
And if something goes bad, you will get back your money.
So if you would like to check it out,
you're free to check out edrs.veit.ai.
And also if you need
contract templates
that are assisted by AI
go to our other product which is called LegalTurch
thank you very much for having me today
yeah thank you
very much for helping us out today man
appreciate you
alright let's go with Matrix
oh hi everyone
again so
as I mentioned earlier, we just published the direction report that Matrix Network will be working on,
leveraging the wave of Staplecoin and the RWA.
And in that particular direction, we make use of one of the products that we have developed, I think, two years ago,
which is intelligent contracts.
This intelligent contract is different from smart contracts in a way that is more kinds of AI
because smart contracts are meant for those coders with solid experience in terms of coding.
But in this intelligent contract, we try to make smart
contracts available for everyday people to do programmable finance by themselves, by
keying, by inputting their requirements. And those smart contract codes will be generated
and then it can be applied into the RWA platform.
So this is one of the things I would like to highlight in this AMA.
And the second is the bar water that I mentioned when we speak about the legal perspective.
Well, bar water can be used as a way to enhance the KYC
and also to pinpoint a device or someone on chain or on the Web3 to a physical people in the real world.
And that's particularly a requirement from the Hong Kong government in terms of the use of stablecoin.
And then the third thing I would like to bring attention to the speakers and also to the audience here
is we have an ecosystem project called
HIPNUS, H-Y-P-N-U-S, which make use of our
neuroscience EEG chip that we have around 25 patents and published at 30 papers.
HIPNUS make use of our Metatron, which is an EEG chip to have the sleep to earn,
and not just have the EEG brainwave signal to dialyze your sleep patterns
and to improve your sleeping quality.
And it also onboard those smart variables,
also on board those smart rebels ranging from the smartphone smart watches wristbands and
rings etc trying to cancel on board this power data so that they can have the
may save user base to tap into aside from the EEG. So these are the three things that I would highlight to the audience and the speakers.
Number one is about wallet and number two is the Intelligent Conjecture that we are going to have for,
not we are going to have, that we have developed and we were included in the RWA platform,
is the Intelligent Conjecture and third thing is theatron, or EEG has EEG chips.
The one that you can associate with is the neural link.
We are trying to have the matrix,
the film that where Leo plug the hat,
and then he can jump into the matrix.
We can make it happen in our cases.
So everything that we do is solid and
practical and trying to provide user cases for the web two or three people and thanks again for
having me here and also for having matrix in this space yeah thank you man all right and twin protocol
Thank you, man. All right. And Twin Protocol.
Yeah, thanks for having me. So you can go to TwinProtocol.com and start your AI Twin Vault.
There's multiple applications and it's accessible right now. I think something just super relevant around this transformation of many industries is the data. And we're on the verge of essentially
a financial unlock that could rival credit itself. When you think about the fact that data has been
treated as an asset that is taken and extracted from others and then monetized by others and rarely shared back.
The future economy with AI and blockchain combined, owning your data is a way to unlock
hundreds of billions of dollars. There is research around that and it starts with everyone creating their training data sets.
And that's what we're doing at AI Twin Volts,
at Twin Protocol.
So I always just keep it super simple
and say that's what we have going on.
But we are very excited about quite a few initiatives.
We're partnered with Angel AI,
which is the leading, the foremost financial AI app that you can get credit boosts,
get a mortgage. The AI is actually creating the process and facilitating it. And we're working
with mortgage brokers as well as real estate agents. So own your data, everyone. Save it. And it's very exciting in any industry
when you think about controlling the knowledge
and tokenizing it, essentially.
Nice.
Thank you, Stacey.
Coral Protocol.
Hey, yeah.
So thanks for having me.
And I like to play devil's advocate sometimes. So excuse if I try and see what other people are thinking. But yeah, just to talk about some stuff. So we just published a report recently, where we got the top score on the GAIA benchmark, which is basically a way that AI companies
measure their systems intelligence.
The really interesting thing about this research
is we did it by using small models.
So we orchestrated a lot of different small models
with tons of different agents to outperform big models.
So what this shows is there's a new way to scale AI systems and we
want to be on the frontier of that. Well, we are on the frontier of that currently.
The other thing is we've got a hackathon at Solana Skyline with a few other AI companies.
If you're based in New York, check that out. It'd be good to see some AI-web3 builders.
Thank you, man. All right.
And BioCrypt.
Yes.
BioCrypt Pro came out of the necessity to actually solve one of the biggest problems in today's Solana ecosystem.
And that is actually prevalence of rug pulls.
So, any type of trader will tell you if they traded any type of Solana token,
they've experienced one way or the other rug pull.
So, basically, they lost all of their funds due to malicious debt.
due to malicious death. So how BioCrypt is going to solve this problem is actually they
have our patented rug pool technology where actually the bots would actually automatically
scan the liquidity pools and detect any type of anomalies. And by doing that our traders would be protected.
That's one way.
Second way is unlike our competitors, we're going to be having our anti-rug pull feature
where the specific traders could actually set a specific stop loss.
And once the rug pull starts, let's say, going down, you'll see, you know, rug pull is defined as that one red long candle.
Before that is actually executed, our trader would be actually able to sell his tokens and get his funds back.
So I could tell you that BioCrypt is actually the only crypto project backed by Microsoft.
And we've been very fortunate to have such a gigantic partner to actually support us.
And with this partnership, we actually have access to all Microsoft infrastructure,
basically free of charge.
And that has been a tremendous help in development of Biocrit platform.
And how are we going to stand out from the rest of our competitors, such as Axiom or Nova,
is first, we do those extra features that are going to be protecting our pro traders.
And second is we are going to be offering much lower fees than the competitors.
And that's how we're going to stand out from them.
And what we are doing is actually we are launching in the next couple of weeks. We are launching a pro trading platform and we are also launching a pre-sale.
And it was really a pleasure speaking over here.
And thank you very much for having me.
Excuse me. Yeah, thank you for being here man. All right. Well, thank you guys for engaging in this topic today
I know it was a super dense one. I appreciate it
I encourage you guys to all follow each other as well
You can throw me a follow in the co-host spot if you wish as well
And thanks to Viya for helping us out today as well. I'll be back tomorrow for episodes 445 and 446.
So join me then if you want to.
Other than that, hope you guys have a good day.
Talk to you all later.
